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