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  * @access public
  17  * @package MediaWiki
  18  * @subpackage DifferenceEngine
  19  */
  20 class DifferenceEngine {
  21         /**#@+
  22          * @access private
  23          */
  24         var $mOldid, $mNewid, $mTitle;
  25         var $mOldtitle, $mNewtitle, $mPagetitle;
  26         var $mOldtext, $mNewtext;
  27         var $mOldUser, $mNewUser;
  28         var $mOldComment, $mNewComment;
  29         var $mOldPage, $mNewPage;
  30         var $mRcidMarkPatrolled;
  31         var $mOldRev, $mNewRev;
  32         var $mRevisionsLoaded = false; // Have the revisions been loaded
  33         var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
  34         /**#@-*/
  35
  36         /**
  37          * Constructor
  38          * @param Title $titleObj Title object that the diff is associated with
  39          * @param integer $old Old ID we want to show and diff with.
  40          * @param string $new Either 'prev' or 'next'.
  41          * @param integer $rcid ??? (default 0)
  42          */
  43         function DifferenceEngine( $titleObj = null, $old = 0, $new = 0, $rcid = 0 ) {
  44                 $this->mTitle = $titleObj;
  45
  46                 if ( 'prev' == $new ) {
  47                         # Show diff between revision $old and the previous one.
  48                         # Get previous one from DB.
  49                         #
  50                         $this->mNewid = intval($old);
  51
  52                         $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
  53
  54                 } elseif ( 'next' == $new ) {
  55                         # Show diff between revision $old and the previous one.
  56                         # Get previous one from DB.
  57                         #
  58                         $this->mOldid = intval($old);
  59                         $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
  60                         if ( false === $this->mNewid ) {
  61                                 # if no result, NewId points to the newest old revision. The only newer
  62                                 # revision is cur, which is "0".
  63                                 $this->mNewid = 0;
  64                         }
  65
  66                 } else {
  67                         $this->mOldid = intval($old);
  68                         $this->mNewid = intval($new);
  69                 }
  70                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  71         }
  72
  73         function showDiffPage() {
  74                 global $wgUser, $wgOut, $wgContLang, $wgOnlySysopsCanPatrol,
  75                        $wgUseExternalEditor, $wgUseRCPatrol;
  76                 $fname = 'DifferenceEngine::showDiffPage';
  77                 wfProfileIn( $fname );
  78
  79                 # If external diffs are enabled both globally and for the user,
  80                 # we'll use the application/x-external-editor interface to call
  81                 # an external diff tool like kompare, kdiff3, etc.
  82                 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
  83                         global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
  84                         $wgOut->disable();
  85                         header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
  86                         $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
  87                         $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
  88                         $special=$wgLang->getNsText(NS_SPECIAL);
  89                         $control=<<<CONTROL
  90 [Process]
  91 Type=Diff text
  92 Engine=MediaWiki
  93 Script={$wgServer}{$wgScript}
  94 Special namespace={$special}
  95
  96 [File]
  97 Extension=wiki
  98 URL=$url1
  99
 100 [File 2]
 101 Extension=wiki
 102 URL=$url2
 103 CONTROL;
 104                         echo($control);
 105                         return;
 106                 }
 107
 108                 # mOldid is false if the difference engine is called with a "vague" query for
 109                 # a diff between a version V and its previous version V' AND the version V
 110                 # is the first version of that article. In that case, V' does not exist.
 111                 if ( $this->mOldid === false ) {
 112                         $this->showFirstRevision();
 113                         wfProfileOut( $fname );
 114                         return;
 115                 }
 116
 117                 $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 118                   "{$this->mNewid})";
 119                 $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 120
 121                 $wgOut->setArticleFlag( false );
 122                 if ( ! $this->loadRevisionData() ) {
 123                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 124                         $wgOut->addWikitext( $mtext );
 125                         wfProfileOut( $fname );
 126                         return;
 127                 }
 128                 if ( $this->mNewRev->isCurrent() ) {
 129                         $wgOut->setArticleFlag( true );
 130                 }
 131
 132                 $wgOut->suppressQuickbar();
 133
 134                 $oldTitle = $this->mOldPage->getPrefixedText();
 135                 $newTitle = $this->mNewPage->getPrefixedText();
 136                 if( $oldTitle == $newTitle ) {
 137                         $wgOut->setPageTitle( $newTitle );
 138                 } else {
 139                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
 140                 }
 141                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 142                 $wgOut->setRobotpolicy( 'noindex,follow' );
 143
 144                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
 145                         $wgOut->loginToUse();
 146                         $wgOut->output();
 147                         wfProfileOut( $fname );
 148                         exit;
 149                 }
 150
 151                 $sk = $wgUser->getSkin();
 152                 $talk = $wgContLang->getNsText( NS_TALK );
 153                 $contribs = wfMsg( 'contribslink' );
 154
 155                 $this->mOldComment = $sk->formatComment($this->mOldComment);
 156                 $this->mNewComment = $sk->formatComment($this->mNewComment);
 157
 158                 $oldUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 159                 $newUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mNewUser ), $this->mNewUser );
 160                 $oldUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ), $talk );
 161                 $newUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mNewUser ), $talk );
 162                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 163                         'target=' . urlencode($this->mOldUser) );
 164                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 165                         'target=' . urlencode($this->mNewUser) );
 166                 if ( $this->mNewRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
 167                         $rollback = '&nbsp;&nbsp;&nbsp;<strong>[' . $sk->makeKnownLinkObj( $this->mTitle, wfMsg( 'rollbacklink' ),
 168                                 'action=rollback&from=' . urlencode($this->mNewUser) .
 169                                 '&token=' . urlencode( $wgUser->editToken( array( $this->mTitle->getPrefixedText(), $this->mNewUser ) ) ) ) .
 170                                 ']</strong>';
 171                 } else {
 172                         $rollback = '';
 173                 }
 174                 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->isLoggedIn() &&
 175                      ( $wgUser->isAllowed('rollback') || !$wgOnlySysopsCanPatrol ) )
 176                 {
 177                         $patrol = ' [' . $sk->makeKnownLinkObj( $this->mTitle, wfMsg( 'markaspatrolleddiff' ),
 178                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 179                 } else {
 180                         $patrol = '';
 181                 }
 182
 183                 $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
 184                         'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
 185                 if ( $this->mNewRev->isCurrent() ) {
 186                         $nextlink = '';
 187                 } else {
 188                         $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
 189                                 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 190                 }
 191
 192                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
 193                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment .
 194                         '<br />' . $prevlink;
 195                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
 196                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment .
 197                         '<br />' . $nextlink . $patrol;
 198
 199                 $this->showDiff( $oldHeader, $newHeader );
 200                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 201
 202                 if( !$this->mNewRev->isCurrent() ) {
 203                         $oldEditSectionSetting = $wgOut->mParserOptions->setEditSection( false );
 204                 }
 205
 206                 $this->loadNewText();
 207                 if( is_object( $this->mNewRev ) ) {
 208                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 209                 }
 210                 $wgOut->addWikiText( $this->mNewtext );
 211
 212                 if( !$this->mNewRev->isCurrent() ) {
 213                         $wgOut->mParserOptions->setEditSection( $oldEditSectionSetting );
 214                 }
 215
 216                 wfProfileOut( $fname );
 217         }
 218
 219         /**
 220          * Show the first revision of an article. Uses normal diff headers in
 221          * contrast to normal "old revision" display style.
 222          */
 223         function showFirstRevision() {
 224                 global $wgOut, $wgUser, $wgLang;
 225
 226                 $fname = 'DifferenceEngine::showFirstRevision';
 227                 wfProfileIn( $fname );
 228
 229                 # Get article text from the DB
 230                 #
 231                 if ( ! $this->loadNewText() ) {
 232                         $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 233                           "{$this->mNewid})";
 234                         $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
 235                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 236                         $wgOut->addWikitext( $mtext );
 237                         wfProfileOut( $fname );
 238                         return;
 239                 }
 240                 if ( $this->mNewRev->isCurrent() ) {
 241                         $wgOut->setArticleFlag( true );
 242                 }
 243
 244                 # Check if user is allowed to look at this page. If not, bail out.
 245                 #
 246                 if ( !( $this->mTitle->userCanRead() ) ) {
 247                         $wgOut->loginToUse();
 248                         $wgOut->output();
 249                         wfProfileOut( $fname );
 250                         exit;
 251                 }
 252
 253                 # Prepare the header box
 254                 #
 255                 $sk = $wgUser->getSkin();
 256
 257                 $uTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ),  $wgLang->getNsText( NS_TALK ) );
 258                 $userLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 259                 $contribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), wfMsg( 'contribslink' ),
 260                         'target=' . urlencode($this->mOldUser) );
 261                 $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
 262                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />$userLink " .
 263                         "($uTLink | $contribs)<br />" . $this->mOldComment .
 264                         '<br />' . $nextlink. "</div>\n";
 265
 266                 $wgOut->addHTML( $header );
 267
 268                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 269                 $wgOut->setRobotpolicy( 'noindex,follow' );
 270
 271
 272                 # Show current revision
 273                 #
 274                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 275                 if( is_object( $this->mNewRev ) ) {
 276                         $wgOut->setRevisionId( $this->mNewRev->getId() );
 277                 }
 278                 $wgOut->addWikiText( $this->mNewtext );
 279
 280                 wfProfileOut( $fname );
 281         }
 282
 283         /**
 284          * Get the diff text, send it to $wgOut
 285          * Returns false if the diff could not be generated, otherwise returns true
 286          */
 287         function showDiff( $otitle, $ntitle ) {
 288                 global $wgOut;
 289                 $diff = $this->getDiff( $otitle, $ntitle );
 290                 if ( $diff === false ) {
 291                         $wgOut->addWikitext( wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" ) );
 292                         return false;
 293                 } else {
 294                         $wgOut->addHTML( $diff );
 295                         return true;
 296                 }
 297         }
 298
 299         /**
 300          * Get diff table, including header
 301          * Note that the interface has changed, it's no longer static.
 302          * Returns false on error
 303          */
 304         function getDiff( $otitle, $ntitle ) {
 305                 $body = $this->getDiffBody();
 306                 if ( $body === false ) {
 307                         return false;
 308                 } else {
 309                         return $this->addHeader( $body, $otitle, $ntitle );
 310                 }
 311         }
 312
 313         /**
 314          * Get the diff table body, without header
 315          * Returns false on error
 316          */
 317         function getDiffBody() {
 318                 global $wgUseExternalDiffEngine, $wgContLang, $wgMemc, $wgDBname;
 319
 320                 // Cacheable?
 321                 $key = false;
 322                 if ( $this->mOldid && $this->mNewid ) {
 323                         // Try cache
 324                         $key = "$wgDBname:diff:oldid:{$this->mOldid}:newid:{$this->mNewid}";
 325                         $difftext = $wgMemc->get( $key );
 326                         if ( $difftext ) {
 327                                 return $difftext;
 328                         }
 329                 }
 330
 331                 if ( !$this->loadText() ) {
 332                         return false;
 333                 }
 334
 335                 $otext = $wgContLang->segmentForDiff($this->mOldtext);
 336                 $ntext = $wgContLang->segmentForDiff($this->mNewtext);
 337                 if ( $wgUseExternalDiffEngine ) {
 338                         # For historical reasons, external diff engine expects
 339                         # input text to be HTML-escaped already
 340                         $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
 341                         $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
 342                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 343                                 dl('php_wikidiff.so');
 344                         }
 345                         $difftext = wikidiff_do_diff( $otext, $ntext, 2 );
 346                 } else {
 347                         $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
 348                         $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
 349                         $diffs =& new Diff( $ota, $nta );
 350                         $formatter =& new TableDiffFormatter();
 351                         $difftext = $formatter->format( $diffs );
 352                 }
 353                 $difftext = $wgContLang->unsegmentForDiff($difftext);
 354
 355                 // Save to cache for 7 days
 356                 if ( $key !== false ) {
 357                         $wgMemc->set( $key, $difftext, 7*86400 );
 358                 }
 359                 return $difftext;
 360         }
 361
 362         /**
 363          * Add the header to a diff body
 364          */
 365         function addHeader( $diff, $otitle, $ntitle ) {
 366                 $out = "
 367                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 368                         <tr>
 369                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 370                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 371                         </tr>
 372                         $diff
 373                         </table>
 374                 ";
 375                 return $out;
 376         }
 377
 378         /**
 379          * Use specified text instead of loading from the database
 380          */
 381         function setText( $oldText, $newText ) {
 382                 $this->mOldtext = $oldText;
 383                 $this->mNewtext = $newText;
 384                 $this->mTextLoaded = 2;
 385         }
 386
 387         /**
 388          * Load revision metadata for the specified articles. If newid is 0, then compare
 389          * the old article in oldid to the current article; if oldid is 0, then
 390          * compare the current article to the immediately previous one (ignoring the
 391          * value of newid).
 392          *
 393          * If oldid is false, leave the corresponding revision object set
 394          * to false. This is impossible via ordinary user input, and is provided for
 395          * API convenience.
 396          */
 397         function loadRevisionData() {
 398                 global $wgLang;
 399                 if ( $this->mRevisionsLoaded ) {
 400                         return true;
 401                 } else {
 402                         // Whether it succeeds or fails, we don't want to try again
 403                         $this->mRevisionsLoaded = true;
 404                 }
 405
 406                 // Load the new revision object
 407                 if( $this->mNewid ) {
 408                         $this->mNewRev = Revision::newFromId( $this->mNewid );
 409                 } else {
 410                         $this->mNewRev = Revision::newFromTitle( $this->mTitle );
 411                 }
 412
 413                 if( is_null( $this->mNewRev ) ) {
 414                         return false;
 415                 }
 416
 417                 // Set assorted variables
 418                 if( $this->mNewRev->isCurrent() ) {
 419                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 420                         $this->mNewPage = $this->mTitle;
 421                         $newLink = $this->mNewPage->escapeLocalUrl();
 422                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 423                 } else {
 424                         $this->mNewPage = $this->mNewRev->getTitle();
 425                         $newLink = $this->mNewPage->escapeLocalUrl ('oldid=' . $this->mNewid );
 426                         $t = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 427                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 428                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 429                 }
 430
 431                 $this->mNewUser = $this->mNewRev->getUserText();
 432                 $this->mNewComment = $this->mNewRev->getComment();
 433
 434                 // Load the old revision object
 435                 $this->mOldRev = false;
 436                 if( $this->mOldid ) {
 437                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 438                 } elseif ( $this->mOldid === 0 ) {
 439                         $this->mOldRev = $this->mNewRev->getPrevious();
 440                         $this->mOldid = $this->mOldRev->getId();
 441                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 442
 443                 if( is_null( $this->mOldRev ) ) {
 444                         return false;
 445                 }
 446
 447                 if ( $this->mOldRev ) {
 448                         $this->mOldPage = $this->mOldRev->getTitle();
 449
 450                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 451                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 452                         $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
 453
 454
 455                         $this->mOldUser = $this->mOldRev->getUserText();
 456                         $this->mOldComment = $this->mOldRev->getComment();
 457                 }
 458
 459                 return true;
 460         }
 461
 462         /**
 463          * Load the text of the revisions, as well as revision data.
 464          */
 465         function loadText() {
 466                 if ( $this->mTextLoaded == 2 ) {
 467                         return true;
 468                 } else {
 469                         // Whether it succeeds or fails, we don't want to try again
 470                         $this->mTextLoaded = 2;
 471                 }
 472
 473                 if ( !$this->loadRevisionData() ) {
 474                         return false;
 475                 }
 476                 if ( $this->mOldRev ) {
 477                         $this->mOldtext = $this->mOldRev->getText();
 478                 }
 479                 if ( $this->mNewRev ) {
 480                         $this->mNewtext = $this->mNewRev->getText();
 481                 }
 482                 return true;
 483         }
 484
 485         /**
 486          * Load the text of the new revision, not the old one
 487          */
 488         function loadNewText() {
 489                 if ( $this->mTextLoaded >= 1 ) {
 490                         return true;
 491                 } else {
 492                         $this->mTextLoaded = 1;
 493                 }
 494                 if ( !$this->loadRevisionData() ) {
 495                         return false;
 496                 }
 497                 $this->mNewtext = $this->mNewRev->getText();
 498                 return true;
 499         }
 500
 501
 502 }
 503
 504 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 505 //
 506 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 507 // You may copy this code freely under the conditions of the GPL.
 508 //
 509
 510 define('USE_ASSERTS', function_exists('assert'));
 511
 512 /**
 513  * @todo document
 514  * @access private
 515  * @package MediaWiki
 516  * @subpackage DifferenceEngine
 517  */
 518 class _DiffOp {
 519         var $type;
 520         var $orig;
 521         var $closing;
 522
 523         function reverse() {
 524                 trigger_error('pure virtual', E_USER_ERROR);
 525         }
 526
 527         function norig() {
 528                 return $this->orig ? sizeof($this->orig) : 0;
 529         }
 530
 531         function nclosing() {
 532                 return $this->closing ? sizeof($this->closing) : 0;
 533         }
 534 }
 535
 536 /**
 537  * @todo document
 538  * @access private
 539  * @package MediaWiki
 540  * @subpackage DifferenceEngine
 541  */
 542 class _DiffOp_Copy extends _DiffOp {
 543         var $type = 'copy';
 544
 545         function _DiffOp_Copy ($orig, $closing = false) {
 546                 if (!is_array($closing))
 547                         $closing = $orig;
 548                 $this->orig = $orig;
 549                 $this->closing = $closing;
 550         }
 551
 552         function reverse() {
 553                 return new _DiffOp_Copy($this->closing, $this->orig);
 554         }
 555 }
 556
 557 /**
 558  * @todo document
 559  * @access private
 560  * @package MediaWiki
 561  * @subpackage DifferenceEngine
 562  */
 563 class _DiffOp_Delete extends _DiffOp {
 564         var $type = 'delete';
 565
 566         function _DiffOp_Delete ($lines) {
 567                 $this->orig = $lines;
 568                 $this->closing = false;
 569         }
 570
 571         function reverse() {
 572                 return new _DiffOp_Add($this->orig);
 573         }
 574 }
 575
 576 /**
 577  * @todo document
 578  * @access private
 579  * @package MediaWiki
 580  * @subpackage DifferenceEngine
 581  */
 582 class _DiffOp_Add extends _DiffOp {
 583         var $type = 'add';
 584
 585         function _DiffOp_Add ($lines) {
 586                 $this->closing = $lines;
 587                 $this->orig = false;
 588         }
 589
 590         function reverse() {
 591                 return new _DiffOp_Delete($this->closing);
 592         }
 593 }
 594
 595 /**
 596  * @todo document
 597  * @access private
 598  * @package MediaWiki
 599  * @subpackage DifferenceEngine
 600  */
 601 class _DiffOp_Change extends _DiffOp {
 602         var $type = 'change';
 603
 604         function _DiffOp_Change ($orig, $closing) {
 605                 $this->orig = $orig;
 606                 $this->closing = $closing;
 607         }
 608
 609         function reverse() {
 610                 return new _DiffOp_Change($this->closing, $this->orig);
 611         }
 612 }
 613
 614
 615 /**
 616  * Class used internally by Diff to actually compute the diffs.
 617  *
 618  * The algorithm used here is mostly lifted from the perl module
 619  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 620  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 621  *
 622  * More ideas are taken from:
 623  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 624  *
 625  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 626  * diffutils-2.7, which can be found at:
 627  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 628  *
 629  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 630  * are my own.
 631  *
 632  * Line length limits for robustness added by Tim Starling, 2005-08-31
 633  *
 634  * @author Geoffrey T. Dairiki, Tim Starling
 635  * @access private
 636  * @package MediaWiki
 637  * @subpackage DifferenceEngine
 638  */
 639 class _DiffEngine
 640 {
 641         function diff ($from_lines, $to_lines) {
 642                 $fname = '_DiffEngine::diff';
 643                 wfProfileIn( $fname );
 644
 645                 $n_from = sizeof($from_lines);
 646                 $n_to = sizeof($to_lines);
 647
 648                 $this->xchanged = $this->ychanged = array();
 649                 $this->xv = $this->yv = array();
 650                 $this->xind = $this->yind = array();
 651                 unset($this->seq);
 652                 unset($this->in_seq);
 653                 unset($this->lcs);
 654
 655                 // Skip leading common lines.
 656                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 657                         if ($from_lines[$skip] !== $to_lines[$skip])
 658                                 break;
 659                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 660                 }
 661                 // Skip trailing common lines.
 662                 $xi = $n_from; $yi = $n_to;
 663                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 664                         if ($from_lines[$xi] !== $to_lines[$yi])
 665                                 break;
 666                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 667                 }
 668
 669                 // Ignore lines which do not exist in both files.
 670                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 671                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 672                 }
 673
 674                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 675                         $line = $to_lines[$yi];
 676                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 677                                 continue;
 678                         $yhash[$this->_line_hash($line)] = 1;
 679                         $this->yv[] = $line;
 680                         $this->yind[] = $yi;
 681                 }
 682                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 683                         $line = $from_lines[$xi];
 684                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 685                                 continue;
 686                         $this->xv[] = $line;
 687                         $this->xind[] = $xi;
 688                 }
 689
 690                 // Find the LCS.
 691                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 692
 693                 // Merge edits when possible
 694                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 695                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 696
 697                 // Compute the edit operations.
 698                 $edits = array();
 699                 $xi = $yi = 0;
 700                 while ($xi < $n_from || $yi < $n_to) {
 701                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 702                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 703
 704                         // Skip matching "snake".
 705                         $copy = array();
 706                         while ( $xi < $n_from && $yi < $n_to
 707                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 708                                 $copy[] = $from_lines[$xi++];
 709                                 ++$yi;
 710                         }
 711                         if ($copy)
 712                                 $edits[] = new _DiffOp_Copy($copy);
 713
 714                         // Find deletes & adds.
 715                         $delete = array();
 716                         while ($xi < $n_from && $this->xchanged[$xi])
 717                                 $delete[] = $from_lines[$xi++];
 718
 719                         $add = array();
 720                         while ($yi < $n_to && $this->ychanged[$yi])
 721                                 $add[] = $to_lines[$yi++];
 722
 723                         if ($delete && $add)
 724                                 $edits[] = new _DiffOp_Change($delete, $add);
 725                         elseif ($delete)
 726                                 $edits[] = new _DiffOp_Delete($delete);
 727                         elseif ($add)
 728                                 $edits[] = new _DiffOp_Add($add);
 729                 }
 730                 wfProfileOut( $fname );
 731                 return $edits;
 732         }
 733
 734         /**
 735          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 736          */
 737         function _line_hash( $line ) {
 738                 if ( strlen( $line ) > MAX_DIFF_XREF_LENGTH ) {
 739                         return md5( $line );
 740                 } else {
 741                         return $line;
 742                 }
 743         }
 744
 745
 746         /* Divide the Largest Common Subsequence (LCS) of the sequences
 747          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 748          * sized segments.
 749          *
 750          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 751          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 752          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 753          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 754          * that (X0, Y0) == (XOFF, YOFF) and
 755          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 756          *
 757          * This function assumes that the first lines of the specified portions
 758          * of the two files do not match, and likewise that the last lines do not
 759          * match.  The caller must trim matching lines from the beginning and end
 760          * of the portions it is going to specify.
 761          */
 762         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 763                 $fname = '_DiffEngine::_diag';
 764                 wfProfileIn( $fname );
 765                 $flip = false;
 766
 767                 if ($xlim - $xoff > $ylim - $yoff) {
 768                         // Things seems faster (I'm not sure I understand why)
 769                                 // when the shortest sequence in X.
 770                                 $flip = true;
 771                         list ($xoff, $xlim, $yoff, $ylim)
 772                         = array( $yoff, $ylim, $xoff, $xlim);
 773                 }
 774
 775                 if ($flip)
 776                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 777                                 $ymatches[$this->xv[$i]][] = $i;
 778                 else
 779                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 780                                 $ymatches[$this->yv[$i]][] = $i;
 781
 782                 $this->lcs = 0;
 783                 $this->seq[0]= $yoff - 1;
 784                 $this->in_seq = array();
 785                 $ymids[0] = array();
 786
 787                 $numer = $xlim - $xoff + $nchunks - 1;
 788                 $x = $xoff;
 789                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 790                         wfProfileIn( "$fname-chunk" );
 791                         if ($chunk > 0)
 792                                 for ($i = 0; $i <= $this->lcs; $i++)
 793                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 794
 795                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 796                         for ( ; $x < $x1; $x++) {
 797                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 798                                         if (empty($ymatches[$line]))
 799                                                 continue;
 800                                 $matches = $ymatches[$line];
 801                                 reset($matches);
 802                                 while (list ($junk, $y) = each($matches))
 803                                         if (empty($this->in_seq[$y])) {
 804                                                 $k = $this->_lcs_pos($y);
 805                                                 USE_ASSERTS && assert($k > 0);
 806                                                 $ymids[$k] = $ymids[$k-1];
 807                                                 break;
 808                                         }
 809                                 while (list ($junk, $y) = each($matches)) {
 810                                         if ($y > $this->seq[$k-1]) {
 811                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 812                                                 // Optimization: this is a common case:
 813                                                 //      next match is just replacing previous match.
 814                                                 $this->in_seq[$this->seq[$k]] = false;
 815                                                 $this->seq[$k] = $y;
 816                                                 $this->in_seq[$y] = 1;
 817                                         } else if (empty($this->in_seq[$y])) {
 818                                                 $k = $this->_lcs_pos($y);
 819                                                 USE_ASSERTS && assert($k > 0);
 820                                                 $ymids[$k] = $ymids[$k-1];
 821                                         }
 822                                 }
 823                         }
 824                         wfProfileOut( "$fname-chunk" );
 825                 }
 826
 827                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 828                 $ymid = $ymids[$this->lcs];
 829                 for ($n = 0; $n < $nchunks - 1; $n++) {
 830                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 831                         $y1 = $ymid[$n] + 1;
 832                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 833                 }
 834                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 835
 836                 wfProfileOut( $fname );
 837                 return array($this->lcs, $seps);
 838         }
 839
 840         function _lcs_pos ($ypos) {
 841                 $fname = '_DiffEngine::_lcs_pos';
 842                 wfProfileIn( $fname );
 843
 844                 $end = $this->lcs;
 845                 if ($end == 0 || $ypos > $this->seq[$end]) {
 846                         $this->seq[++$this->lcs] = $ypos;
 847                         $this->in_seq[$ypos] = 1;
 848                         wfProfileOut( $fname );
 849                         return $this->lcs;
 850                 }
 851
 852                 $beg = 1;
 853                 while ($beg < $end) {
 854                         $mid = (int)(($beg + $end) / 2);
 855                         if ( $ypos > $this->seq[$mid] )
 856                                 $beg = $mid + 1;
 857                         else
 858                                 $end = $mid;
 859                 }
 860
 861                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 862
 863                 $this->in_seq[$this->seq[$end]] = false;
 864                 $this->seq[$end] = $ypos;
 865                 $this->in_seq[$ypos] = 1;
 866                 wfProfileOut( $fname );
 867                 return $end;
 868         }
 869
 870         /* Find LCS of two sequences.
 871          *
 872          * The results are recorded in the vectors $this->{x,y}changed[], by
 873          * storing a 1 in the element for each line that is an insertion
 874          * or deletion (ie. is not in the LCS).
 875          *
 876          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 877          *
 878          * Note that XLIM, YLIM are exclusive bounds.
 879          * All line numbers are origin-0 and discarded lines are not counted.
 880          */
 881         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 882                 $fname = '_DiffEngine::_compareseq';
 883                 wfProfileIn( $fname );
 884
 885                 // Slide down the bottom initial diagonal.
 886                 while ($xoff < $xlim && $yoff < $ylim
 887                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 888                         ++$xoff;
 889                         ++$yoff;
 890                 }
 891
 892                 // Slide up the top initial diagonal.
 893                 while ($xlim > $xoff && $ylim > $yoff
 894                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 895                         --$xlim;
 896                         --$ylim;
 897                 }
 898
 899                 if ($xoff == $xlim || $yoff == $ylim)
 900                         $lcs = 0;
 901                 else {
 902                         // This is ad hoc but seems to work well.
 903                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 904                         //$nchunks = max(2,min(8,(int)$nchunks));
 905                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 906                         list ($lcs, $seps)
 907                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 908                 }
 909
 910                 if ($lcs == 0) {
 911                         // X and Y sequences have no common subsequence:
 912                         // mark all changed.
 913                         while ($yoff < $ylim)
 914                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 915                         while ($xoff < $xlim)
 916                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 917                 } else {
 918                         // Use the partitions to split this problem into subproblems.
 919                         reset($seps);
 920                         $pt1 = $seps[0];
 921                         while ($pt2 = next($seps)) {
 922                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 923                                 $pt1 = $pt2;
 924                         }
 925                 }
 926                 wfProfileOut( $fname );
 927         }
 928
 929         /* Adjust inserts/deletes of identical lines to join changes
 930          * as much as possible.
 931          *
 932          * We do something when a run of changed lines include a
 933          * line at one end and has an excluded, identical line at the other.
 934          * We are free to choose which identical line is included.
 935          * `compareseq' usually chooses the one at the beginning,
 936          * but usually it is cleaner to consider the following identical line
 937          * to be the "change".
 938          *
 939          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 940          */
 941         function _shift_boundaries ($lines, &$changed, $other_changed) {
 942                 $fname = '_DiffEngine::_shift_boundaries';
 943                 wfProfileIn( $fname );
 944                 $i = 0;
 945                 $j = 0;
 946
 947                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 948                 $len = sizeof($lines);
 949                 $other_len = sizeof($other_changed);
 950
 951                 while (1) {
 952                         /*
 953                          * Scan forwards to find beginning of another run of changes.
 954                          * Also keep track of the corresponding point in the other file.
 955                          *
 956                          * Throughout this code, $i and $j are adjusted together so that
 957                          * the first $i elements of $changed and the first $j elements
 958                          * of $other_changed both contain the same number of zeros
 959                          * (unchanged lines).
 960                          * Furthermore, $j is always kept so that $j == $other_len or
 961                          * $other_changed[$j] == false.
 962                          */
 963                         while ($j < $other_len && $other_changed[$j])
 964                                 $j++;
 965
 966                         while ($i < $len && ! $changed[$i]) {
 967                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 968                                 $i++; $j++;
 969                                 while ($j < $other_len && $other_changed[$j])
 970                                         $j++;
 971                         }
 972
 973                         if ($i == $len)
 974                                 break;
 975
 976                         $start = $i;
 977
 978                         // Find the end of this run of changes.
 979                         while (++$i < $len && $changed[$i])
 980                                 continue;
 981
 982                         do {
 983                                 /*
 984                                  * Record the length of this run of changes, so that
 985                                  * we can later determine whether the run has grown.
 986                                  */
 987                                 $runlength = $i - $start;
 988
 989                                 /*
 990                                  * Move the changed region back, so long as the
 991                                  * previous unchanged line matches the last changed one.
 992                                  * This merges with previous changed regions.
 993                                  */
 994                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 995                                         $changed[--$start] = 1;
 996                                         $changed[--$i] = false;
 997                                         while ($start > 0 && $changed[$start - 1])
 998                                                 $start--;
 999                                         USE_ASSERTS && assert('$j > 0');
1000                                         while ($other_changed[--$j])
1001                                                 continue;
1002                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1003                                 }
1004
1005                                 /*
1006                                  * Set CORRESPONDING to the end of the changed run, at the last
1007                                  * point where it corresponds to a changed run in the other file.
1008                                  * CORRESPONDING == LEN means no such point has been found.
1009                                  */
1010                                 $corresponding = $j < $other_len ? $i : $len;
1011
1012                                 /*
1013                                  * Move the changed region forward, so long as the
1014                                  * first changed line matches the following unchanged one.
1015                                  * This merges with following changed regions.
1016                                  * Do this second, so that if there are no merges,
1017                                  * the changed region is moved forward as far as possible.
1018                                  */
1019                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1020                                         $changed[$start++] = false;
1021                                         $changed[$i++] = 1;
1022                                         while ($i < $len && $changed[$i])
1023                                                 $i++;
1024
1025                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1026                                         $j++;
1027                                         if ($j < $other_len && $other_changed[$j]) {
1028                                                 $corresponding = $i;
1029                                                 while ($j < $other_len && $other_changed[$j])
1030                                                         $j++;
1031                                         }
1032                                 }
1033                         } while ($runlength != $i - $start);
1034
1035                         /*
1036                          * If possible, move the fully-merged run of changes
1037                          * back to a corresponding run in the other file.
1038                          */
1039                         while ($corresponding < $i) {
1040                                 $changed[--$start] = 1;
1041                                 $changed[--$i] = 0;
1042                                 USE_ASSERTS && assert('$j > 0');
1043                                 while ($other_changed[--$j])
1044                                         continue;
1045                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1046                         }
1047                 }
1048                 wfProfileOut( $fname );
1049         }
1050 }
1051
1052 /**
1053  * Class representing a 'diff' between two sequences of strings.
1054  * @todo document
1055  * @access private
1056  * @package MediaWiki
1057  * @subpackage DifferenceEngine
1058  */
1059 class Diff
1060 {
1061         var $edits;
1062
1063         /**
1064          * Constructor.
1065          * Computes diff between sequences of strings.
1066          *
1067          * @param $from_lines array An array of strings.
1068          *                (Typically these are lines from a file.)
1069          * @param $to_lines array An array of strings.
1070          */
1071         function Diff($from_lines, $to_lines) {
1072                 $eng = new _DiffEngine;
1073                 $this->edits = $eng->diff($from_lines, $to_lines);
1074                 //$this->_check($from_lines, $to_lines);
1075         }
1076
1077         /**
1078          * Compute reversed Diff.
1079          *
1080          * SYNOPSIS:
1081          *
1082          *      $diff = new Diff($lines1, $lines2);
1083          *      $rev = $diff->reverse();
1084          * @return object A Diff object representing the inverse of the
1085          *                                original diff.
1086          */
1087         function reverse () {
1088                 $rev = $this;
1089                 $rev->edits = array();
1090                 foreach ($this->edits as $edit) {
1091                         $rev->edits[] = $edit->reverse();
1092                 }
1093                 return $rev;
1094         }
1095
1096         /**
1097          * Check for empty diff.
1098          *
1099          * @return bool True iff two sequences were identical.
1100          */
1101         function isEmpty () {
1102                 foreach ($this->edits as $edit) {
1103                         if ($edit->type != 'copy')
1104                                 return false;
1105                 }
1106                 return true;
1107         }
1108
1109         /**
1110          * Compute the length of the Longest Common Subsequence (LCS).
1111          *
1112          * This is mostly for diagnostic purposed.
1113          *
1114          * @return int The length of the LCS.
1115          */
1116         function lcs () {
1117                 $lcs = 0;
1118                 foreach ($this->edits as $edit) {
1119                         if ($edit->type == 'copy')
1120                                 $lcs += sizeof($edit->orig);
1121                 }
1122                 return $lcs;
1123         }
1124
1125         /**
1126          * Get the original set of lines.
1127          *
1128          * This reconstructs the $from_lines parameter passed to the
1129          * constructor.
1130          *
1131          * @return array The original sequence of strings.
1132          */
1133         function orig() {
1134                 $lines = array();
1135
1136                 foreach ($this->edits as $edit) {
1137                         if ($edit->orig)
1138                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1139                 }
1140                 return $lines;
1141         }
1142
1143         /**
1144          * Get the closing set of lines.
1145          *
1146          * This reconstructs the $to_lines parameter passed to the
1147          * constructor.
1148          *
1149          * @return array The sequence of strings.
1150          */
1151         function closing() {
1152                 $lines = array();
1153
1154                 foreach ($this->edits as $edit) {
1155                         if ($edit->closing)
1156                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1157                 }
1158                 return $lines;
1159         }
1160
1161         /**
1162          * Check a Diff for validity.
1163          *
1164          * This is here only for debugging purposes.
1165          */
1166         function _check ($from_lines, $to_lines) {
1167                 $fname = 'Diff::_check';
1168                 wfProfileIn( $fname );
1169                 if (serialize($from_lines) != serialize($this->orig()))
1170                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1171                 if (serialize($to_lines) != serialize($this->closing()))
1172                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1173
1174                 $rev = $this->reverse();
1175                 if (serialize($to_lines) != serialize($rev->orig()))
1176                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1177                 if (serialize($from_lines) != serialize($rev->closing()))
1178                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1179
1180
1181                 $prevtype = 'none';
1182                 foreach ($this->edits as $edit) {
1183                         if ( $prevtype == $edit->type )
1184                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1185                         $prevtype = $edit->type;
1186                 }
1187
1188                 $lcs = $this->lcs();
1189                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1190                 wfProfileOut( $fname );
1191         }
1192 }
1193
1194 /**
1195  * FIXME: bad name.
1196  * @todo document
1197  * @access private
1198  * @package MediaWiki
1199  * @subpackage DifferenceEngine
1200  */
1201 class MappedDiff extends Diff
1202 {
1203         /**
1204          * Constructor.
1205          *
1206          * Computes diff between sequences of strings.
1207          *
1208          * This can be used to compute things like
1209          * case-insensitve diffs, or diffs which ignore
1210          * changes in white-space.
1211          *
1212          * @param $from_lines array An array of strings.
1213          *      (Typically these are lines from a file.)
1214          *
1215          * @param $to_lines array An array of strings.
1216          *
1217          * @param $mapped_from_lines array This array should
1218          *      have the same size number of elements as $from_lines.
1219          *      The elements in $mapped_from_lines and
1220          *      $mapped_to_lines are what is actually compared
1221          *      when computing the diff.
1222          *
1223          * @param $mapped_to_lines array This array should
1224          *      have the same number of elements as $to_lines.
1225          */
1226         function MappedDiff($from_lines, $to_lines,
1227                                                 $mapped_from_lines, $mapped_to_lines) {
1228                 $fname = 'MappedDiff::MappedDiff';
1229                 wfProfileIn( $fname );
1230
1231                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1232                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1233
1234                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1235
1236                 $xi = $yi = 0;
1237                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1238                         $orig = &$this->edits[$i]->orig;
1239                         if (is_array($orig)) {
1240                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1241                                 $xi += sizeof($orig);
1242                         }
1243
1244                         $closing = &$this->edits[$i]->closing;
1245                         if (is_array($closing)) {
1246                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1247                                 $yi += sizeof($closing);
1248                         }
1249                 }
1250                 wfProfileOut( $fname );
1251         }
1252 }
1253
1254 /**
1255  * A class to format Diffs
1256  *
1257  * This class formats the diff in classic diff format.
1258  * It is intended that this class be customized via inheritance,
1259  * to obtain fancier outputs.
1260  * @todo document
1261  * @access private
1262  * @package MediaWiki
1263  * @subpackage DifferenceEngine
1264  */
1265 class DiffFormatter
1266 {
1267         /**
1268          * Number of leading context "lines" to preserve.
1269          *
1270          * This should be left at zero for this class, but subclasses
1271          * may want to set this to other values.
1272          */
1273         var $leading_context_lines = 0;
1274
1275         /**
1276          * Number of trailing context "lines" to preserve.
1277          *
1278          * This should be left at zero for this class, but subclasses
1279          * may want to set this to other values.
1280          */
1281         var $trailing_context_lines = 0;
1282
1283         /**
1284          * Format a diff.
1285          *
1286          * @param $diff object A Diff object.
1287          * @return string The formatted output.
1288          */
1289         function format($diff) {
1290                 $fname = 'DiffFormatter::format';
1291                 wfProfileIn( $fname );
1292
1293                 $xi = $yi = 1;
1294                 $block = false;
1295                 $context = array();
1296
1297                 $nlead = $this->leading_context_lines;
1298                 $ntrail = $this->trailing_context_lines;
1299
1300                 $this->_start_diff();
1301
1302                 foreach ($diff->edits as $edit) {
1303                         if ($edit->type == 'copy') {
1304                                 if (is_array($block)) {
1305                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1306                                                 $block[] = $edit;
1307                                         }
1308                                         else{
1309                                                 if ($ntrail) {
1310                                                         $context = array_slice($edit->orig, 0, $ntrail);
1311                                                         $block[] = new _DiffOp_Copy($context);
1312                                                 }
1313                                                 $this->_block($x0, $ntrail + $xi - $x0,
1314                                                                           $y0, $ntrail + $yi - $y0,
1315                                                                           $block);
1316                                                 $block = false;
1317                                         }
1318                                 }
1319                                 $context = $edit->orig;
1320                         }
1321                         else {
1322                                 if (! is_array($block)) {
1323                                         $context = array_slice($context, sizeof($context) - $nlead);
1324                                         $x0 = $xi - sizeof($context);
1325                                         $y0 = $yi - sizeof($context);
1326                                         $block = array();
1327                                         if ($context)
1328                                                 $block[] = new _DiffOp_Copy($context);
1329                                 }
1330                                 $block[] = $edit;
1331                         }
1332
1333                         if ($edit->orig)
1334                                 $xi += sizeof($edit->orig);
1335                         if ($edit->closing)
1336                                 $yi += sizeof($edit->closing);
1337                 }
1338
1339                 if (is_array($block))
1340                         $this->_block($x0, $xi - $x0,
1341                                                   $y0, $yi - $y0,
1342                                                   $block);
1343
1344                 $end = $this->_end_diff();
1345                 wfProfileOut( $fname );
1346                 return $end;
1347         }
1348
1349         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1350                 $fname = 'DiffFormatter::_block';
1351                 wfProfileIn( $fname );
1352                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1353                 foreach ($edits as $edit) {
1354                         if ($edit->type == 'copy')
1355                                 $this->_context($edit->orig);
1356                         elseif ($edit->type == 'add')
1357                                 $this->_added($edit->closing);
1358                         elseif ($edit->type == 'delete')
1359                                 $this->_deleted($edit->orig);
1360                         elseif ($edit->type == 'change')
1361                                 $this->_changed($edit->orig, $edit->closing);
1362                         else
1363                                 trigger_error('Unknown edit type', E_USER_ERROR);
1364                 }
1365                 $this->_end_block();
1366                 wfProfileOut( $fname );
1367         }
1368
1369         function _start_diff() {
1370                 ob_start();
1371         }
1372
1373         function _end_diff() {
1374                 $val = ob_get_contents();
1375                 ob_end_clean();
1376                 return $val;
1377         }
1378
1379         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1380                 if ($xlen > 1)
1381                         $xbeg .= "," . ($xbeg + $xlen - 1);
1382                 if ($ylen > 1)
1383                         $ybeg .= "," . ($ybeg + $ylen - 1);
1384
1385                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1386         }
1387
1388         function _start_block($header) {
1389                 echo $header;
1390         }
1391
1392         function _end_block() {
1393         }
1394
1395         function _lines($lines, $prefix = ' ') {
1396                 foreach ($lines as $line)
1397                         echo "$prefix $line\n";
1398         }
1399
1400         function _context($lines) {
1401                 $this->_lines($lines);
1402         }
1403
1404         function _added($lines) {
1405                 $this->_lines($lines, '>');
1406         }
1407         function _deleted($lines) {
1408                 $this->_lines($lines, '<');
1409         }
1410
1411         function _changed($orig, $closing) {
1412                 $this->_deleted($orig);
1413                 echo "---\n";
1414                 $this->_added($closing);
1415         }
1416 }
1417
1418
1419 /**
1420  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1421  *
1422  */
1423
1424 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1425
1426 /**
1427  * @todo document
1428  * @access private
1429  * @package MediaWiki
1430  * @subpackage DifferenceEngine
1431  */
1432 class _HWLDF_WordAccumulator {
1433         function _HWLDF_WordAccumulator () {
1434                 $this->_lines = array();
1435                 $this->_line = '';
1436                 $this->_group = '';
1437                 $this->_tag = '';
1438         }
1439
1440         function _flushGroup ($new_tag) {
1441                 if ($this->_group !== '') {
1442                         if ($this->_tag == 'mark')
1443                                 $this->_line .= '<span class="diffchange">' .
1444                                         htmlspecialchars ( $this->_group ) . '</span>';
1445                         else
1446                                 $this->_line .= htmlspecialchars ( $this->_group );
1447                 }
1448                 $this->_group = '';
1449                 $this->_tag = $new_tag;
1450         }
1451
1452         function _flushLine ($new_tag) {
1453                 $this->_flushGroup($new_tag);
1454                 if ($this->_line != '')
1455                         array_push ( $this->_lines, $this->_line );
1456                 else
1457                         # make empty lines visible by inserting an NBSP
1458                         array_push ( $this->_lines, NBSP );
1459                 $this->_line = '';
1460         }
1461
1462         function addWords ($words, $tag = '') {
1463                 if ($tag != $this->_tag)
1464                         $this->_flushGroup($tag);
1465
1466                 foreach ($words as $word) {
1467                         // new-line should only come as first char of word.
1468                         if ($word == '')
1469                                 continue;
1470                         if ($word[0] == "\n") {
1471                                 $this->_flushLine($tag);
1472                                 $word = substr($word, 1);
1473                         }
1474                         assert(!strstr($word, "\n"));
1475                         $this->_group .= $word;
1476                 }
1477         }
1478
1479         function getLines() {
1480                 $this->_flushLine('~done');
1481                 return $this->_lines;
1482         }
1483 }
1484
1485 /**
1486  * @todo document
1487  * @access private
1488  * @package MediaWiki
1489  * @subpackage DifferenceEngine
1490  */
1491 class WordLevelDiff extends MappedDiff
1492 {
1493         function WordLevelDiff ($orig_lines, $closing_lines) {
1494                 $fname = 'WordLevelDiff::WordLevelDiff';
1495                 wfProfileIn( $fname );
1496
1497                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1498                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1499
1500                 $this->MappedDiff($orig_words, $closing_words,
1501                                                   $orig_stripped, $closing_stripped);
1502                 wfProfileOut( $fname );
1503         }
1504
1505         function _split($lines) {
1506                 $fname = 'WordLevelDiff::_split';
1507                 wfProfileIn( $fname );
1508
1509                 $words = array();
1510                 $stripped = array();
1511                 $first = true;
1512                 foreach ( $lines as $line ) {
1513                         # If the line is too long, just pretend the entire line is one big word
1514                         # This prevents resource exhaustion problems
1515                         if ( $first ) {
1516                                 $first = false;
1517                         } else {
1518                                 $words[] = "\n";
1519                                 $stripped[] = "\n";
1520                         }
1521                         if ( strlen( $line ) > MAX_DIFF_LINE ) {
1522                                 $words[] = $line;
1523                                 $stripped[] = $line;
1524                         } else {
1525                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1526                                         $line, $m))
1527                                 {
1528                                         $words = array_merge( $words, $m[0] );
1529                                         $stripped = array_merge( $stripped, $m[1] );
1530                                 }
1531                         }
1532                 }
1533                 wfProfileOut( $fname );
1534                 return array($words, $stripped);
1535         }
1536
1537         function orig () {
1538                 $fname = 'WordLevelDiff::orig';
1539                 wfProfileIn( $fname );
1540                 $orig = new _HWLDF_WordAccumulator;
1541
1542                 foreach ($this->edits as $edit) {
1543                         if ($edit->type == 'copy')
1544                                 $orig->addWords($edit->orig);
1545                         elseif ($edit->orig)
1546                                 $orig->addWords($edit->orig, 'mark');
1547                 }
1548                 $lines = $orig->getLines();
1549                 wfProfileOut( $fname );
1550                 return $lines;
1551         }
1552
1553         function closing () {
1554                 $fname = 'WordLevelDiff::closing';
1555                 wfProfileIn( $fname );
1556                 $closing = new _HWLDF_WordAccumulator;
1557
1558                 foreach ($this->edits as $edit) {
1559                         if ($edit->type == 'copy')
1560                                 $closing->addWords($edit->closing);
1561                         elseif ($edit->closing)
1562                                 $closing->addWords($edit->closing, 'mark');
1563                 }
1564                 $lines = $closing->getLines();
1565                 wfProfileOut( $fname );
1566                 return $lines;
1567         }
1568 }
1569
1570 /**
1571  *      Wikipedia Table style diff formatter.
1572  * @todo document
1573  * @access private
1574  * @package MediaWiki
1575  * @subpackage DifferenceEngine
1576  */
1577 class TableDiffFormatter extends DiffFormatter
1578 {
1579         function TableDiffFormatter() {
1580                 $this->leading_context_lines = 2;
1581                 $this->trailing_context_lines = 2;
1582         }
1583
1584         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1585                 $l1 = wfMsg( 'lineno', $xbeg );
1586                 $l2 = wfMsg( 'lineno', $ybeg );
1587
1588                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1589                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1590                 return $r;
1591         }
1592
1593         function _start_block( $header ) {
1594                 global $wgOut;
1595                 echo $header;
1596         }
1597
1598         function _end_block() {
1599         }
1600
1601         function _lines( $lines, $prefix=' ', $color='white' ) {
1602         }
1603
1604         # HTML-escape parameter before calling this
1605         function addedLine( $line ) {
1606                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1607         }
1608
1609         # HTML-escape parameter before calling this
1610         function deletedLine( $line ) {
1611                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1612         }
1613
1614         # HTML-escape parameter before calling this
1615         function contextLine( $line ) {
1616                 return "<td> </td><td class='diff-context'>{$line}</td>";
1617         }
1618
1619         function emptyLine() {
1620                 return '<td colspan="2">&nbsp;</td>';
1621         }
1622
1623         function _added( $lines ) {
1624                 foreach ($lines as $line) {
1625                         echo '<tr>' . $this->emptyLine() .
1626                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1627                 }
1628         }
1629
1630         function _deleted($lines) {
1631                 foreach ($lines as $line) {
1632                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1633                           $this->emptyLine() . "</tr>\n";
1634                 }
1635         }
1636
1637         function _context( $lines ) {
1638                 foreach ($lines as $line) {
1639                         echo '<tr>' .
1640                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1641                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1642                 }
1643         }
1644
1645         function _changed( $orig, $closing ) {
1646                 $fname = 'TableDiffFormatter::_changed';
1647                 wfProfileIn( $fname );
1648
1649                 $diff = new WordLevelDiff( $orig, $closing );
1650                 $del = $diff->orig();
1651                 $add = $diff->closing();
1652
1653                 # Notice that WordLevelDiff returns HTML-escaped output.
1654                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1655
1656                 while ( $line = array_shift( $del ) ) {
1657                         $aline = array_shift( $add );
1658                         echo '<tr>' . $this->deletedLine( $line ) .
1659                                 $this->addedLine( $aline ) . "</tr>\n";
1660                 }
1661                 foreach ($add as $line) {       # If any leftovers
1662                         echo '<tr>' . $this->emptyLine() .
1663                                 $this->addedLine( $line ) . "</tr>\n";
1664                 }
1665                 wfProfileOut( $fname );
1666         }
1667 }
1668
1669 ?>