includes/DifferenceEngine.php

   1 <?php
   2 /**
   3  * See diff.doc
   4  * @package MediaWiki
   5  * @subpackage DifferenceEngine
   6  */
   7
   8 /**
   9  * @todo document
  10  * @access public
  11  */
  12 class DifferenceEngine {
  13         /* private */ var $mOldid, $mNewid;
  14         /* private */ var $mOldtitle, $mNewtitle, $mPagetitle;
  15         /* private */ var $mOldtext, $mNewtext;
  16         /* private */ var $mOldUser, $mNewUser;
  17         /* private */ var $mOldComment, $mNewComment;
  18         /* private */ var $mOldPage, $mNewPage;
  19         /* private */ var $mRcidMarkPatrolled;
  20
  21         function DifferenceEngine( $old, $new, $rcid = 0 )
  22         {
  23                 global $wgTitle;
  24                 if ( 'prev' == $new ) {
  25                         # Show diff between revision $old and the previous one.
  26                         # Get previous one from DB.
  27                         #
  28                         $this->mNewid = intval($old);
  29
  30                         $this->mOldid = $wgTitle->getPreviousRevisionID( $this->mNewid );
  31
  32                 } elseif ( 'next' == $new ) {
  33
  34                         # Show diff between revision $old and the previous one.
  35                         # Get previous one from DB.
  36                         #
  37                         $this->mOldid = intval($old);
  38                         $this->mNewid = $wgTitle->getNextRevisionID( $this->mOldid );
  39                         if ( false === $this->mNewid ) {
  40                                 # if no result, NewId points to the newest old revision. The only newer
  41                                 # revision is cur, which is "0".
  42                                 $this->mNewid = 0;
  43                         }
  44
  45                 } else {
  46
  47                         $this->mOldid = intval($old);
  48                         $this->mNewid = intval($new);
  49                 }
  50                 $this->mRcidMarkPatrolled = intval($rcid);  # force it to be an integer
  51         }
  52
  53         function showDiffPage()
  54         {
  55                 global $wgUser, $wgTitle, $wgOut, $wgContLang, $wgOnlySysopsCanPatrol, $wgUseRCPatrol;
  56                 $fname = 'DifferenceEngine::showDiffPage';
  57                 wfProfileIn( $fname );
  58
  59                 # mOldid is false if the difference engine is called with a "vague" query for
  60                 # a diff between a version V and its previous version V' AND the version V
  61                 # is the first version of that article. In that case, V' does not exist.
  62                 if ( $this->mOldid === false ) {
  63                         $this->showFirstRevision();
  64                         wfProfileOut( $fname );
  65                         return;
  66                 }
  67
  68                 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
  69                   "{$this->mNewid})";
  70                 $mtext = wfMsg( 'missingarticle', $t );
  71
  72                 $wgOut->setArticleFlag( false );
  73                 if ( ! $this->loadText() ) {
  74                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
  75                         $wgOut->addHTML( $mtext );
  76                         wfProfileOut( $fname );
  77                         return;
  78                 }
  79                 $wgOut->suppressQuickbar();
  80
  81                 $oldTitle = $this->mOldPage->getPrefixedText();
  82                 $newTitle = $this->mNewPage->getPrefixedText();
  83                 if( $oldTitle == $newTitle ) {
  84                         $wgOut->setPageTitle( $newTitle );
  85                 } else {
  86                         $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
  87                 }
  88                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
  89                 $wgOut->setRobotpolicy( 'noindex,follow' );
  90
  91                 if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
  92                         $wgOut->loginToUse();
  93                         $wgOut->output();
  94                         wfProfileOut( $fname );
  95                         exit;
  96                 }
  97
  98                 $sk = $wgUser->getSkin();
  99                 $talk = $wgContLang->getNsText( NS_TALK );
 100                 $contribs = wfMsg( 'contribslink' );
 101
 102                 $this->mOldComment = $sk->formatComment($this->mOldComment);
 103                 $this->mNewComment = $sk->formatComment($this->mNewComment);
 104
 105                 $oldUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 106                 $newUserLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mNewUser ), $this->mNewUser );
 107                 $oldUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ), $talk );
 108                 $newUTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mNewUser ), $talk );
 109                 $oldContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 110                         'target=' . urlencode($this->mOldUser) );
 111                 $newContribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), $contribs,
 112                         'target=' . urlencode($this->mNewUser) );
 113                 if ( !$this->mNewid && $wgUser->isAllowed('rollback') ) {
 114                         $rollback = '&nbsp;&nbsp;&nbsp;<strong>[' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'rollbacklink' ),
 115                                 'action=rollback&from=' . urlencode($this->mNewUser) ) . ']</strong>';
 116                 } else {
 117                         $rollback = '';
 118                 }
 119                 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->getID() != 0 &&
 120                      ( $wgUser->isAllowed('rollback') || !$wgOnlySysopsCanPatrol ) )
 121                 {
 122                         $patrol = ' [' . $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'markaspatrolleddiff' ),
 123                                 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
 124                 } else {
 125                         $patrol = '';
 126                 }
 127
 128                 $prevlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'previousdiff' ), 'diff=prev&oldid='.$this->mOldid );
 129                 if ( $this->mNewid == 0 ) {
 130                         $nextlink = '';
 131                 } else {
 132                         $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 133                 }
 134
 135                 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
 136                         "($oldUTLink | $oldContribs)<br />" . $this->mOldComment .
 137                         '<br />' . $prevlink;
 138                 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
 139                         "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment .
 140                         '<br />' . $nextlink . $patrol;
 141
 142                 DifferenceEngine::showDiff( $this->mOldtext, $this->mNewtext,
 143                   $oldHeader, $newHeader );
 144                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 145                 $wgOut->addWikiText( $this->mNewtext );
 146
 147                 wfProfileOut( $fname );
 148         }
 149
 150         # Show the first revision of an article. Uses normal diff headers in contrast to normal
 151         # "old revision" display style.
 152         #
 153         function showFirstRevision()
 154         {
 155                 global $wgOut, $wgTitle, $wgUser, $wgLang;
 156
 157                 $fname = 'DifferenceEngine::showFirstRevision';
 158                 wfProfileIn( $fname );
 159
 160
 161                 $this->mOldid = $this->mNewid; # hack to make loadText() work.
 162
 163                 # Get article text from the DB
 164                 #
 165                 if ( ! $this->loadText() ) {
 166                         $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
 167                           "{$this->mNewid})";
 168                         $mtext = wfMsg( 'missingarticle', $t );
 169                         $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
 170                         $wgOut->addHTML( $mtext );
 171                         wfProfileOut( $fname );
 172                         return;
 173                 }
 174
 175                 # Check if user is allowed to look at this page. If not, bail out.
 176                 #
 177                 if ( !( $this->mOldPage->userCanRead() ) ) {
 178                         $wgOut->loginToUse();
 179                         $wgOut->output();
 180                         wfProfileOut( $fname );
 181                         exit;
 182                 }
 183
 184                 # Prepare the header box
 185                 #
 186                 $sk = $wgUser->getSkin();
 187
 188                 $uTLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER_TALK, $this->mOldUser ),  $wgLang->getNsText( NS_TALK ) );
 189                 $userLink = $sk->makeLinkObj( Title::makeTitleSafe( NS_USER, $this->mOldUser ), $this->mOldUser );
 190                 $contribs = $sk->makeKnownLinkObj( Title::makeTitle( NS_SPECIAL, 'Contributions' ), wfMsg( 'contribslink' ),
 191                         'target=' . urlencode($this->mOldUser) );
 192                 $nextlink = $sk->makeKnownLinkObj( $wgTitle, wfMsg( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid );
 193                 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />$userLink " .
 194                         "($uTLink | $contribs)<br />" . $this->mOldComment .
 195                         '<br />' . $nextlink. "</div>\n";
 196
 197                 $wgOut->addHTML( $header );
 198
 199                 $wgOut->setSubtitle( wfMsg( 'difference' ) );
 200                 $wgOut->setRobotpolicy( 'noindex,follow' );
 201
 202
 203                 # Show current revision
 204                 #
 205                 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
 206                 $wgOut->addWikiText( $this->mNewtext );
 207
 208                 wfProfileOut( $fname );
 209         }
 210
 211         function showDiff( $otext, $ntext, $otitle, $ntitle )
 212         {
 213                 global $wgOut;
 214                 $wgOut->addHTML( DifferenceEngine::getDiff( $otext, $ntext, $otitle, $ntitle ) );
 215         }
 216
 217         function getDiff( $otext, $ntext, $otitle, $ntitle ) {
 218                 global $wgUseExternalDiffEngine, $wgContLang;
 219                 $out = "
 220                         <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
 221                         <tr>
 222                                 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
 223                                 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
 224                         </tr>
 225                 ";
 226                 $otext = $wgContLang->segmentForDiff($otext);
 227                 $ntext = $wgContLang->segmentForDiff($ntext);
 228                 $difftext='';
 229                 if ( $wgUseExternalDiffEngine ) {
 230                         # For historical reasons, external diff engine expects
 231                         # input text to be HTML-escaped already
 232                         $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
 233                         $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
 234                         if( !function_exists( 'wikidiff_do_diff' ) ) {
 235                                 dl('php_wikidiff.so');
 236                         }
 237                         $difftext = wikidiff_do_diff( $otext, $ntext, 2 );
 238                 } else {
 239                         $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
 240                         $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
 241                         $diffs =& new Diff( $ota, $nta );
 242                         $formatter =& new TableDiffFormatter();
 243                         $difftext = $formatter->format( $diffs );
 244                 }
 245                 $difftext = $wgContLang->unsegmentForDiff($difftext);
 246                 $out .= $difftext."</table>\n";
 247                 return $out;
 248         }
 249
 250         # Load the text of the articles to compare.  If newid is 0, then compare
 251         # the old article in oldid to the current article; if oldid is 0, then
 252         # compare the current article to the immediately previous one (ignoring
 253         # the value of newid).
 254         #
 255         function loadText()
 256         {
 257                 global $wgTitle, $wgOut, $wgLang;
 258                 $fname = 'DifferenceEngine::loadText';
 259
 260                 $dbr =& wfGetDB( DB_SLAVE );
 261                 if ( 0 == $this->mNewid || 0 == $this->mOldid ) {
 262                         /* Fetch current revision */
 263                         $wgOut->setArticleFlag( true );
 264                         $newLink = $wgTitle->escapeLocalUrl();
 265                         $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
 266                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 267                         $id = $wgTitle->getArticleID();
 268
 269                         $s = $dbr->selectRow( array( 'page', 'revision', 'text' ),
 270                                 array( 'old_flags', 'old_text', 'rev_user_text', 'rev_comment' ),
 271                                 "page_id=$id AND page_id=rev_page AND rev_id=old_id", $fname );
 272                         if ( $s === false ) {
 273                                 wfDebug( "Unable to load page_id $id\n" );
 274                                 return false;
 275                         }
 276
 277                         $this->mNewPage = &$wgTitle;
 278                         $this->mNewtext = Article::getRevisionText( $s );
 279                         $this->mNewUser = $s->rev_user_text;
 280                         $this->mNewComment = $s->rev_comment;
 281                 } else {
 282                         $s = $dbr->selectRow( array( 'page', 'revision', 'text' ),
 283                                 array( 'page_namespace','page_title','rev_timestamp', 'old_text',
 284                                 'old_flags','rev_user_text','rev_comment' ),
 285                                 "page_id=rev_page AND rev_id=old_id AND rev_id={$this->mNewid}",
 286                                 $fname );
 287
 288                         if ( $s === false ) {
 289                                 wfDebug( "Unable to load old_id {$this->mNewid}\n" );
 290                                 return false;
 291                         }
 292
 293                         $this->mNewtext = Article::getRevisionText( $s );
 294
 295                         $t = $wgLang->timeanddate( $s->rev_timestamp, true );
 296                         $this->mNewPage = Title::MakeTitle( $s->page_namespace, $s->page_title );
 297                         $newLink = $wgTitle->escapeLocalUrl ('oldid=' . $this->mNewid);
 298                         $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $t ) );
 299                         $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>";
 300                         $this->mNewUser = $s->rev_user_text;
 301                         $this->mNewComment = $s->rev_comment;
 302                 }
 303                 if ( 0 == $this->mOldid ) {
 304                         $s = $dbr->selectRow( 'old',
 305                                 array( 'old_namespace','old_title','old_timestamp','old_text', 'old_flags','old_user_text','old_comment' ),
 306                                 array( /* WHERE */
 307                                         'old_namespace' => $this->mNewPage->getNamespace(),
 308                                         'old_title' => $this->mNewPage->getDBkey()
 309                                 ), $fname, array( 'ORDER BY' => 'inverse_timestamp', 'USE INDEX' => 'name_title_timestamp' )
 310                         );
 311                         if ( $s === false ) {
 312                                 wfDebug( 'Unable to load ' . $this->mNewPage->getPrefixedDBkey() . " from old\n" );
 313                                 return false;
 314                         }
 315                 } else {
 316                         $s = $dbr->selectRow( array( 'page', 'revision', 'text' ),
 317                                 array( 'page_namespace','page_title','rev_timestamp','old_text','old_flags','rev_user_text','rev_comment'),
 318                                 "page_id=rev_page AND rev_id=old_id AND rev_id={$this->mOldid}",
 319                                 $fname
 320                         );
 321                         if ( $s === false ) {
 322                                 wfDebug( "Unable to load old_id {$this->mOldid}\n" );
 323                                 return false;
 324                         }
 325                 }
 326                 $this->mOldPage = Title::MakeTitle( $s->page_namespace, $s->page_title );
 327                 $this->mOldtext = Article::getRevisionText( $s );
 328
 329                 $t = $wgLang->timeanddate( $s->rev_timestamp, true );
 330                 $oldLink = $this->mOldPage->escapeLocalUrl ('oldid=' . $this->mOldid);
 331                 $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
 332                 $this->mOldUser = $s->rev_user_text;
 333                 $this->mOldComment = $s->rev_comment;
 334
 335                 return true;
 336         }
 337 }
 338
 339 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
 340 //
 341 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
 342 // You may copy this code freely under the conditions of the GPL.
 343 //
 344
 345 define('USE_ASSERTS', function_exists('assert'));
 346
 347 /**
 348  * @todo document
 349  * @access private
 350  */
 351 class _DiffOp {
 352         var $type;
 353         var $orig;
 354         var $closing;
 355
 356         function reverse() {
 357                 trigger_error('pure virtual', E_USER_ERROR);
 358         }
 359
 360         function norig() {
 361                 return $this->orig ? sizeof($this->orig) : 0;
 362         }
 363
 364         function nclosing() {
 365                 return $this->closing ? sizeof($this->closing) : 0;
 366         }
 367 }
 368
 369 /**
 370  * @todo document
 371  * @access private
 372  */
 373 class _DiffOp_Copy extends _DiffOp {
 374         var $type = 'copy';
 375
 376         function _DiffOp_Copy ($orig, $closing = false) {
 377                 if (!is_array($closing))
 378                         $closing = $orig;
 379                 $this->orig = $orig;
 380                 $this->closing = $closing;
 381         }
 382
 383         function reverse() {
 384                 return new _DiffOp_Copy($this->closing, $this->orig);
 385         }
 386 }
 387
 388 /**
 389  * @todo document
 390  * @access private
 391  */
 392 class _DiffOp_Delete extends _DiffOp {
 393         var $type = 'delete';
 394
 395         function _DiffOp_Delete ($lines) {
 396                 $this->orig = $lines;
 397                 $this->closing = false;
 398         }
 399
 400         function reverse() {
 401                 return new _DiffOp_Add($this->orig);
 402         }
 403 }
 404
 405 /**
 406  * @todo document
 407  * @access private
 408  */
 409 class _DiffOp_Add extends _DiffOp {
 410         var $type = 'add';
 411
 412         function _DiffOp_Add ($lines) {
 413                 $this->closing = $lines;
 414                 $this->orig = false;
 415         }
 416
 417         function reverse() {
 418                 return new _DiffOp_Delete($this->closing);
 419         }
 420 }
 421
 422 /**
 423  * @todo document
 424  * @access private
 425  */
 426 class _DiffOp_Change extends _DiffOp {
 427         var $type = 'change';
 428
 429         function _DiffOp_Change ($orig, $closing) {
 430                 $this->orig = $orig;
 431                 $this->closing = $closing;
 432         }
 433
 434         function reverse() {
 435                 return new _DiffOp_Change($this->closing, $this->orig);
 436         }
 437 }
 438
 439
 440 /**
 441  * Class used internally by Diff to actually compute the diffs.
 442  *
 443  * The algorithm used here is mostly lifted from the perl module
 444  * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
 445  *       http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
 446  *
 447  * More ideas are taken from:
 448  *       http://www.ics.uci.edu/~eppstein/161/960229.html
 449  *
 450  * Some ideas are (and a bit of code) are from from analyze.c, from GNU
 451  * diffutils-2.7, which can be found at:
 452  *       ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
 453  *
 454  * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
 455  * are my own.
 456  *
 457  * @author Geoffrey T. Dairiki
 458  * @access private
 459  */
 460 class _DiffEngine
 461 {
 462         function diff ($from_lines, $to_lines) {
 463                 $fname = '_DiffEngine::diff';
 464                 wfProfileIn( $fname );
 465
 466                 $n_from = sizeof($from_lines);
 467                 $n_to = sizeof($to_lines);
 468
 469                 $this->xchanged = $this->ychanged = array();
 470                 $this->xv = $this->yv = array();
 471                 $this->xind = $this->yind = array();
 472                 unset($this->seq);
 473                 unset($this->in_seq);
 474                 unset($this->lcs);
 475
 476                 // Skip leading common lines.
 477                 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
 478                         if ($from_lines[$skip] != $to_lines[$skip])
 479                                 break;
 480                         $this->xchanged[$skip] = $this->ychanged[$skip] = false;
 481                 }
 482                 // Skip trailing common lines.
 483                 $xi = $n_from; $yi = $n_to;
 484                 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
 485                         if ($from_lines[$xi] != $to_lines[$yi])
 486                                 break;
 487                         $this->xchanged[$xi] = $this->ychanged[$yi] = false;
 488                 }
 489
 490                 // Ignore lines which do not exist in both files.
 491                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++)
 492                         $xhash[$from_lines[$xi]] = 1;
 493                 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
 494                         $line = $to_lines[$yi];
 495                         if ( ($this->ychanged[$yi] = empty($xhash[$line])) )
 496                                 continue;
 497                         $yhash[$line] = 1;
 498                         $this->yv[] = $line;
 499                         $this->yind[] = $yi;
 500                 }
 501                 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
 502                         $line = $from_lines[$xi];
 503                         if ( ($this->xchanged[$xi] = empty($yhash[$line])) )
 504                                 continue;
 505                         $this->xv[] = $line;
 506                         $this->xind[] = $xi;
 507                 }
 508
 509                 // Find the LCS.
 510                 $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
 511
 512                 // Merge edits when possible
 513                 $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
 514                 $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
 515
 516                 // Compute the edit operations.
 517                 $edits = array();
 518                 $xi = $yi = 0;
 519                 while ($xi < $n_from || $yi < $n_to) {
 520                         USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
 521                         USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
 522
 523                         // Skip matching "snake".
 524                         $copy = array();
 525                         while ( $xi < $n_from && $yi < $n_to
 526                                         && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
 527                                 $copy[] = $from_lines[$xi++];
 528                                 ++$yi;
 529                         }
 530                         if ($copy)
 531                                 $edits[] = new _DiffOp_Copy($copy);
 532
 533                         // Find deletes & adds.
 534                         $delete = array();
 535                         while ($xi < $n_from && $this->xchanged[$xi])
 536                                 $delete[] = $from_lines[$xi++];
 537
 538                         $add = array();
 539                         while ($yi < $n_to && $this->ychanged[$yi])
 540                                 $add[] = $to_lines[$yi++];
 541
 542                         if ($delete && $add)
 543                                 $edits[] = new _DiffOp_Change($delete, $add);
 544                         elseif ($delete)
 545                                 $edits[] = new _DiffOp_Delete($delete);
 546                         elseif ($add)
 547                                 $edits[] = new _DiffOp_Add($add);
 548                 }
 549                 wfProfileOut( $fname );
 550                 return $edits;
 551         }
 552
 553
 554         /* Divide the Largest Common Subsequence (LCS) of the sequences
 555          * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
 556          * sized segments.
 557          *
 558          * Returns (LCS, PTS).  LCS is the length of the LCS. PTS is an
 559          * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
 560          * sub sequences.  The first sub-sequence is contained in [X0, X1),
 561          * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on.  Note
 562          * that (X0, Y0) == (XOFF, YOFF) and
 563          * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
 564          *
 565          * This function assumes that the first lines of the specified portions
 566          * of the two files do not match, and likewise that the last lines do not
 567          * match.  The caller must trim matching lines from the beginning and end
 568          * of the portions it is going to specify.
 569          */
 570         function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
 571                 $fname = '_DiffEngine::_diag';
 572                 wfProfileIn( $fname );
 573                 $flip = false;
 574
 575                 if ($xlim - $xoff > $ylim - $yoff) {
 576                         // Things seems faster (I'm not sure I understand why)
 577                                 // when the shortest sequence in X.
 578                                 $flip = true;
 579                         list ($xoff, $xlim, $yoff, $ylim)
 580                         = array( $yoff, $ylim, $xoff, $xlim);
 581                 }
 582
 583                 if ($flip)
 584                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 585                                 $ymatches[$this->xv[$i]][] = $i;
 586                 else
 587                         for ($i = $ylim - 1; $i >= $yoff; $i--)
 588                                 $ymatches[$this->yv[$i]][] = $i;
 589
 590                 $this->lcs = 0;
 591                 $this->seq[0]= $yoff - 1;
 592                 $this->in_seq = array();
 593                 $ymids[0] = array();
 594
 595                 $numer = $xlim - $xoff + $nchunks - 1;
 596                 $x = $xoff;
 597                 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
 598                         wfProfileIn( "$fname-chunk" );
 599                         if ($chunk > 0)
 600                                 for ($i = 0; $i <= $this->lcs; $i++)
 601                                         $ymids[$i][$chunk-1] = $this->seq[$i];
 602
 603                         $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
 604                         for ( ; $x < $x1; $x++) {
 605                                 $line = $flip ? $this->yv[$x] : $this->xv[$x];
 606                                         if (empty($ymatches[$line]))
 607                                                 continue;
 608                                 $matches = $ymatches[$line];
 609                                 reset($matches);
 610                                 while (list ($junk, $y) = each($matches))
 611                                         if (empty($this->in_seq[$y])) {
 612                                                 $k = $this->_lcs_pos($y);
 613                                                 USE_ASSERTS && assert($k > 0);
 614                                                 $ymids[$k] = $ymids[$k-1];
 615                                                 break;
 616                                         }
 617                                 while (list ($junk, $y) = each($matches)) {
 618                                         if ($y > $this->seq[$k-1]) {
 619                                                 USE_ASSERTS && assert($y < $this->seq[$k]);
 620                                                 // Optimization: this is a common case:
 621                                                 //      next match is just replacing previous match.
 622                                                 $this->in_seq[$this->seq[$k]] = false;
 623                                                 $this->seq[$k] = $y;
 624                                                 $this->in_seq[$y] = 1;
 625                                         } else if (empty($this->in_seq[$y])) {
 626                                                 $k = $this->_lcs_pos($y);
 627                                                 USE_ASSERTS && assert($k > 0);
 628                                                 $ymids[$k] = $ymids[$k-1];
 629                                         }
 630                                 }
 631                         }
 632                         wfProfileOut( "$fname-chunk" );
 633                 }
 634
 635                 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
 636                 $ymid = $ymids[$this->lcs];
 637                 for ($n = 0; $n < $nchunks - 1; $n++) {
 638                         $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
 639                         $y1 = $ymid[$n] + 1;
 640                         $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
 641                 }
 642                 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
 643
 644                 wfProfileOut( $fname );
 645                 return array($this->lcs, $seps);
 646         }
 647
 648         function _lcs_pos ($ypos) {
 649                 $fname = '_DiffEngine::_lcs_pos';
 650                 wfProfileIn( $fname );
 651
 652                 $end = $this->lcs;
 653                 if ($end == 0 || $ypos > $this->seq[$end]) {
 654                         $this->seq[++$this->lcs] = $ypos;
 655                         $this->in_seq[$ypos] = 1;
 656                         wfProfileOut( $fname );
 657                         return $this->lcs;
 658                 }
 659
 660                 $beg = 1;
 661                 while ($beg < $end) {
 662                         $mid = (int)(($beg + $end) / 2);
 663                         if ( $ypos > $this->seq[$mid] )
 664                                 $beg = $mid + 1;
 665                         else
 666                                 $end = $mid;
 667                 }
 668
 669                 USE_ASSERTS && assert($ypos != $this->seq[$end]);
 670
 671                 $this->in_seq[$this->seq[$end]] = false;
 672                 $this->seq[$end] = $ypos;
 673                 $this->in_seq[$ypos] = 1;
 674                 wfProfileOut( $fname );
 675                 return $end;
 676         }
 677
 678         /* Find LCS of two sequences.
 679          *
 680          * The results are recorded in the vectors $this->{x,y}changed[], by
 681          * storing a 1 in the element for each line that is an insertion
 682          * or deletion (ie. is not in the LCS).
 683          *
 684          * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
 685          *
 686          * Note that XLIM, YLIM are exclusive bounds.
 687          * All line numbers are origin-0 and discarded lines are not counted.
 688          */
 689         function _compareseq ($xoff, $xlim, $yoff, $ylim) {
 690                 $fname = '_DiffEngine::_compareseq';
 691                 wfProfileIn( $fname );
 692
 693                 // Slide down the bottom initial diagonal.
 694                 while ($xoff < $xlim && $yoff < $ylim
 695                            && $this->xv[$xoff] == $this->yv[$yoff]) {
 696                         ++$xoff;
 697                         ++$yoff;
 698                 }
 699
 700                 // Slide up the top initial diagonal.
 701                 while ($xlim > $xoff && $ylim > $yoff
 702                            && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
 703                         --$xlim;
 704                         --$ylim;
 705                 }
 706
 707                 if ($xoff == $xlim || $yoff == $ylim)
 708                         $lcs = 0;
 709                 else {
 710                         // This is ad hoc but seems to work well.
 711                         //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
 712                         //$nchunks = max(2,min(8,(int)$nchunks));
 713                         $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
 714                         list ($lcs, $seps)
 715                         = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
 716                 }
 717
 718                 if ($lcs == 0) {
 719                         // X and Y sequences have no common subsequence:
 720                         // mark all changed.
 721                         while ($yoff < $ylim)
 722                                 $this->ychanged[$this->yind[$yoff++]] = 1;
 723                         while ($xoff < $xlim)
 724                                 $this->xchanged[$this->xind[$xoff++]] = 1;
 725                 } else {
 726                         // Use the partitions to split this problem into subproblems.
 727                         reset($seps);
 728                         $pt1 = $seps[0];
 729                         while ($pt2 = next($seps)) {
 730                                 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
 731                                 $pt1 = $pt2;
 732                         }
 733                 }
 734                 wfProfileOut( $fname );
 735         }
 736
 737         /* Adjust inserts/deletes of identical lines to join changes
 738          * as much as possible.
 739          *
 740          * We do something when a run of changed lines include a
 741          * line at one end and has an excluded, identical line at the other.
 742          * We are free to choose which identical line is included.
 743          * `compareseq' usually chooses the one at the beginning,
 744          * but usually it is cleaner to consider the following identical line
 745          * to be the "change".
 746          *
 747          * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
 748          */
 749         function _shift_boundaries ($lines, &$changed, $other_changed) {
 750                 $fname = '_DiffEngine::_shift_boundaries';
 751                 wfProfileIn( $fname );
 752                 $i = 0;
 753                 $j = 0;
 754
 755                 USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
 756                 $len = sizeof($lines);
 757                 $other_len = sizeof($other_changed);
 758
 759                 while (1) {
 760                         /*
 761                          * Scan forwards to find beginning of another run of changes.
 762                          * Also keep track of the corresponding point in the other file.
 763                          *
 764                          * Throughout this code, $i and $j are adjusted together so that
 765                          * the first $i elements of $changed and the first $j elements
 766                          * of $other_changed both contain the same number of zeros
 767                          * (unchanged lines).
 768                          * Furthermore, $j is always kept so that $j == $other_len or
 769                          * $other_changed[$j] == false.
 770                          */
 771                         while ($j < $other_len && $other_changed[$j])
 772                                 $j++;
 773
 774                         while ($i < $len && ! $changed[$i]) {
 775                                 USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 776                                 $i++; $j++;
 777                                 while ($j < $other_len && $other_changed[$j])
 778                                         $j++;
 779                         }
 780
 781                         if ($i == $len)
 782                                 break;
 783
 784                         $start = $i;
 785
 786                         // Find the end of this run of changes.
 787                         while (++$i < $len && $changed[$i])
 788                                 continue;
 789
 790                         do {
 791                                 /*
 792                                  * Record the length of this run of changes, so that
 793                                  * we can later determine whether the run has grown.
 794                                  */
 795                                 $runlength = $i - $start;
 796
 797                                 /*
 798                                  * Move the changed region back, so long as the
 799                                  * previous unchanged line matches the last changed one.
 800                                  * This merges with previous changed regions.
 801                                  */
 802                                 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
 803                                         $changed[--$start] = 1;
 804                                         $changed[--$i] = false;
 805                                         while ($start > 0 && $changed[$start - 1])
 806                                                 $start--;
 807                                         USE_ASSERTS && assert('$j > 0');
 808                                         while ($other_changed[--$j])
 809                                                 continue;
 810                                         USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 811                                 }
 812
 813                                 /*
 814                                  * Set CORRESPONDING to the end of the changed run, at the last
 815                                  * point where it corresponds to a changed run in the other file.
 816                                  * CORRESPONDING == LEN means no such point has been found.
 817                                  */
 818                                 $corresponding = $j < $other_len ? $i : $len;
 819
 820                                 /*
 821                                  * Move the changed region forward, so long as the
 822                                  * first changed line matches the following unchanged one.
 823                                  * This merges with following changed regions.
 824                                  * Do this second, so that if there are no merges,
 825                                  * the changed region is moved forward as far as possible.
 826                                  */
 827                                 while ($i < $len && $lines[$start] == $lines[$i]) {
 828                                         $changed[$start++] = false;
 829                                         $changed[$i++] = 1;
 830                                         while ($i < $len && $changed[$i])
 831                                                 $i++;
 832
 833                                         USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
 834                                         $j++;
 835                                         if ($j < $other_len && $other_changed[$j]) {
 836                                                 $corresponding = $i;
 837                                                 while ($j < $other_len && $other_changed[$j])
 838                                                         $j++;
 839                                         }
 840                                 }
 841                         } while ($runlength != $i - $start);
 842
 843                         /*
 844                          * If possible, move the fully-merged run of changes
 845                          * back to a corresponding run in the other file.
 846                          */
 847                         while ($corresponding < $i) {
 848                                 $changed[--$start] = 1;
 849                                 $changed[--$i] = 0;
 850                                 USE_ASSERTS && assert('$j > 0');
 851                                 while ($other_changed[--$j])
 852                                         continue;
 853                                 USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
 854                         }
 855                 }
 856                 wfProfileOut( $fname );
 857         }
 858 }
 859
 860 /**
 861  * Class representing a 'diff' between two sequences of strings.
 862  * @todo document
 863  * @access private
 864  */
 865 class Diff
 866 {
 867         var $edits;
 868
 869         /**
 870          * Constructor.
 871          * Computes diff between sequences of strings.
 872          *
 873          * @param $from_lines array An array of strings.
 874          *                (Typically these are lines from a file.)
 875          * @param $to_lines array An array of strings.
 876          */
 877         function Diff($from_lines, $to_lines) {
 878                 $eng = new _DiffEngine;
 879                 $this->edits = $eng->diff($from_lines, $to_lines);
 880                 //$this->_check($from_lines, $to_lines);
 881         }
 882
 883         /**
 884          * Compute reversed Diff.
 885          *
 886          * SYNOPSIS:
 887          *
 888          *      $diff = new Diff($lines1, $lines2);
 889          *      $rev = $diff->reverse();
 890          * @return object A Diff object representing the inverse of the
 891          *                                original diff.
 892          */
 893         function reverse () {
 894                 $rev = $this;
 895                 $rev->edits = array();
 896                 foreach ($this->edits as $edit) {
 897                         $rev->edits[] = $edit->reverse();
 898                 }
 899                 return $rev;
 900         }
 901
 902         /**
 903          * Check for empty diff.
 904          *
 905          * @return bool True iff two sequences were identical.
 906          */
 907         function isEmpty () {
 908                 foreach ($this->edits as $edit) {
 909                         if ($edit->type != 'copy')
 910                                 return false;
 911                 }
 912                 return true;
 913         }
 914
 915         /**
 916          * Compute the length of the Longest Common Subsequence (LCS).
 917          *
 918          * This is mostly for diagnostic purposed.
 919          *
 920          * @return int The length of the LCS.
 921          */
 922         function lcs () {
 923                 $lcs = 0;
 924                 foreach ($this->edits as $edit) {
 925                         if ($edit->type == 'copy')
 926                                 $lcs += sizeof($edit->orig);
 927                 }
 928                 return $lcs;
 929         }
 930
 931         /**
 932          * Get the original set of lines.
 933          *
 934          * This reconstructs the $from_lines parameter passed to the
 935          * constructor.
 936          *
 937          * @return array The original sequence of strings.
 938          */
 939         function orig() {
 940                 $lines = array();
 941
 942                 foreach ($this->edits as $edit) {
 943                         if ($edit->orig)
 944                                 array_splice($lines, sizeof($lines), 0, $edit->orig);
 945                 }
 946                 return $lines;
 947         }
 948
 949         /**
 950          * Get the closing set of lines.
 951          *
 952          * This reconstructs the $to_lines parameter passed to the
 953          * constructor.
 954          *
 955          * @return array The sequence of strings.
 956          */
 957         function closing() {
 958                 $lines = array();
 959
 960                 foreach ($this->edits as $edit) {
 961                         if ($edit->closing)
 962                                 array_splice($lines, sizeof($lines), 0, $edit->closing);
 963                 }
 964                 return $lines;
 965         }
 966
 967         /**
 968          * Check a Diff for validity.
 969          *
 970          * This is here only for debugging purposes.
 971          */
 972         function _check ($from_lines, $to_lines) {
 973                 $fname = 'Diff::_check';
 974                 wfProfileIn( $fname );
 975                 if (serialize($from_lines) != serialize($this->orig()))
 976                         trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
 977                 if (serialize($to_lines) != serialize($this->closing()))
 978                         trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
 979
 980                 $rev = $this->reverse();
 981                 if (serialize($to_lines) != serialize($rev->orig()))
 982                         trigger_error("Reversed original doesn't match", E_USER_ERROR);
 983                 if (serialize($from_lines) != serialize($rev->closing()))
 984                         trigger_error("Reversed closing doesn't match", E_USER_ERROR);
 985
 986
 987                 $prevtype = 'none';
 988                 foreach ($this->edits as $edit) {
 989                         if ( $prevtype == $edit->type )
 990                                 trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
 991                         $prevtype = $edit->type;
 992                 }
 993
 994                 $lcs = $this->lcs();
 995                 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
 996                 wfProfileOut( $fname );
 997         }
 998 }
 999
1000 /**
1001  * FIXME: bad name.
1002  * @todo document
1003  * @access private
1004  */
1005 class MappedDiff extends Diff
1006 {
1007         /**
1008          * Constructor.
1009          *
1010          * Computes diff between sequences of strings.
1011          *
1012          * This can be used to compute things like
1013          * case-insensitve diffs, or diffs which ignore
1014          * changes in white-space.
1015          *
1016          * @param $from_lines array An array of strings.
1017          *      (Typically these are lines from a file.)
1018          *
1019          * @param $to_lines array An array of strings.
1020          *
1021          * @param $mapped_from_lines array This array should
1022          *      have the same size number of elements as $from_lines.
1023          *      The elements in $mapped_from_lines and
1024          *      $mapped_to_lines are what is actually compared
1025          *      when computing the diff.
1026          *
1027          * @param $mapped_to_lines array This array should
1028          *      have the same number of elements as $to_lines.
1029          */
1030         function MappedDiff($from_lines, $to_lines,
1031                                                 $mapped_from_lines, $mapped_to_lines) {
1032                 $fname = 'MappedDiff::MappedDiff';
1033                 wfProfileIn( $fname );
1034
1035                 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1036                 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1037
1038                 $this->Diff($mapped_from_lines, $mapped_to_lines);
1039
1040                 $xi = $yi = 0;
1041                 for ($i = 0; $i < sizeof($this->edits); $i++) {
1042                         $orig = &$this->edits[$i]->orig;
1043                         if (is_array($orig)) {
1044                                 $orig = array_slice($from_lines, $xi, sizeof($orig));
1045                                 $xi += sizeof($orig);
1046                         }
1047
1048                         $closing = &$this->edits[$i]->closing;
1049                         if (is_array($closing)) {
1050                                 $closing = array_slice($to_lines, $yi, sizeof($closing));
1051                                 $yi += sizeof($closing);
1052                         }
1053                 }
1054                 wfProfileOut( $fname );
1055         }
1056 }
1057
1058 /**
1059  * A class to format Diffs
1060  *
1061  * This class formats the diff in classic diff format.
1062  * It is intended that this class be customized via inheritance,
1063  * to obtain fancier outputs.
1064  * @todo document
1065  * @access private
1066  */
1067 class DiffFormatter
1068 {
1069         /**
1070          * Number of leading context "lines" to preserve.
1071          *
1072          * This should be left at zero for this class, but subclasses
1073          * may want to set this to other values.
1074          */
1075         var $leading_context_lines = 0;
1076
1077         /**
1078          * Number of trailing context "lines" to preserve.
1079          *
1080          * This should be left at zero for this class, but subclasses
1081          * may want to set this to other values.
1082          */
1083         var $trailing_context_lines = 0;
1084
1085         /**
1086          * Format a diff.
1087          *
1088          * @param $diff object A Diff object.
1089          * @return string The formatted output.
1090          */
1091         function format($diff) {
1092                 $fname = 'DiffFormatter::format';
1093                 wfProfileIn( $fname );
1094
1095                 $xi = $yi = 1;
1096                 $block = false;
1097                 $context = array();
1098
1099                 $nlead = $this->leading_context_lines;
1100                 $ntrail = $this->trailing_context_lines;
1101
1102                 $this->_start_diff();
1103
1104                 foreach ($diff->edits as $edit) {
1105                         if ($edit->type == 'copy') {
1106                                 if (is_array($block)) {
1107                                         if (sizeof($edit->orig) <= $nlead + $ntrail) {
1108                                                 $block[] = $edit;
1109                                         }
1110                                         else{
1111                                                 if ($ntrail) {
1112                                                         $context = array_slice($edit->orig, 0, $ntrail);
1113                                                         $block[] = new _DiffOp_Copy($context);
1114                                                 }
1115                                                 $this->_block($x0, $ntrail + $xi - $x0,
1116                                                                           $y0, $ntrail + $yi - $y0,
1117                                                                           $block);
1118                                                 $block = false;
1119                                         }
1120                                 }
1121                                 $context = $edit->orig;
1122                         }
1123                         else {
1124                                 if (! is_array($block)) {
1125                                         $context = array_slice($context, sizeof($context) - $nlead);
1126                                         $x0 = $xi - sizeof($context);
1127                                         $y0 = $yi - sizeof($context);
1128                                         $block = array();
1129                                         if ($context)
1130                                                 $block[] = new _DiffOp_Copy($context);
1131                                 }
1132                                 $block[] = $edit;
1133                         }
1134
1135                         if ($edit->orig)
1136                                 $xi += sizeof($edit->orig);
1137                         if ($edit->closing)
1138                                 $yi += sizeof($edit->closing);
1139                 }
1140
1141                 if (is_array($block))
1142                         $this->_block($x0, $xi - $x0,
1143                                                   $y0, $yi - $y0,
1144                                                   $block);
1145
1146                 $end = $this->_end_diff();
1147                 wfProfileOut( $fname );
1148                 return $end;
1149         }
1150
1151         function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1152                 $fname = 'DiffFormatter::_block';
1153                 wfProfileIn( $fname );
1154                 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1155                 foreach ($edits as $edit) {
1156                         if ($edit->type == 'copy')
1157                                 $this->_context($edit->orig);
1158                         elseif ($edit->type == 'add')
1159                                 $this->_added($edit->closing);
1160                         elseif ($edit->type == 'delete')
1161                                 $this->_deleted($edit->orig);
1162                         elseif ($edit->type == 'change')
1163                                 $this->_changed($edit->orig, $edit->closing);
1164                         else
1165                                 trigger_error('Unknown edit type', E_USER_ERROR);
1166                 }
1167                 $this->_end_block();
1168                 wfProfileOut( $fname );
1169         }
1170
1171         function _start_diff() {
1172                 ob_start();
1173         }
1174
1175         function _end_diff() {
1176                 $val = ob_get_contents();
1177                 ob_end_clean();
1178                 return $val;
1179         }
1180
1181         function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1182                 if ($xlen > 1)
1183                         $xbeg .= "," . ($xbeg + $xlen - 1);
1184                 if ($ylen > 1)
1185                         $ybeg .= "," . ($ybeg + $ylen - 1);
1186
1187                 return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1188         }
1189
1190         function _start_block($header) {
1191                 echo $header;
1192         }
1193
1194         function _end_block() {
1195         }
1196
1197         function _lines($lines, $prefix = ' ') {
1198                 foreach ($lines as $line)
1199                         echo "$prefix $line\n";
1200         }
1201
1202         function _context($lines) {
1203                 $this->_lines($lines);
1204         }
1205
1206         function _added($lines) {
1207                 $this->_lines($lines, '>');
1208         }
1209         function _deleted($lines) {
1210                 $this->_lines($lines, '<');
1211         }
1212
1213         function _changed($orig, $closing) {
1214                 $this->_deleted($orig);
1215                 echo "---\n";
1216                 $this->_added($closing);
1217         }
1218 }
1219
1220
1221 /**
1222  *      Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1223  *
1224  */
1225
1226 define('NBSP', '&#160;');                       // iso-8859-x non-breaking space.
1227
1228 /**
1229  * @todo document
1230  * @access private
1231  */
1232 class _HWLDF_WordAccumulator {
1233         function _HWLDF_WordAccumulator () {
1234                 $this->_lines = array();
1235                 $this->_line = '';
1236                 $this->_group = '';
1237                 $this->_tag = '';
1238         }
1239
1240         function _flushGroup ($new_tag) {
1241                 if ($this->_group !== '') {
1242                         if ($this->_tag == 'mark')
1243                                 $this->_line .= '<span class="diffchange">' .
1244                                         htmlspecialchars ( $this->_group ) . '</span>';
1245                         else
1246                                 $this->_line .= htmlspecialchars ( $this->_group );
1247                 }
1248                 $this->_group = '';
1249                 $this->_tag = $new_tag;
1250         }
1251
1252         function _flushLine ($new_tag) {
1253                 $this->_flushGroup($new_tag);
1254                 if ($this->_line != '')
1255                         array_push ( $this->_lines, $this->_line );
1256                 else
1257                         # make empty lines visible by inserting an NBSP
1258                         array_push ( $this->_lines, NBSP );
1259                 $this->_line = '';
1260         }
1261
1262         function addWords ($words, $tag = '') {
1263                 if ($tag != $this->_tag)
1264                         $this->_flushGroup($tag);
1265
1266                 foreach ($words as $word) {
1267                         // new-line should only come as first char of word.
1268                         if ($word == '')
1269                                 continue;
1270                         if ($word[0] == "\n") {
1271                                 $this->_flushLine($tag);
1272                                 $word = substr($word, 1);
1273                         }
1274                         assert(!strstr($word, "\n"));
1275                         $this->_group .= $word;
1276                 }
1277         }
1278
1279         function getLines() {
1280                 $this->_flushLine('~done');
1281                 return $this->_lines;
1282         }
1283 }
1284
1285 /**
1286  * @todo document
1287  * @access private
1288  */
1289 class WordLevelDiff extends MappedDiff
1290 {
1291         function WordLevelDiff ($orig_lines, $closing_lines) {
1292                 $fname = 'WordLevelDiff::WordLevelDiff';
1293                 wfProfileIn( $fname );
1294
1295                 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1296                 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1297
1298
1299                 $this->MappedDiff($orig_words, $closing_words,
1300                                                   $orig_stripped, $closing_stripped);
1301                 wfProfileOut( $fname );
1302         }
1303
1304         function _split($lines) {
1305                 $fname = 'WordLevelDiff::_split';
1306                 wfProfileIn( $fname );
1307                 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1308                                                         implode("\n", $lines),
1309                                                         $m)) {
1310                         wfProfileOut( $fname );
1311                         return array(array(''), array(''));
1312                 }
1313                 wfProfileOut( $fname );
1314                 return array($m[0], $m[1]);
1315         }
1316
1317         function orig () {
1318                 $fname = 'WordLevelDiff::orig';
1319                 wfProfileIn( $fname );
1320                 $orig = new _HWLDF_WordAccumulator;
1321
1322                 foreach ($this->edits as $edit) {
1323                         if ($edit->type == 'copy')
1324                                 $orig->addWords($edit->orig);
1325                         elseif ($edit->orig)
1326                                 $orig->addWords($edit->orig, 'mark');
1327                 }
1328                 $lines = $orig->getLines();
1329                 wfProfileOut( $fname );
1330                 return $lines;
1331         }
1332
1333         function closing () {
1334                 $fname = 'WordLevelDiff::closing';
1335                 wfProfileIn( $fname );
1336                 $closing = new _HWLDF_WordAccumulator;
1337
1338                 foreach ($this->edits as $edit) {
1339                         if ($edit->type == 'copy')
1340                                 $closing->addWords($edit->closing);
1341                         elseif ($edit->closing)
1342                                 $closing->addWords($edit->closing, 'mark');
1343                 }
1344                 $lines = $closing->getLines();
1345                 wfProfileOut( $fname );
1346                 return $lines;
1347         }
1348 }
1349
1350 /**
1351  *      Wikipedia Table style diff formatter.
1352  * @todo document
1353  * @access private
1354  */
1355 class TableDiffFormatter extends DiffFormatter
1356 {
1357         function TableDiffFormatter() {
1358                 $this->leading_context_lines = 2;
1359                 $this->trailing_context_lines = 2;
1360         }
1361
1362         function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1363                 $l1 = wfMsg( 'lineno', $xbeg );
1364                 $l2 = wfMsg( 'lineno', $ybeg );
1365
1366                 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1367                   '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1368                 return $r;
1369         }
1370
1371         function _start_block( $header ) {
1372                 global $wgOut;
1373                 echo $header;
1374         }
1375
1376         function _end_block() {
1377         }
1378
1379         function _lines( $lines, $prefix=' ', $color='white' ) {
1380         }
1381
1382         # HTML-escape parameter before calling this
1383         function addedLine( $line ) {
1384                 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1385         }
1386
1387         # HTML-escape parameter before calling this
1388         function deletedLine( $line ) {
1389                 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1390         }
1391
1392         # HTML-escape parameter before calling this
1393         function contextLine( $line ) {
1394                 return "<td> </td><td class='diff-context'>{$line}</td>";
1395         }
1396
1397         function emptyLine() {
1398                 return '<td colspan="2">&nbsp;</td>';
1399         }
1400
1401         function _added( $lines ) {
1402                 foreach ($lines as $line) {
1403                         echo '<tr>' . $this->emptyLine() .
1404                                 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1405                 }
1406         }
1407
1408         function _deleted($lines) {
1409                 foreach ($lines as $line) {
1410                         echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1411                           $this->emptyLine() . "</tr>\n";
1412                 }
1413         }
1414
1415         function _context( $lines ) {
1416                 foreach ($lines as $line) {
1417                         echo '<tr>' .
1418                                 $this->contextLine( htmlspecialchars ( $line ) ) .
1419                                 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1420                 }
1421         }
1422
1423         function _changed( $orig, $closing ) {
1424                 $fname = 'TableDiffFormatter::_changed';
1425                 wfProfileIn( $fname );
1426
1427                 $diff = new WordLevelDiff( $orig, $closing );
1428                 $del = $diff->orig();
1429                 $add = $diff->closing();
1430
1431                 # Notice that WordLevelDiff returns HTML-escaped output.
1432                 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1433
1434                 while ( $line = array_shift( $del ) ) {
1435                         $aline = array_shift( $add );
1436                         echo '<tr>' . $this->deletedLine( $line ) .
1437                                 $this->addedLine( $aline ) . "</tr>\n";
1438                 }
1439                 foreach ($add as $line) {       # If any leftovers
1440                         echo '<tr>' . $this->emptyLine() .
1441                                 $this->addedLine( $line ) . "</tr>\n";
1442                 }
1443                 wfProfileOut( $fname );
1444         }
1445 }
1446
1447 ?>