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->addSecondaryWikiText( $this->mNewtext );
 211
 212                 if( !$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->addSecondaryWikiText( $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                                 wfIncrStats( 'diff_cache_hit' );
 328                                 $difftext .= "\n<!-- diff cache key $key -->\n";
 329                                 return $difftext;
 330                         }
 331                 }
 332
 333                 if ( !$this->loadText() ) {
 334                         return false;
 335                 }
 336
 337                 $otext = $wgContLang->segmentForDiff($this->mOldtext);
 338                 $ntext = $wgContLang->segmentForDiff($this->mNewtext);
 339                 if ( $wgUseExternalDiffEngine ) {
 340                         # For historical reasons, external diff engine expects
 341                         # input text to be HTML-escaped already
 342                         $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
 343                         $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
 344                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 345                                 dl('php_wikidiff.so');
 346                         }
 347                         $difftext = wikidiff_do_diff( $otext, $ntext, 2 );
 348                 } else {
 349                         $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
 350                         $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
 351                         $diffs =& new Diff( $ota, $nta );
 352                         $formatter =& new TableDiffFormatter();
 353                         $difftext = $formatter->format( $diffs );
 354                 }
 355                 $difftext = $wgContLang->unsegmentForDiff($difftext);
 356
 357                 // Save to cache for 7 days
 358                 if ( $key !== false ) {
 359                         wfIncrStats( 'diff_cache_miss' );
 360                         $wgMemc->set( $key, $difftext, 7*86400 );
 361                 } else {
 362                         wfIncrStats( 'diff_uncacheable' );
 363                 }
 364                 return $difftext;
 365         }
 366
 367         /**
 368          * Add the header to a diff body
 369          */
 370         function addHeader( $diff, $otitle, $ntitle ) {
 371                 $out = "
 372                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 373                         <tr>
 374                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 375                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 376                         </tr>
 377                         $diff
 378                         </table>
 379                 ";
 380                 return $out;
 381         }
 382
 383         /**
 384          * Use specified text instead of loading from the database
 385          */
 386         function setText( $oldText, $newText ) {
 387                 $this->mOldtext = $oldText;
 388                 $this->mNewtext = $newText;
 389                 $this->mTextLoaded = 2;
 390         }
 391
 392         /**
 393          * Load revision metadata for the specified articles. If newid is 0, then compare
 394          * the old article in oldid to the current article; if oldid is 0, then
 395          * compare the current article to the immediately previous one (ignoring the
 396          * value of newid).
 397          *
 398          * If oldid is false, leave the corresponding revision object set
 399          * to false. This is impossible via ordinary user input, and is provided for
 400          * API convenience.
 401          */
 402         function loadRevisionData() {
 403                 global $wgLang;
 404                 if ( $this->mRevisionsLoaded ) {
 405                         return true;
 406                 } else {
 407                         // Whether it succeeds or fails, we don't want to try again
 408                         $this->mRevisionsLoaded = true;
 409                 }
 410
 411                 // Load the new revision object
 412                 if( $this->mNewid ) {
 413                         $this->mNewRev = Revision::newFromId( $this->mNewid );
 414                 } else {
 415                         $this->mNewRev = Revision::newFromTitle( $this->mTitle );
 416                 }
 417
 418                 if( is_null( $this->mNewRev ) ) {
 419                         return false;
 420                 }
 421
 422                 // Set assorted variables
 423                 if( $this->mNewRev->isCurrent() ) {
 424                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 425                         $this->mNewPage = $this->mTitle;
 426                         $newLink = $this->mNewPage->escapeLocalUrl();
 427                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 428                 } else {
 429                         $this->mNewPage = $this->mNewRev->getTitle();
 430                         $newLink = $this->mNewPage->escapeLocalUrl ('oldid=' . $this->mNewid );
 431                         $t = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
 432                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 433                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 434                 }
 435
 436                 $this->mNewUser = $this->mNewRev->getUserText();
 437                 $this->mNewComment = $this->mNewRev->getComment();
 438
 439                 // Load the old revision object
 440                 $this->mOldRev = false;
 441                 if( $this->mOldid ) {
 442                         $this->mOldRev = Revision::newFromId( $this->mOldid );
 443                 } elseif ( $this->mOldid === 0 ) {
 444                         $this->mOldRev = $this->mNewRev->getPrevious();
 445                         $this->mOldid = $this->mOldRev->getId();
 446                 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
 447
 448                 if( is_null( $this->mOldRev ) ) {
 449                         return false;
 450                 }
 451
 452                 if ( $this->mOldRev ) {
 453                         $this->mOldPage = $this->mOldRev->getTitle();
 454
 455                         $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
 456                         $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
 457                         $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
 458
 459
 460                         $this->mOldUser = $this->mOldRev->getUserText();
 461                         $this->mOldComment = $this->mOldRev->getComment();
 462                 }
 463
 464                 return true;
 465         }
 466
 467         /**
 468          * Load the text of the revisions, as well as revision data.
 469          */
 470         function loadText() {
 471                 if ( $this->mTextLoaded == 2 ) {
 472                         return true;
 473                 } else {
 474                         // Whether it succeeds or fails, we don't want to try again
 475                         $this->mTextLoaded = 2;
 476                 }
 477
 478                 if ( !$this->loadRevisionData() ) {
 479                         return false;
 480                 }
 481                 if ( $this->mOldRev ) {
 482                         $this->mOldtext = $this->mOldRev->getText();
 483                         if ( $this->mOldtext === false ) {
 484                                 return false;
 485                         }
 486                 }
 487                 if ( $this->mNewRev ) {
 488                         $this->mNewtext = $this->mNewRev->getText();
 489                         if ( $this->mNewtext === false ) {
 490                                 return false;
 491                         }
 492                 }
 493                 return true;
 494         }
 495
 496         /**
 497          * Load the text of the new revision, not the old one
 498          */
 499         function loadNewText() {
 500                 if ( $this->mTextLoaded >= 1 ) {
 501                         return true;
 502                 } else {
 503                         $this->mTextLoaded = 1;
 504                 }
 505                 if ( !$this->loadRevisionData() ) {
 506                         return false;
 507                 }
 508                 $this->mNewtext = $this->mNewRev->getText();
 509                 return true;
 510         }
 511
 512
 513 }
 514
 515 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 516 //
 517 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 518 // You may copy this code freely under the conditions of the GPL.
 519 //
 520
 521 define('USE_ASSERTS', function_exists('assert'));
 522
 523 /**
 524  * @todo document
 525  * @access private
 526  * @package MediaWiki
 527  * @subpackage DifferenceEngine
 528  */
 529 class _DiffOp {
 530         var $type;
 531         var $orig;
 532         var $closing;
 533
 534         function reverse() {
 535                 trigger_error('pure virtual', E_USER_ERROR);
 536         }
 537
 538         function norig() {
 539                 return $this->orig ? sizeof($this->orig) : 0;
 540         }
 541
 542         function nclosing() {
 543                 return $this->closing ? sizeof($this->closing) : 0;
 544         }
 545 }
 546
 547 /**
 548  * @todo document
 549  * @access private
 550  * @package MediaWiki
 551  * @subpackage DifferenceEngine
 552  */
 553 class _DiffOp_Copy extends _DiffOp {
 554         var $type = 'copy';
 555
 556         function _DiffOp_Copy ($orig, $closing = false) {
 557                 if (!is_array($closing))
 558                         $closing = $orig;
 559                 $this->orig = $orig;
 560                 $this->closing = $closing;
 561         }
 562
 563         function reverse() {
 564                 return new _DiffOp_Copy($this->closing, $this->orig);
 565         }
 566 }
 567
 568 /**
 569  * @todo document
 570  * @access private
 571  * @package MediaWiki
 572  * @subpackage DifferenceEngine
 573  */
 574 class _DiffOp_Delete extends _DiffOp {
 575         var $type = 'delete';
 576
 577         function _DiffOp_Delete ($lines) {
 578                 $this->orig = $lines;
 579                 $this->closing = false;
 580         }
 581
 582         function reverse() {
 583                 return new _DiffOp_Add($this->orig);
 584         }
 585 }
 586
 587 /**
 588  * @todo document
 589  * @access private
 590  * @package MediaWiki
 591  * @subpackage DifferenceEngine
 592  */
 593 class _DiffOp_Add extends _DiffOp {
 594         var $type = 'add';
 595
 596         function _DiffOp_Add ($lines) {
 597                 $this->closing = $lines;
 598                 $this->orig = false;
 599         }
 600
 601         function reverse() {
 602                 return new _DiffOp_Delete($this->closing);
 603         }
 604 }
 605
 606 /**
 607  * @todo document
 608  * @access private
 609  * @package MediaWiki
 610  * @subpackage DifferenceEngine
 611  */
 612 class _DiffOp_Change extends _DiffOp {
 613         var $type = 'change';
 614
 615         function _DiffOp_Change ($orig, $closing) {
 616                 $this->orig = $orig;
 617                 $this->closing = $closing;
 618         }
 619
 620         function reverse() {
 621                 return new _DiffOp_Change($this->closing, $this->orig);
 622         }
 623 }
 624
 625
 626 /**
 627  * Class used internally by Diff to actually compute the diffs.
 628  *
 629  * The algorithm used here is mostly lifted from the perl module
 630  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 631  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 632  *
 633  * More ideas are taken from:
 634  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 635  *
 636  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 637  * diffutils-2.7, which can be found at:
 638  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 639  *
 640  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 641  * are my own.
 642  *
 643  * Line length limits for robustness added by Tim Starling, 2005-08-31
 644  *
 645  * @author Geoffrey T. Dairiki, Tim Starling
 646  * @access private
 647  * @package MediaWiki
 648  * @subpackage DifferenceEngine
 649  */
 650 class _DiffEngine
 651 {
 652         function diff ($from_lines, $to_lines) {
 653                 $fname = '_DiffEngine::diff';
 654                 wfProfileIn( $fname );
 655
 656                 $n_from = sizeof($from_lines);
 657                 $n_to = sizeof($to_lines);
 658
 659                 $this->xchanged = $this->ychanged = array();
 660                 $this->xv = $this->yv = array();
 661                 $this->xind = $this->yind = array();
 662                 unset($this->seq);
 663                 unset($this->in_seq);
 664                 unset($this->lcs);
 665
 666                 // Skip leading common lines.
 667                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 668                         if ($from_lines[$skip] !== $to_lines[$skip])
 669                                 break;
 670                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 671                 }
 672                 // Skip trailing common lines.
 673                 $xi = $n_from; $yi = $n_to;
 674                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 675                         if ($from_lines[$xi] !== $to_lines[$yi])
 676                                 break;
 677                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 678                 }
 679
 680                 // Ignore lines which do not exist in both files.
 681                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 682                         $xhash[$this->_line_hash($from_lines[$xi])] = 1;
 683                 }
 684
 685                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 686                         $line = $to_lines[$yi];
 687                         if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
 688                                 continue;
 689                         $yhash[$this->_line_hash($line)] = 1;
 690                         $this->yv[] = $line;
 691                         $this->yind[] = $yi;
 692                 }
 693                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 694                         $line = $from_lines[$xi];
 695                         if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
 696                                 continue;
 697                         $this->xv[] = $line;
 698                         $this->xind[] = $xi;
 699                 }
 700
 701                 // Find the LCS.
 702                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 703
 704                 // Merge edits when possible
 705                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 706                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 707
 708                 // Compute the edit operations.
 709                 $edits = array();
 710                 $xi = $yi = 0;
 711                 while ($xi < $n_from || $yi < $n_to) {
 712                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 713                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 714
 715                         // Skip matching "snake".
 716                         $copy = array();
 717                         while ( $xi < $n_from && $yi < $n_to
 718                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 719                                 $copy[] = $from_lines[$xi++];
 720                                 ++$yi;
 721                         }
 722                         if ($copy)
 723                                 $edits[] = new _DiffOp_Copy($copy);
 724
 725                         // Find deletes & adds.
 726                         $delete = array();
 727                         while ($xi < $n_from && $this->xchanged[$xi])
 728                                 $delete[] = $from_lines[$xi++];
 729
 730                         $add = array();
 731                         while ($yi < $n_to && $this->ychanged[$yi])
 732                                 $add[] = $to_lines[$yi++];
 733
 734                         if ($delete && $add)
 735                                 $edits[] = new _DiffOp_Change($delete, $add);
 736                         elseif ($delete)
 737                                 $edits[] = new _DiffOp_Delete($delete);
 738                         elseif ($add)
 739                                 $edits[] = new _DiffOp_Add($add);
 740                 }
 741                 wfProfileOut( $fname );
 742                 return $edits;
 743         }
 744
 745         /**
 746          * Returns the whole line if it's small enough, or the MD5 hash otherwise
 747          */
 748         function _line_hash( $line ) {
 749                 if ( strlen( $line ) > MAX_DIFF_XREF_LENGTH ) {
 750                         return md5( $line );
 751                 } else {
 752                         return $line;
 753                 }
 754         }
 755
 756
 757         /* Divide the Largest Common Subsequence (LCS) of the sequences
 758          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 759          * sized segments.
 760          *
 761          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 762          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 763          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 764          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 765          * that (X0, Y0) == (XOFF, YOFF) and
 766          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 767          *
 768          * This function assumes that the first lines of the specified portions
 769          * of the two files do not match, and likewise that the last lines do not
 770          * match.  The caller must trim matching lines from the beginning and end
 771          * of the portions it is going to specify.
 772          */
 773         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 774                 $fname = '_DiffEngine::_diag';
 775                 wfProfileIn( $fname );
 776                 $flip = false;
 777
 778                 if ($xlim - $xoff > $ylim - $yoff) {
 779                         // Things seems faster (I'm not sure I understand why)
 780                                 // when the shortest sequence in X.
 781                                 $flip = true;
 782                         list ($xoff, $xlim, $yoff, $ylim)
 783                         = array( $yoff, $ylim, $xoff, $xlim);
 784                 }
 785
 786                 if ($flip)
 787                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 788                                 $ymatches[$this->xv[$i]][] = $i;
 789                 else
 790                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 791                                 $ymatches[$this->yv[$i]][] = $i;
 792
 793                 $this->lcs = 0;
 794                 $this->seq[0]= $yoff - 1;
 795                 $this->in_seq = array();
 796                 $ymids[0] = array();
 797
 798                 $numer = $xlim - $xoff + $nchunks - 1;
 799                 $x = $xoff;
 800                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 801                         wfProfileIn( "$fname-chunk" );
 802                         if ($chunk > 0)
 803                                 for ($i = 0; $i <= $this->lcs; $i++)
 804                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 805
 806                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 807                         for ( ; $x < $x1; $x++) {
 808                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 809                                         if (empty($ymatches[$line]))
 810                                                 continue;
 811                                 $matches = $ymatches[$line];
 812                                 reset($matches);
 813                                 while (list ($junk, $y) = each($matches))
 814                                         if (empty($this->in_seq[$y])) {
 815                                                 $k = $this->_lcs_pos($y);
 816                                                 USE_ASSERTS && assert($k > 0);
 817                                                 $ymids[$k] = $ymids[$k-1];
 818                                                 break;
 819                                         }
 820                                 while (list ($junk, $y) = each($matches)) {
 821                                         if ($y > $this->seq[$k-1]) {
 822                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 823                                                 // Optimization: this is a common case:
 824                                                 //      next match is just replacing previous match.
 825                                                 $this->in_seq[$this->seq[$k]] = false;
 826                                                 $this->seq[$k] = $y;
 827                                                 $this->in_seq[$y] = 1;
 828                                         } else if (empty($this->in_seq[$y])) {
 829                                                 $k = $this->_lcs_pos($y);
 830                                                 USE_ASSERTS && assert($k > 0);
 831                                                 $ymids[$k] = $ymids[$k-1];
 832                                         }
 833                                 }
 834                         }
 835                         wfProfileOut( "$fname-chunk" );
 836                 }
 837
 838                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 839                 $ymid = $ymids[$this->lcs];
 840                 for ($n = 0; $n < $nchunks - 1; $n++) {
 841                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 842                         $y1 = $ymid[$n] + 1;
 843                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 844                 }
 845                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 846
 847                 wfProfileOut( $fname );
 848                 return array($this->lcs, $seps);
 849         }
 850
 851         function _lcs_pos ($ypos) {
 852                 $fname = '_DiffEngine::_lcs_pos';
 853                 wfProfileIn( $fname );
 854
 855                 $end = $this->lcs;
 856                 if ($end == 0 || $ypos > $this->seq[$end]) {
 857                         $this->seq[++$this->lcs] = $ypos;
 858                         $this->in_seq[$ypos] = 1;
 859                         wfProfileOut( $fname );
 860                         return $this->lcs;
 861                 }
 862
 863                 $beg = 1;
 864                 while ($beg < $end) {
 865                         $mid = (int)(($beg + $end) / 2);
 866                         if ( $ypos > $this->seq[$mid] )
 867                                 $beg = $mid + 1;
 868                         else
 869                                 $end = $mid;
 870                 }
 871
 872                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 873
 874                 $this->in_seq[$this->seq[$end]] = false;
 875                 $this->seq[$end] = $ypos;
 876                 $this->in_seq[$ypos] = 1;
 877                 wfProfileOut( $fname );
 878                 return $end;
 879         }
 880
 881         /* Find LCS of two sequences.
 882          *
 883          * The results are recorded in the vectors $this->{x,y}changed[], by
 884          * storing a 1 in the element for each line that is an insertion
 885          * or deletion (ie. is not in the LCS).
 886          *
 887          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 888          *
 889          * Note that XLIM, YLIM are exclusive bounds.
 890          * All line numbers are origin-0 and discarded lines are not counted.
 891          */
 892         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 893                 $fname = '_DiffEngine::_compareseq';
 894                 wfProfileIn( $fname );
 895
 896                 // Slide down the bottom initial diagonal.
 897                 while ($xoff < $xlim && $yoff < $ylim
 898                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 899                         ++$xoff;
 900                         ++$yoff;
 901                 }
 902
 903                 // Slide up the top initial diagonal.
 904                 while ($xlim > $xoff && $ylim > $yoff
 905                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 906                         --$xlim;
 907                         --$ylim;
 908                 }
 909
 910                 if ($xoff == $xlim || $yoff == $ylim)
 911                         $lcs = 0;
 912                 else {
 913                         // This is ad hoc but seems to work well.
 914                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 915                         //$nchunks = max(2,min(8,(int)$nchunks));
 916                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 917                         list ($lcs, $seps)
 918                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 919                 }
 920
 921                 if ($lcs == 0) {
 922                         // X and Y sequences have no common subsequence:
 923                         // mark all changed.
 924                         while ($yoff < $ylim)
 925                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 926                         while ($xoff < $xlim)
 927                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 928                 } else {
 929                         // Use the partitions to split this problem into subproblems.
 930                         reset($seps);
 931                         $pt1 = $seps[0];
 932                         while ($pt2 = next($seps)) {
 933                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 934                                 $pt1 = $pt2;
 935                         }
 936                 }
 937                 wfProfileOut( $fname );
 938         }
 939
 940         /* Adjust inserts/deletes of identical lines to join changes
 941          * as much as possible.
 942          *
 943          * We do something when a run of changed lines include a
 944          * line at one end and has an excluded, identical line at the other.
 945          * We are free to choose which identical line is included.
 946          * `compareseq' usually chooses the one at the beginning,
 947          * but usually it is cleaner to consider the following identical line
 948          * to be the "change".
 949          *
 950          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 951          */
 952         function _shift_boundaries ($lines, &$changed, $other_changed) {
 953                 $fname = '_DiffEngine::_shift_boundaries';
 954                 wfProfileIn( $fname );
 955                 $i = 0;
 956                 $j = 0;
 957
 958                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 959                 $len = sizeof($lines);
 960                 $other_len = sizeof($other_changed);
 961
 962                 while (1) {
 963                         /*
 964                          * Scan forwards to find beginning of another run of changes.
 965                          * Also keep track of the corresponding point in the other file.
 966                          *
 967                          * Throughout this code, $i and $j are adjusted together so that
 968                          * the first $i elements of $changed and the first $j elements
 969                          * of $other_changed both contain the same number of zeros
 970                          * (unchanged lines).
 971                          * Furthermore, $j is always kept so that $j == $other_len or
 972                          * $other_changed[$j] == false.
 973                          */
 974                         while ($j < $other_len && $other_changed[$j])
 975                                 $j++;
 976
 977                         while ($i < $len && ! $changed[$i]) {
 978                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 979                                 $i++; $j++;
 980                                 while ($j < $other_len && $other_changed[$j])
 981                                         $j++;
 982                         }
 983
 984                         if ($i == $len)
 985                                 break;
 986
 987                         $start = $i;
 988
 989                         // Find the end of this run of changes.
 990                         while (++$i < $len && $changed[$i])
 991                                 continue;
 992
 993                         do {
 994                                 /*
 995                                  * Record the length of this run of changes, so that
 996                                  * we can later determine whether the run has grown.
 997                                  */
 998                                 $runlength = $i - $start;
 999
1000                                 /*
1001                                  * Move the changed region back, so long as the
1002                                  * previous unchanged line matches the last changed one.
1003                                  * This merges with previous changed regions.
1004                                  */
1005                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1006                                         $changed[--$start] = 1;
1007                                         $changed[--$i] = false;
1008                                         while ($start > 0 && $changed[$start - 1])
1009                                                 $start--;
1010                                         USE_ASSERTS && assert('$j > 0');
1011                                         while ($other_changed[--$j])
1012                                                 continue;
1013                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1014                                 }
1015
1016                                 /*
1017                                  * Set CORRESPONDING to the end of the changed run, at the last
1018                                  * point where it corresponds to a changed run in the other file.
1019                                  * CORRESPONDING == LEN means no such point has been found.
1020                                  */
1021                                 $corresponding = $j < $other_len ? $i : $len;
1022
1023                                 /*
1024                                  * Move the changed region forward, so long as the
1025                                  * first changed line matches the following unchanged one.
1026                                  * This merges with following changed regions.
1027                                  * Do this second, so that if there are no merges,
1028                                  * the changed region is moved forward as far as possible.
1029                                  */
1030                                 while ($i < $len && $lines[$start] == $lines[$i]) {
1031                                         $changed[$start++] = false;
1032                                         $changed[$i++] = 1;
1033                                         while ($i < $len && $changed[$i])
1034                                                 $i++;
1035
1036                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1037                                         $j++;
1038                                         if ($j < $other_len && $other_changed[$j]) {
1039                                                 $corresponding = $i;
1040                                                 while ($j < $other_len && $other_changed[$j])
1041                                                         $j++;
1042                                         }
1043                                 }
1044                         } while ($runlength != $i - $start);
1045
1046                         /*
1047                          * If possible, move the fully-merged run of changes
1048                          * back to a corresponding run in the other file.
1049                          */
1050                         while ($corresponding < $i) {
1051                                 $changed[--$start] = 1;
1052                                 $changed[--$i] = 0;
1053                                 USE_ASSERTS && assert('$j > 0');
1054                                 while ($other_changed[--$j])
1055                                         continue;
1056                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1057                         }
1058                 }
1059                 wfProfileOut( $fname );
1060         }
1061 }
1062
1063 /**
1064  * Class representing a 'diff' between two sequences of strings.
1065  * @todo document
1066  * @access private
1067  * @package MediaWiki
1068  * @subpackage DifferenceEngine
1069  */
1070 class Diff
1071 {
1072         var $edits;
1073
1074         /**
1075          * Constructor.
1076          * Computes diff between sequences of strings.
1077          *
1078          * @param $from_lines array An array of strings.
1079          *                (Typically these are lines from a file.)
1080          * @param $to_lines array An array of strings.
1081          */
1082         function Diff($from_lines, $to_lines) {
1083                 $eng = new _DiffEngine;
1084                 $this->edits = $eng->diff($from_lines, $to_lines);
1085                 //$this->_check($from_lines, $to_lines);
1086         }
1087
1088         /**
1089          * Compute reversed Diff.
1090          *
1091          * SYNOPSIS:
1092          *
1093          *      $diff = new Diff($lines1, $lines2);
1094          *      $rev = $diff->reverse();
1095          * @return object A Diff object representing the inverse of the
1096          *                                original diff.
1097          */
1098         function reverse () {
1099                 $rev = $this;
1100                 $rev->edits = array();
1101                 foreach ($this->edits as $edit) {
1102                         $rev->edits[] = $edit->reverse();
1103                 }
1104                 return $rev;
1105         }
1106
1107         /**
1108          * Check for empty diff.
1109          *
1110          * @return bool True iff two sequences were identical.
1111          */
1112         function isEmpty () {
1113                 foreach ($this->edits as $edit) {
1114                         if ($edit->type != 'copy')
1115                                 return false;
1116                 }
1117                 return true;
1118         }
1119
1120         /**
1121          * Compute the length of the Longest Common Subsequence (LCS).
1122          *
1123          * This is mostly for diagnostic purposed.
1124          *
1125          * @return int The length of the LCS.
1126          */
1127         function lcs () {
1128                 $lcs = 0;
1129                 foreach ($this->edits as $edit) {
1130                         if ($edit->type == 'copy')
1131                                 $lcs += sizeof($edit->orig);
1132                 }
1133                 return $lcs;
1134         }
1135
1136         /**
1137          * Get the original set of lines.
1138          *
1139          * This reconstructs the $from_lines parameter passed to the
1140          * constructor.
1141          *
1142          * @return array The original sequence of strings.
1143          */
1144         function orig() {
1145                 $lines = array();
1146
1147                 foreach ($this->edits as $edit) {
1148                         if ($edit->orig)
1149                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
1150                 }
1151                 return $lines;
1152         }
1153
1154         /**
1155          * Get the closing set of lines.
1156          *
1157          * This reconstructs the $to_lines parameter passed to the
1158          * constructor.
1159          *
1160          * @return array The sequence of strings.
1161          */
1162         function closing() {
1163                 $lines = array();
1164
1165                 foreach ($this->edits as $edit) {
1166                         if ($edit->closing)
1167                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
1168                 }
1169                 return $lines;
1170         }
1171
1172         /**
1173          * Check a Diff for validity.
1174          *
1175          * This is here only for debugging purposes.
1176          */
1177         function _check ($from_lines, $to_lines) {
1178                 $fname = 'Diff::_check';
1179                 wfProfileIn( $fname );
1180                 if (serialize($from_lines) != serialize($this->orig()))
1181                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1182                 if (serialize($to_lines) != serialize($this->closing()))
1183                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1184
1185                 $rev = $this->reverse();
1186                 if (serialize($to_lines) != serialize($rev->orig()))
1187                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
1188                 if (serialize($from_lines) != serialize($rev->closing()))
1189                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1190
1191
1192                 $prevtype = 'none';
1193                 foreach ($this->edits as $edit) {
1194                         if ( $prevtype == $edit->type )
1195                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1196                         $prevtype = $edit->type;
1197                 }
1198
1199                 $lcs = $this->lcs();
1200                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1201                 wfProfileOut( $fname );
1202         }
1203 }
1204
1205 /**
1206  * FIXME: bad name.
1207  * @todo document
1208  * @access private
1209  * @package MediaWiki
1210  * @subpackage DifferenceEngine
1211  */
1212 class MappedDiff extends Diff
1213 {
1214         /**
1215          * Constructor.
1216          *
1217          * Computes diff between sequences of strings.
1218          *
1219          * This can be used to compute things like
1220          * case-insensitve diffs, or diffs which ignore
1221          * changes in white-space.
1222          *
1223          * @param $from_lines array An array of strings.
1224          *      (Typically these are lines from a file.)
1225          *
1226          * @param $to_lines array An array of strings.
1227          *
1228          * @param $mapped_from_lines array This array should
1229          *      have the same size number of elements as $from_lines.
1230          *      The elements in $mapped_from_lines and
1231          *      $mapped_to_lines are what is actually compared
1232          *      when computing the diff.
1233          *
1234          * @param $mapped_to_lines array This array should
1235          *      have the same number of elements as $to_lines.
1236          */
1237         function MappedDiff($from_lines, $to_lines,
1238                                                 $mapped_from_lines, $mapped_to_lines) {
1239                 $fname = 'MappedDiff::MappedDiff';
1240                 wfProfileIn( $fname );
1241
1242                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1243                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1244
1245                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1246
1247                 $xi = $yi = 0;
1248                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1249                         $orig = &$this->edits[$i]->orig;
1250                         if (is_array($orig)) {
1251                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1252                                 $xi += sizeof($orig);
1253                         }
1254
1255                         $closing = &$this->edits[$i]->closing;
1256                         if (is_array($closing)) {
1257                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1258                                 $yi += sizeof($closing);
1259                         }
1260                 }
1261                 wfProfileOut( $fname );
1262         }
1263 }
1264
1265 /**
1266  * A class to format Diffs
1267  *
1268  * This class formats the diff in classic diff format.
1269  * It is intended that this class be customized via inheritance,
1270  * to obtain fancier outputs.
1271  * @todo document
1272  * @access private
1273  * @package MediaWiki
1274  * @subpackage DifferenceEngine
1275  */
1276 class DiffFormatter
1277 {
1278         /**
1279          * Number of leading context "lines" to preserve.
1280          *
1281          * This should be left at zero for this class, but subclasses
1282          * may want to set this to other values.
1283          */
1284         var $leading_context_lines = 0;
1285
1286         /**
1287          * Number of trailing context "lines" to preserve.
1288          *
1289          * This should be left at zero for this class, but subclasses
1290          * may want to set this to other values.
1291          */
1292         var $trailing_context_lines = 0;
1293
1294         /**
1295          * Format a diff.
1296          *
1297          * @param $diff object A Diff object.
1298          * @return string The formatted output.
1299          */
1300         function format($diff) {
1301                 $fname = 'DiffFormatter::format';
1302                 wfProfileIn( $fname );
1303
1304                 $xi = $yi = 1;
1305                 $block = false;
1306                 $context = array();
1307
1308                 $nlead = $this->leading_context_lines;
1309                 $ntrail = $this->trailing_context_lines;
1310
1311                 $this->_start_diff();
1312
1313                 foreach ($diff->edits as $edit) {
1314                         if ($edit->type == 'copy') {
1315                                 if (is_array($block)) {
1316                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1317                                                 $block[] = $edit;
1318                                         }
1319                                         else{
1320                                                 if ($ntrail) {
1321                                                         $context = array_slice($edit->orig, 0, $ntrail);
1322                                                         $block[] = new _DiffOp_Copy($context);
1323                                                 }
1324                                                 $this->_block($x0, $ntrail + $xi - $x0,
1325                                                                           $y0, $ntrail + $yi - $y0,
1326                                                                           $block);
1327                                                 $block = false;
1328                                         }
1329                                 }
1330                                 $context = $edit->orig;
1331                         }
1332                         else {
1333                                 if (! is_array($block)) {
1334                                         $context = array_slice($context, sizeof($context) - $nlead);
1335                                         $x0 = $xi - sizeof($context);
1336                                         $y0 = $yi - sizeof($context);
1337                                         $block = array();
1338                                         if ($context)
1339                                                 $block[] = new _DiffOp_Copy($context);
1340                                 }
1341                                 $block[] = $edit;
1342                         }
1343
1344                         if ($edit->orig)
1345                                 $xi += sizeof($edit->orig);
1346                         if ($edit->closing)
1347                                 $yi += sizeof($edit->closing);
1348                 }
1349
1350                 if (is_array($block))
1351                         $this->_block($x0, $xi - $x0,
1352                                                   $y0, $yi - $y0,
1353                                                   $block);
1354
1355                 $end = $this->_end_diff();
1356                 wfProfileOut( $fname );
1357                 return $end;
1358         }
1359
1360         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1361                 $fname = 'DiffFormatter::_block';
1362                 wfProfileIn( $fname );
1363                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1364                 foreach ($edits as $edit) {
1365                         if ($edit->type == 'copy')
1366                                 $this->_context($edit->orig);
1367                         elseif ($edit->type == 'add')
1368                                 $this->_added($edit->closing);
1369                         elseif ($edit->type == 'delete')
1370                                 $this->_deleted($edit->orig);
1371                         elseif ($edit->type == 'change')
1372                                 $this->_changed($edit->orig, $edit->closing);
1373                         else
1374                                 trigger_error('Unknown edit type', E_USER_ERROR);
1375                 }
1376                 $this->_end_block();
1377                 wfProfileOut( $fname );
1378         }
1379
1380         function _start_diff() {
1381                 ob_start();
1382         }
1383
1384         function _end_diff() {
1385                 $val = ob_get_contents();
1386                 ob_end_clean();
1387                 return $val;
1388         }
1389
1390         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1391                 if ($xlen > 1)
1392                         $xbeg .= "," . ($xbeg + $xlen - 1);
1393                 if ($ylen > 1)
1394                         $ybeg .= "," . ($ybeg + $ylen - 1);
1395
1396                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1397         }
1398
1399         function _start_block($header) {
1400                 echo $header;
1401         }
1402
1403         function _end_block() {
1404         }
1405
1406         function _lines($lines, $prefix = ' ') {
1407                 foreach ($lines as $line)
1408                         echo "$prefix $line\n";
1409         }
1410
1411         function _context($lines) {
1412                 $this->_lines($lines);
1413         }
1414
1415         function _added($lines) {
1416                 $this->_lines($lines, '>');
1417         }
1418         function _deleted($lines) {
1419                 $this->_lines($lines, '<');
1420         }
1421
1422         function _changed($orig, $closing) {
1423                 $this->_deleted($orig);
1424                 echo "---\n";
1425                 $this->_added($closing);
1426         }
1427 }
1428
1429
1430 /**
1431  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1432  *
1433  */
1434
1435 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1436
1437 /**
1438  * @todo document
1439  * @access private
1440  * @package MediaWiki
1441  * @subpackage DifferenceEngine
1442  */
1443 class _HWLDF_WordAccumulator {
1444         function _HWLDF_WordAccumulator () {
1445                 $this->_lines = array();
1446                 $this->_line = '';
1447                 $this->_group = '';
1448                 $this->_tag = '';
1449         }
1450
1451         function _flushGroup ($new_tag) {
1452                 if ($this->_group !== '') {
1453                         if ($this->_tag == 'mark')
1454                                 $this->_line .= '<span class="diffchange">' .
1455                                         htmlspecialchars ( $this->_group ) . '</span>';
1456                         else
1457                                 $this->_line .= htmlspecialchars ( $this->_group );
1458                 }
1459                 $this->_group = '';
1460                 $this->_tag = $new_tag;
1461         }
1462
1463         function _flushLine ($new_tag) {
1464                 $this->_flushGroup($new_tag);
1465                 if ($this->_line != '')
1466                         array_push ( $this->_lines, $this->_line );
1467                 else
1468                         # make empty lines visible by inserting an NBSP
1469                         array_push ( $this->_lines, NBSP );
1470                 $this->_line = '';
1471         }
1472
1473         function addWords ($words, $tag = '') {
1474                 if ($tag != $this->_tag)
1475                         $this->_flushGroup($tag);
1476
1477                 foreach ($words as $word) {
1478                         // new-line should only come as first char of word.
1479                         if ($word == '')
1480                                 continue;
1481                         if ($word[0] == "\n") {
1482                                 $this->_flushLine($tag);
1483                                 $word = substr($word, 1);
1484                         }
1485                         assert(!strstr($word, "\n"));
1486                         $this->_group .= $word;
1487                 }
1488         }
1489
1490         function getLines() {
1491                 $this->_flushLine('~done');
1492                 return $this->_lines;
1493         }
1494 }
1495
1496 /**
1497  * @todo document
1498  * @access private
1499  * @package MediaWiki
1500  * @subpackage DifferenceEngine
1501  */
1502 class WordLevelDiff extends MappedDiff
1503 {
1504         function WordLevelDiff ($orig_lines, $closing_lines) {
1505                 $fname = 'WordLevelDiff::WordLevelDiff';
1506                 wfProfileIn( $fname );
1507
1508                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1509                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1510
1511                 $this->MappedDiff($orig_words, $closing_words,
1512                                                   $orig_stripped, $closing_stripped);
1513                 wfProfileOut( $fname );
1514         }
1515
1516         function _split($lines) {
1517                 $fname = 'WordLevelDiff::_split';
1518                 wfProfileIn( $fname );
1519
1520                 $words = array();
1521                 $stripped = array();
1522                 $first = true;
1523                 foreach ( $lines as $line ) {
1524                         # If the line is too long, just pretend the entire line is one big word
1525                         # This prevents resource exhaustion problems
1526                         if ( $first ) {
1527                                 $first = false;
1528                         } else {
1529                                 $words[] = "\n";
1530                                 $stripped[] = "\n";
1531                         }
1532                         if ( strlen( $line ) > MAX_DIFF_LINE ) {
1533                                 $words[] = $line;
1534                                 $stripped[] = $line;
1535                         } else {
1536                                 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1537                                         $line, $m))
1538                                 {
1539                                         $words = array_merge( $words, $m[0] );
1540                                         $stripped = array_merge( $stripped, $m[1] );
1541                                 }
1542                         }
1543                 }
1544                 wfProfileOut( $fname );
1545                 return array($words, $stripped);
1546         }
1547
1548         function orig () {
1549                 $fname = 'WordLevelDiff::orig';
1550                 wfProfileIn( $fname );
1551                 $orig = new _HWLDF_WordAccumulator;
1552
1553                 foreach ($this->edits as $edit) {
1554                         if ($edit->type == 'copy')
1555                                 $orig->addWords($edit->orig);
1556                         elseif ($edit->orig)
1557                                 $orig->addWords($edit->orig, 'mark');
1558                 }
1559                 $lines = $orig->getLines();
1560                 wfProfileOut( $fname );
1561                 return $lines;
1562         }
1563
1564         function closing () {
1565                 $fname = 'WordLevelDiff::closing';
1566                 wfProfileIn( $fname );
1567                 $closing = new _HWLDF_WordAccumulator;
1568
1569                 foreach ($this->edits as $edit) {
1570                         if ($edit->type == 'copy')
1571                                 $closing->addWords($edit->closing);
1572                         elseif ($edit->closing)
1573                                 $closing->addWords($edit->closing, 'mark');
1574                 }
1575                 $lines = $closing->getLines();
1576                 wfProfileOut( $fname );
1577                 return $lines;
1578         }
1579 }
1580
1581 /**
1582  *      Wikipedia Table style diff formatter.
1583  * @todo document
1584  * @access private
1585  * @package MediaWiki
1586  * @subpackage DifferenceEngine
1587  */
1588 class TableDiffFormatter extends DiffFormatter
1589 {
1590         function TableDiffFormatter() {
1591                 $this->leading_context_lines = 2;
1592                 $this->trailing_context_lines = 2;
1593         }
1594
1595         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1596                 $l1 = wfMsg( 'lineno', $xbeg );
1597                 $l2 = wfMsg( 'lineno', $ybeg );
1598
1599                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1600                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1601                 return $r;
1602         }
1603
1604         function _start_block( $header ) {
1605                 global $wgOut;
1606                 echo $header;
1607         }
1608
1609         function _end_block() {
1610         }
1611
1612         function _lines( $lines, $prefix=' ', $color='white' ) {
1613         }
1614
1615         # HTML-escape parameter before calling this
1616         function addedLine( $line ) {
1617                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1618         }
1619
1620         # HTML-escape parameter before calling this
1621         function deletedLine( $line ) {
1622                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1623         }
1624
1625         # HTML-escape parameter before calling this
1626         function contextLine( $line ) {
1627                 return "<td> </td><td class='diff-context'>{$line}</td>";
1628         }
1629
1630         function emptyLine() {
1631                 return '<td colspan="2">&nbsp;</td>';
1632         }
1633
1634         function _added( $lines ) {
1635                 foreach ($lines as $line) {
1636                         echo '<tr>' . $this->emptyLine() .
1637                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1638                 }
1639         }
1640
1641         function _deleted($lines) {
1642                 foreach ($lines as $line) {
1643                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1644                           $this->emptyLine() . "</tr>\n";
1645                 }
1646         }
1647
1648         function _context( $lines ) {
1649                 foreach ($lines as $line) {
1650                         echo '<tr>' .
1651                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1652                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1653                 }
1654         }
1655
1656         function _changed( $orig, $closing ) {
1657                 $fname = 'TableDiffFormatter::_changed';
1658                 wfProfileIn( $fname );
1659
1660                 $diff = new WordLevelDiff( $orig, $closing );
1661                 $del = $diff->orig();
1662                 $add = $diff->closing();
1663
1664                 # Notice that WordLevelDiff returns HTML-escaped output.
1665                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1666
1667                 while ( $line = array_shift( $del ) ) {
1668                         $aline = array_shift( $add );
1669                         echo '<tr>' . $this->deletedLine( $line ) .
1670                                 $this->addedLine( $aline ) . "</tr>\n";
1671                 }
1672                 foreach ($add as $line) {       # If any leftovers
1673                         echo '<tr>' . $this->emptyLine() .
1674                                 $this->addedLine( $line ) . "</tr>\n";
1675                 }
1676                 wfProfileOut( $fname );
1677         }
1678 }
1679
1680 ?>