ILIAS  eassessment Revision 61809
 All Data Structures Namespaces Files Functions Variables Groups Pages
class.WordLevelDiff.php
Go to the documentation of this file.
1 <?php
2 
3 // Copyright holded by MediaWiki contributers, Licensed under GPL version 2 or later
4 
5 
17 class DifferenceEngine {
21  var $mOldid, $mNewid, $mTitle;
22  var $mOldtitle, $mNewtitle, $mPagetitle;
23  var $mOldtext, $mNewtext;
24  var $mOldPage, $mNewPage;
25  var $mRcidMarkPatrolled;
26  var $mOldRev, $mNewRev;
27  var $mRevisionsLoaded = false; // Have the revisions been loaded
28  var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
38  function DifferenceEngine( $titleObj = null, $old = 0, $new = 0, $rcid = 0 ) {
39  $this->mTitle = $titleObj;
40  wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
41 
42  if ( 'prev' === $new ) {
43  # Show diff between revision $old and the previous one.
44  # Get previous one from DB.
45  #
46  $this->mNewid = intval($old);
47 
48  $this->mOldid = $this->mTitle->getPreviousRevisionID( $this->mNewid );
49 
50  } elseif ( 'next' === $new ) {
51  # Show diff between revision $old and the previous one.
52  # Get previous one from DB.
53  #
54  $this->mOldid = intval($old);
55  $this->mNewid = $this->mTitle->getNextRevisionID( $this->mOldid );
56  if ( false === $this->mNewid ) {
57  # if no result, NewId points to the newest old revision. The only newer
58  # revision is cur, which is "0".
59  $this->mNewid = 0;
60  }
61 
62  } else {
63  $this->mOldid = intval($old);
64  $this->mNewid = intval($new);
65  }
66  $this->mRcidMarkPatrolled = intval($rcid); # force it to be an integer
67  }
68 
69  function showDiffPage( $diffOnly = false ) {
70  global $wgUser, $wgOut, $wgUseExternalEditor, $wgUseRCPatrol;
71  $fname = 'DifferenceEngine::showDiffPage';
72  //wfProfileIn( $fname );
73 
74  # If external diffs are enabled both globally and for the user,
75  # we'll use the application/x-external-editor interface to call
76  # an external diff tool like kompare, kdiff3, etc.
77  if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
78  global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
79  $wgOut->disable();
80  header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
81  $url1=$this->mTitle->getFullURL("action=raw&oldid=".$this->mOldid);
82  $url2=$this->mTitle->getFullURL("action=raw&oldid=".$this->mNewid);
83  $special=$wgLang->getNsText(NS_SPECIAL);
84  $control=<<<CONTROL
85 [Process]
86 Type=Diff text
87 Engine=MediaWiki
88 Script={$wgServer}{$wgScript}
89 Special namespace={$special}
90 
91 [File]
92 Extension=wiki
93 URL=$url1
94 
95 [File 2]
96 Extension=wiki
97 URL=$url2
98 CONTROL;
99  echo($control);
100  return;
101  }
102 
103  $wgOut->setArticleFlag( false );
104  if ( ! $this->loadRevisionData() ) {
105  $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, {$this->mNewid})";
106  $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
107  $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
108  $wgOut->addWikitext( $mtext );
109  //wfProfileOut( $fname );
110  return;
111  }
112 
113  wfRunHooks( 'DiffViewHeader', array( $this, $this->mOldRev, $this->mNewRev ) );
114 
115  if ( $this->mNewRev->isCurrent() ) {
116  $wgOut->setArticleFlag( true );
117  }
118 
119  # mOldid is false if the difference engine is called with a "vague" query for
120  # a diff between a version V and its previous version V' AND the version V
121  # is the first version of that article. In that case, V' does not exist.
122  if ( $this->mOldid === false ) {
123  $this->showFirstRevision();
124  $this->renderNewRevision(); // should we respect $diffOnly here or not?
125  //wfProfileOut( $fname );
126  return;
127  }
128 
129  $wgOut->suppressQuickbar();
130 
131  $oldTitle = $this->mOldPage->getPrefixedText();
132  $newTitle = $this->mNewPage->getPrefixedText();
133  if( $oldTitle == $newTitle ) {
134  $wgOut->setPageTitle( $newTitle );
135  } else {
136  $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
137  }
138  $wgOut->setSubtitle( wfMsg( 'difference' ) );
139  $wgOut->setRobotpolicy( 'noindex,nofollow' );
140 
141  if ( !( $this->mOldPage->userCanRead() && $this->mNewPage->userCanRead() ) ) {
142  $wgOut->loginToUse();
143  $wgOut->output();
144  //wfProfileOut( $fname );
145  exit;
146  }
147 
148  $sk = $wgUser->getSkin();
149 
150  if ( $this->mNewRev->isCurrent() && $wgUser->isAllowed('rollback') ) {
151  $rollback = '&nbsp;&nbsp;&nbsp;' . $sk->generateRollback( $this->mNewRev );
152  } else {
153  $rollback = '';
154  }
155  if( $wgUseRCPatrol && $this->mRcidMarkPatrolled != 0 && $wgUser->isAllowed( 'patrol' ) ) {
156  $patrol = ' [' . $sk->makeKnownLinkObj( $this->mTitle, wfMsg( 'markaspatrolleddiff' ), "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
157  } else {
158  $patrol = '';
159  }
160 
161  $prevlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'previousdiff' ),
162  'diff=prev&oldid='.$this->mOldid, '', '', 'id="differences-prevlink"' );
163  if ( $this->mNewRev->isCurrent() ) {
164  $nextlink = '&nbsp;';
165  } else {
166  $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ),
167  'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
168  }
169 
170  $oldminor = '';
171  $newminor = '';
172 
173  if ($this->mOldRev->mMinorEdit == 1) {
174  $oldminor = wfElement( 'span', array( 'class' => 'minor' ),
175  wfMsg( 'minoreditletter') ) . ' ';
176  }
177 
178  if ($this->mNewRev->mMinorEdit == 1) {
179  $newminor = wfElement( 'span', array( 'class' => 'minor' ),
180  wfMsg( 'minoreditletter') ) . ' ';
181  }
182 
183  $oldHeader = "<strong>{$this->mOldtitle}</strong><br />" .
184  $sk->revUserTools( $this->mOldRev ) . "<br />" .
185  $oldminor . $sk->revComment( $this->mOldRev, !$diffOnly ) . "<br />" .
186  $prevlink;
187  $newHeader = "<strong>{$this->mNewtitle}</strong><br />" .
188  $sk->revUserTools( $this->mNewRev ) . " $rollback<br />" .
189  $newminor . $sk->revComment( $this->mNewRev, !$diffOnly ) . "<br />" .
190  $nextlink . $patrol;
191 
192  $this->showDiff( $oldHeader, $newHeader );
193 
194  if ( !$diffOnly )
195  $this->renderNewRevision();
196 
197  //wfProfileOut( $fname );
198  }
199 
203  function renderNewRevision() {
204  global $wgOut;
205  $fname = 'DifferenceEngine::renderNewRevision';
206  //wfProfileIn( $fname );
207 
208  $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
209  #add deleted rev tag if needed
210  if ( !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
211  $wgOut->addWikiText( wfMsg( 'rev-deleted-text-permission' ) );
212  }
213 
214  if( !$this->mNewRev->isCurrent() ) {
215  $oldEditSectionSetting = $wgOut->parserOptions()->setEditSection( false );
216  }
217 
218  $this->loadNewText();
219  if( is_object( $this->mNewRev ) ) {
220  $wgOut->setRevisionId( $this->mNewRev->getId() );
221  }
222 
223  $wgOut->addWikiTextTidy( $this->mNewtext );
224 
225  if( !$this->mNewRev->isCurrent() ) {
226  $wgOut->parserOptions()->setEditSection( $oldEditSectionSetting );
227  }
228 
229  //wfProfileOut( $fname );
230  }
231 
236  function showFirstRevision() {
237  global $wgOut, $wgUser;
238 
239  $fname = 'DifferenceEngine::showFirstRevision';
240  //wfProfileIn( $fname );
241 
242  # Get article text from the DB
243  #
244  if ( ! $this->loadNewText() ) {
245  $t = $this->mTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
246  "{$this->mNewid})";
247  $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
248  $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
249  $wgOut->addWikitext( $mtext );
250  //wfProfileOut( $fname );
251  return;
252  }
253  if ( $this->mNewRev->isCurrent() ) {
254  $wgOut->setArticleFlag( true );
255  }
256 
257  # Check if user is allowed to look at this page. If not, bail out.
258  #
259  if ( !( $this->mTitle->userCanRead() ) ) {
260  $wgOut->loginToUse();
261  $wgOut->output();
262  //wfProfileOut( $fname );
263  exit;
264  }
265 
266  # Prepare the header box
267  #
268  $sk = $wgUser->getSkin();
269 
270  $nextlink = $sk->makeKnownLinkObj( $this->mTitle, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid, '', '', 'id="differences-nextlink"' );
271  $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />" .
272  $sk->revUserTools( $this->mNewRev ) . "<br />" .
273  $sk->revComment( $this->mNewRev ) . "<br />" .
274  $nextlink . "</div>\n";
275 
276  $wgOut->addHTML( $header );
277 
278  $wgOut->setSubtitle( wfMsg( 'difference' ) );
279  $wgOut->setRobotpolicy( 'noindex,nofollow' );
280 
281  //wfProfileOut( $fname );
282  }
283 
288  function showDiff( $otitle, $ntitle ) {
289  global $wgOut;
290  $diff = $this->getDiff( $otitle, $ntitle );
291  if ( $diff === false ) {
292  $wgOut->addWikitext( wfMsg( 'missingarticle', "<nowiki>(fixme, bug)</nowiki>" ) );
293  return false;
294  } else {
295  $wgOut->addHTML( $diff );
296  return true;
297  }
298  }
299 
305  function getDiff( $otitle, $ntitle ) {
306  $body = $this->getDiffBody();
307  if ( $body === false ) {
308  return false;
309  } else {
310  $multi = $this->getMultiNotice();
311  return $this->addHeader( $body, $otitle, $ntitle, $multi );
312  }
313  }
314 
320  function getDiffBody() {
321  global $wgMemc;
322  $fname = 'DifferenceEngine::getDiffBody';
323  //wfProfileIn( $fname );
324 
325  // Cacheable?
326  $key = false;
327  if ( $this->mOldid && $this->mNewid ) {
328  // Try cache
329  $key = wfMemcKey( 'diff', 'oldid', $this->mOldid, 'newid', $this->mNewid );
330  $difftext = $wgMemc->get( $key );
331  if ( $difftext ) {
332  wfIncrStats( 'diff_cache_hit' );
333  $difftext = $this->localiseLineNumbers( $difftext );
334  $difftext .= "\n<!-- diff cache key $key -->\n";
335  //wfProfileOut( $fname );
336  return $difftext;
337  }
338  }
339 
340  #loadtext is permission safe, this just clears out the diff
341  if ( !$this->loadText() ) {
342  //wfProfileOut( $fname );
343  return false;
344  } else if ( $this->mOldRev && !$this->mOldRev->userCan(Revision::DELETED_TEXT) ) {
345  return '';
346  } else if ( $this->mNewRev && !$this->mNewRev->userCan(Revision::DELETED_TEXT) ) {
347  return '';
348  }
349 
350  $difftext = $this->generateDiffBody( $this->mOldtext, $this->mNewtext );
351 
352  // Save to cache for 7 days
353  if ( $key !== false && $difftext !== false ) {
354  wfIncrStats( 'diff_cache_miss' );
355  $wgMemc->set( $key, $difftext, 7*86400 );
356  } else {
357  wfIncrStats( 'diff_uncacheable' );
358  }
359  // Replace line numbers with the text in the user's language
360  if ( $difftext !== false ) {
361  $difftext = $this->localiseLineNumbers( $difftext );
362  }
363  //wfProfileOut( $fname );
364  return $difftext;
365  }
366 
371  function generateDiffBody( $otext, $ntext ) {
372  global $wgExternalDiffEngine, $wgContLang;
373  $fname = 'DifferenceEngine::generateDiffBody';
374 
375  $otext = str_replace( "\r\n", "\n", $otext );
376  $ntext = str_replace( "\r\n", "\n", $ntext );
377 
378  if ( $wgExternalDiffEngine == 'wikidiff' ) {
379  # For historical reasons, external diff engine expects
380  # input text to be HTML-escaped already
381  $otext = htmlspecialchars ( $wgContLang->segmentForDiff( $otext ) );
382  $ntext = htmlspecialchars ( $wgContLang->segmentForDiff( $ntext ) );
383  if( !function_exists( 'wikidiff_do_diff' ) ) {
384  dl('php_wikidiff.so');
385  }
386  return $wgContLang->unsegementForDiff( wikidiff_do_diff( $otext, $ntext, 2 ) );
387  }
388 
389  if ( $wgExternalDiffEngine == 'wikidiff2' ) {
390  # Better external diff engine, the 2 may some day be dropped
391  # This one does the escaping and segmenting itself
392  if ( !function_exists( 'wikidiff2_do_diff' ) ) {
393  //wfProfileIn( "$fname-dl" );
394  @dl('php_wikidiff2.so');
395  //wfProfileOut( "$fname-dl" );
396  }
397  if ( function_exists( 'wikidiff2_do_diff' ) ) {
398  //wfProfileIn( 'wikidiff2_do_diff' );
399  $text = wikidiff2_do_diff( $otext, $ntext, 2 );
400  //wfProfileOut( 'wikidiff2_do_diff' );
401  return $text;
402  }
403  }
404  if ( $wgExternalDiffEngine !== false ) {
405  # Diff via the shell
406  global $wgTmpDirectory;
407  $tempName1 = tempnam( $wgTmpDirectory, 'diff_' );
408  $tempName2 = tempnam( $wgTmpDirectory, 'diff_' );
409 
410  $tempFile1 = fopen( $tempName1, "w" );
411  if ( !$tempFile1 ) {
412  //wfProfileOut( $fname );
413  return false;
414  }
415  $tempFile2 = fopen( $tempName2, "w" );
416  if ( !$tempFile2 ) {
417  //wfProfileOut( $fname );
418  return false;
419  }
420  fwrite( $tempFile1, $otext );
421  fwrite( $tempFile2, $ntext );
422  fclose( $tempFile1 );
423  fclose( $tempFile2 );
424  $cmd = wfEscapeShellArg( $wgExternalDiffEngine, $tempName1, $tempName2 );
425  //wfProfileIn( "$fname-shellexec" );
426  $difftext = wfShellExec( $cmd );
427  //wfProfileOut( "$fname-shellexec" );
428  unlink( $tempName1 );
429  unlink( $tempName2 );
430  return $difftext;
431  }
432 
433  # Native PHP diff
434  $ota = explode( "\n", $wgContLang->segmentForDiff( $otext ) );
435  $nta = explode( "\n", $wgContLang->segmentForDiff( $ntext ) );
436  $diffs = new Diff( $ota, $nta );
437  $formatter = new TableDiffFormatter();
438  return $wgContLang->unsegmentForDiff( $formatter->format( $diffs ) );
439  }
440 
441 
445  function localiseLineNumbers( $text ) {
446  return preg_replace_callback( '/<!--LINE (\d+)-->/',
447  array( &$this, 'localiseLineNumbersCb' ), $text );
448  }
449 
450  function localiseLineNumbersCb( $matches ) {
451  global $wgLang;
452  return wfMsgExt( 'lineno', array('parseinline'), $wgLang->formatNum( $matches[1] ) );
453  }
454 
455 
459  function getMultiNotice() {
460  if ( !is_object($this->mOldRev) || !is_object($this->mNewRev) )
461  return '';
462 
463  if( !$this->mOldPage->equals( $this->mNewPage ) ) {
464  // Comparing two different pages? Count would be meaningless.
465  return '';
466  }
467 
468  $oldid = $this->mOldRev->getId();
469  $newid = $this->mNewRev->getId();
470  if ( $oldid > $newid ) {
471  $tmp = $oldid; $oldid = $newid; $newid = $tmp;
472  }
473 
474  $n = $this->mTitle->countRevisionsBetween( $oldid, $newid );
475  if ( !$n )
476  return '';
477 
478  return wfMsgExt( 'diff-multi', array( 'parseinline' ), $n );
479  }
480 
481 
485  function addHeader( $diff, $otitle, $ntitle, $multi = '' ) {
486  global $wgOut;
487 
488  if ( $this->mOldRev && $this->mOldRev->isDeleted(Revision::DELETED_TEXT) ) {
489  $otitle = '<span class="history-deleted">'.$otitle.'</span>';
490  }
491  if ( $this->mNewRev && $this->mNewRev->isDeleted(Revision::DELETED_TEXT) ) {
492  $ntitle = '<span class="history-deleted">'.$ntitle.'</span>';
493  }
494  $header = "
495  <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
496  <tr>
497  <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
498  <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
499  </tr>
500  ";
501 
502  if ( $multi != '' )
503  $header .= "<tr><td colspan='4' align='center' class='diff-multi'>{$multi}</td></tr>";
504 
505  return $header . $diff . "</table>";
506  }
507 
511  function setText( $oldText, $newText ) {
512  $this->mOldtext = $oldText;
513  $this->mNewtext = $newText;
514  $this->mTextLoaded = 2;
515  }
516 
527  function loadRevisionData() {
528  global $wgLang;
529  if ( $this->mRevisionsLoaded ) {
530  return true;
531  } else {
532  // Whether it succeeds or fails, we don't want to try again
533  $this->mRevisionsLoaded = true;
534  }
535 
536  // Load the new revision object
537  if( $this->mNewid ) {
538  $this->mNewRev = Revision::newFromId( $this->mNewid );
539  } else {
540  $this->mNewRev = Revision::newFromTitle( $this->mTitle );
541  }
542 
543  if( is_null( $this->mNewRev ) ) {
544  return false;
545  }
546 
547  // Set assorted variables
548  $timestamp = $wgLang->timeanddate( $this->mNewRev->getTimestamp(), true );
549  $this->mNewPage = $this->mNewRev->getTitle();
550  if( $this->mNewRev->isCurrent() ) {
551  $newLink = $this->mNewPage->escapeLocalUrl();
552  $this->mPagetitle = htmlspecialchars( wfMsg( 'currentrev' ) );
553  $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit' );
554 
555  $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a> ($timestamp)"
556  . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
557 
558  } else {
559  $newLink = $this->mNewPage->escapeLocalUrl( 'oldid=' . $this->mNewid );
560  $newEdit = $this->mNewPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mNewid );
561  $this->mPagetitle = htmlspecialchars( wfMsg( 'revisionasof', $timestamp ) );
562 
563  $this->mNewtitle = "<a href='$newLink'>{$this->mPagetitle}</a>"
564  . " (<a href='$newEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
565  }
566 
567  // Load the old revision object
568  $this->mOldRev = false;
569  if( $this->mOldid ) {
570  $this->mOldRev = Revision::newFromId( $this->mOldid );
571  } elseif ( $this->mOldid === 0 ) {
572  $rev = $this->mNewRev->getPrevious();
573  if( $rev ) {
574  $this->mOldid = $rev->getId();
575  $this->mOldRev = $rev;
576  } else {
577  // No previous revision; mark to show as first-version only.
578  $this->mOldid = false;
579  $this->mOldRev = false;
580  }
581  }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
582 
583  if( is_null( $this->mOldRev ) ) {
584  return false;
585  }
586 
587  if ( $this->mOldRev ) {
588  $this->mOldPage = $this->mOldRev->getTitle();
589 
590  $t = $wgLang->timeanddate( $this->mOldRev->getTimestamp(), true );
591  $oldLink = $this->mOldPage->escapeLocalUrl( 'oldid=' . $this->mOldid );
592  $oldEdit = $this->mOldPage->escapeLocalUrl( 'action=edit&oldid=' . $this->mOldid );
593  $this->mOldtitle = "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) )
594  . "</a> (<a href='$oldEdit'>" . htmlspecialchars( wfMsg( 'editold' ) ) . "</a>)";
595  //now that we considered old rev, we can make undo link (bug 8133, multi-edit undo)
596  $newUndo = $this->mNewPage->escapeLocalUrl( 'action=edit&undoafter=' . $this->mOldid . '&undo=' . $this->mNewid);
597  $this->mNewtitle .= " (<a href='$newUndo'>" . htmlspecialchars( wfMsg( 'editundo' ) ) . "</a>)";
598  }
599 
600  return true;
601  }
602 
606  function loadText() {
607  if ( $this->mTextLoaded == 2 ) {
608  return true;
609  } else {
610  // Whether it succeeds or fails, we don't want to try again
611  $this->mTextLoaded = 2;
612  }
613 
614  if ( !$this->loadRevisionData() ) {
615  return false;
616  }
617  if ( $this->mOldRev ) {
618  // FIXME: permission tests
619  $this->mOldtext = $this->mOldRev->revText();
620  if ( $this->mOldtext === false ) {
621  return false;
622  }
623  }
624  if ( $this->mNewRev ) {
625  $this->mNewtext = $this->mNewRev->revText();
626  if ( $this->mNewtext === false ) {
627  return false;
628  }
629  }
630  return true;
631  }
632 
636  function loadNewText() {
637  if ( $this->mTextLoaded >= 1 ) {
638  return true;
639  } else {
640  $this->mTextLoaded = 1;
641  }
642  if ( !$this->loadRevisionData() ) {
643  return false;
644  }
645  $this->mNewtext = $this->mNewRev->getText();
646  return true;
647  }
648 
649 
650 }
651 
652 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
653 //
654 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
655 // You may copy this code freely under the conditions of the GPL.
656 //
657 
658 define('USE_ASSERTS', function_exists('assert'));
659 
665 class _DiffOp {
666  var $type;
667  var $orig;
668  var $closing;
669 
670  function reverse() {
671  trigger_error('pure virtual', E_USER_ERROR);
672  }
673 
674  function norig() {
675  return $this->orig ? sizeof($this->orig) : 0;
676  }
677 
678  function nclosing() {
679  return $this->closing ? sizeof($this->closing) : 0;
680  }
681 }
682 
688 class _DiffOp_Copy extends _DiffOp {
689  var $type = 'copy';
690 
691  function _DiffOp_Copy ($orig, $closing = false) {
692  if (!is_array($closing))
693  $closing = $orig;
694  $this->orig = $orig;
695  $this->closing = $closing;
696  }
697 
698  function reverse() {
699  return new _DiffOp_Copy($this->closing, $this->orig);
700  }
701 }
702 
708 class _DiffOp_Delete extends _DiffOp {
709  var $type = 'delete';
710 
711  function _DiffOp_Delete ($lines) {
712  $this->orig = $lines;
713  $this->closing = false;
714  }
715 
716  function reverse() {
717  return new _DiffOp_Add($this->orig);
718  }
719 }
720 
726 class _DiffOp_Add extends _DiffOp {
727  var $type = 'add';
728 
729  function _DiffOp_Add ($lines) {
730  $this->closing = $lines;
731  $this->orig = false;
732  }
733 
734  function reverse() {
735  return new _DiffOp_Delete($this->closing);
736  }
737 }
738 
744 class _DiffOp_Change extends _DiffOp {
745  var $type = 'change';
746 
747  function _DiffOp_Change ($orig, $closing) {
748  $this->orig = $orig;
749  $this->closing = $closing;
750  }
751 
752  function reverse() {
753  return new _DiffOp_Change($this->closing, $this->orig);
754  }
755 }
756 
757 
781 class _DiffEngine
782 {
783  const MAX_XREF_LENGTH = 10000;
784 
785  function diff ($from_lines, $to_lines) {
786  $fname = '_DiffEngine::diff';
787  //wfProfileIn( $fname );
788 
789  $n_from = sizeof($from_lines);
790  $n_to = sizeof($to_lines);
791 
792  $this->xchanged = $this->ychanged = array();
793  $this->xv = $this->yv = array();
794  $this->xind = $this->yind = array();
795  unset($this->seq);
796  unset($this->in_seq);
797  unset($this->lcs);
798 
799  // Skip leading common lines.
800  for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
801  if ($from_lines[$skip] !== $to_lines[$skip])
802  break;
803  $this->xchanged[$skip] = $this->ychanged[$skip] = false;
804  }
805  // Skip trailing common lines.
806  $xi = $n_from; $yi = $n_to;
807  for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
808  if ($from_lines[$xi] !== $to_lines[$yi])
809  break;
810  $this->xchanged[$xi] = $this->ychanged[$yi] = false;
811  }
812 
813  // Ignore lines which do not exist in both files.
814  for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
815  $xhash[$this->_line_hash($from_lines[$xi])] = 1;
816  }
817 
818  for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
819  $line = $to_lines[$yi];
820  if ( ($this->ychanged[$yi] = empty($xhash[$this->_line_hash($line)])) )
821  continue;
822  $yhash[$this->_line_hash($line)] = 1;
823  $this->yv[] = $line;
824  $this->yind[] = $yi;
825  }
826  for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
827  $line = $from_lines[$xi];
828  if ( ($this->xchanged[$xi] = empty($yhash[$this->_line_hash($line)])) )
829  continue;
830  $this->xv[] = $line;
831  $this->xind[] = $xi;
832  }
833 
834  // Find the LCS.
835  $this->_compareseq(0, sizeof($this->xv), 0, sizeof($this->yv));
836 
837  // Merge edits when possible
838  $this->_shift_boundaries($from_lines, $this->xchanged, $this->ychanged);
839  $this->_shift_boundaries($to_lines, $this->ychanged, $this->xchanged);
840 
841  // Compute the edit operations.
842  $edits = array();
843  $xi = $yi = 0;
844  while ($xi < $n_from || $yi < $n_to) {
845  USE_ASSERTS && assert($yi < $n_to || $this->xchanged[$xi]);
846  USE_ASSERTS && assert($xi < $n_from || $this->ychanged[$yi]);
847 
848  // Skip matching "snake".
849  $copy = array();
850  while ( $xi < $n_from && $yi < $n_to
851  && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
852  $copy[] = $from_lines[$xi++];
853  ++$yi;
854  }
855  if ($copy)
856  $edits[] = new _DiffOp_Copy($copy);
857 
858  // Find deletes & adds.
859  $delete = array();
860  while ($xi < $n_from && $this->xchanged[$xi])
861  $delete[] = $from_lines[$xi++];
862 
863  $add = array();
864  while ($yi < $n_to && $this->ychanged[$yi])
865  $add[] = $to_lines[$yi++];
866 
867  if ($delete && $add)
868  $edits[] = new _DiffOp_Change($delete, $add);
869  elseif ($delete)
870  $edits[] = new _DiffOp_Delete($delete);
871  elseif ($add)
872  $edits[] = new _DiffOp_Add($add);
873  }
874  //wfProfileOut( $fname );
875  return $edits;
876  }
877 
881  function _line_hash( $line ) {
882  if ( strlen( $line ) > self::MAX_XREF_LENGTH ) {
883  return md5( $line );
884  } else {
885  return $line;
886  }
887  }
888 
889 
890  /* Divide the Largest Common Subsequence (LCS) of the sequences
891  * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
892  * sized segments.
893  *
894  * Returns (LCS, PTS). LCS is the length of the LCS. PTS is an
895  * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
896  * sub sequences. The first sub-sequence is contained in [X0, X1),
897  * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on. Note
898  * that (X0, Y0) == (XOFF, YOFF) and
899  * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
900  *
901  * This function assumes that the first lines of the specified portions
902  * of the two files do not match, and likewise that the last lines do not
903  * match. The caller must trim matching lines from the beginning and end
904  * of the portions it is going to specify.
905  */
906  function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
907  $fname = '_DiffEngine::_diag';
908  //wfProfileIn( $fname );
909  $flip = false;
910 
911  if ($xlim - $xoff > $ylim - $yoff) {
912  // Things seems faster (I'm not sure I understand why)
913  // when the shortest sequence in X.
914  $flip = true;
915  list ($xoff, $xlim, $yoff, $ylim)
916  = array( $yoff, $ylim, $xoff, $xlim);
917  }
918 
919  if ($flip)
920  for ($i = $ylim - 1; $i >= $yoff; $i--)
921  $ymatches[$this->xv[$i]][] = $i;
922  else
923  for ($i = $ylim - 1; $i >= $yoff; $i--)
924  $ymatches[$this->yv[$i]][] = $i;
925 
926  $this->lcs = 0;
927  $this->seq[0]= $yoff - 1;
928  $this->in_seq = array();
929  $ymids[0] = array();
930 
931  $numer = $xlim - $xoff + $nchunks - 1;
932  $x = $xoff;
933  for ($chunk = 0; $chunk < $nchunks; $chunk++) {
934  //wfProfileIn( "$fname-chunk" );
935  if ($chunk > 0)
936  for ($i = 0; $i <= $this->lcs; $i++)
937  $ymids[$i][$chunk-1] = $this->seq[$i];
938 
939  $x1 = $xoff + (int)(($numer + ($xlim-$xoff)*$chunk) / $nchunks);
940  for ( ; $x < $x1; $x++) {
941  $line = $flip ? $this->yv[$x] : $this->xv[$x];
942  if (empty($ymatches[$line]))
943  continue;
944  $matches = $ymatches[$line];
945  reset($matches);
946  while (list ($junk, $y) = each($matches))
947  if (empty($this->in_seq[$y])) {
948  $k = $this->_lcs_pos($y);
949  USE_ASSERTS && assert($k > 0);
950  $ymids[$k] = $ymids[$k-1];
951  break;
952  }
953  while (list ( /* $junk */, $y) = each($matches)) {
954  if ($y > $this->seq[$k-1]) {
955  USE_ASSERTS && assert($y < $this->seq[$k]);
956  // Optimization: this is a common case:
957  // next match is just replacing previous match.
958  $this->in_seq[$this->seq[$k]] = false;
959  $this->seq[$k] = $y;
960  $this->in_seq[$y] = 1;
961  } else if (empty($this->in_seq[$y])) {
962  $k = $this->_lcs_pos($y);
963  USE_ASSERTS && assert($k > 0);
964  $ymids[$k] = $ymids[$k-1];
965  }
966  }
967  }
968  //wfProfileOut( "$fname-chunk" );
969  }
970 
971  $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
972  $ymid = $ymids[$this->lcs];
973  for ($n = 0; $n < $nchunks - 1; $n++) {
974  $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
975  $y1 = $ymid[$n] + 1;
976  $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
977  }
978  $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
979 
980  //wfProfileOut( $fname );
981  return array($this->lcs, $seps);
982  }
983 
984  function _lcs_pos ($ypos) {
985  $fname = '_DiffEngine::_lcs_pos';
986  //wfProfileIn( $fname );
987 
988  $end = $this->lcs;
989  if ($end == 0 || $ypos > $this->seq[$end]) {
990  $this->seq[++$this->lcs] = $ypos;
991  $this->in_seq[$ypos] = 1;
992  //wfProfileOut( $fname );
993  return $this->lcs;
994  }
995 
996  $beg = 1;
997  while ($beg < $end) {
998  $mid = (int)(($beg + $end) / 2);
999  if ( $ypos > $this->seq[$mid] )
1000  $beg = $mid + 1;
1001  else
1002  $end = $mid;
1003  }
1004 
1005  USE_ASSERTS && assert($ypos != $this->seq[$end]);
1006 
1007  $this->in_seq[$this->seq[$end]] = false;
1008  $this->seq[$end] = $ypos;
1009  $this->in_seq[$ypos] = 1;
1010  //wfProfileOut( $fname );
1011  return $end;
1012  }
1013 
1014  /* Find LCS of two sequences.
1015  *
1016  * The results are recorded in the vectors $this->{x,y}changed[], by
1017  * storing a 1 in the element for each line that is an insertion
1018  * or deletion (ie. is not in the LCS).
1019  *
1020  * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
1021  *
1022  * Note that XLIM, YLIM are exclusive bounds.
1023  * All line numbers are origin-0 and discarded lines are not counted.
1024  */
1025  function _compareseq ($xoff, $xlim, $yoff, $ylim) {
1026  $fname = '_DiffEngine::_compareseq';
1027  //wfProfileIn( $fname );
1028 
1029  // Slide down the bottom initial diagonal.
1030  while ($xoff < $xlim && $yoff < $ylim
1031  && $this->xv[$xoff] == $this->yv[$yoff]) {
1032  ++$xoff;
1033  ++$yoff;
1034  }
1035 
1036  // Slide up the top initial diagonal.
1037  while ($xlim > $xoff && $ylim > $yoff
1038  && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
1039  --$xlim;
1040  --$ylim;
1041  }
1042 
1043  if ($xoff == $xlim || $yoff == $ylim)
1044  $lcs = 0;
1045  else {
1046  // This is ad hoc but seems to work well.
1047  //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
1048  //$nchunks = max(2,min(8,(int)$nchunks));
1049  $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
1050  list ($lcs, $seps)
1051  = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
1052  }
1053 
1054  if ($lcs == 0) {
1055  // X and Y sequences have no common subsequence:
1056  // mark all changed.
1057  while ($yoff < $ylim)
1058  $this->ychanged[$this->yind[$yoff++]] = 1;
1059  while ($xoff < $xlim)
1060  $this->xchanged[$this->xind[$xoff++]] = 1;
1061  } else {
1062  // Use the partitions to split this problem into subproblems.
1063  reset($seps);
1064  $pt1 = $seps[0];
1065  while ($pt2 = next($seps)) {
1066  $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
1067  $pt1 = $pt2;
1068  }
1069  }
1070  //wfProfileOut( $fname );
1071  }
1072 
1073  /* Adjust inserts/deletes of identical lines to join changes
1074  * as much as possible.
1075  *
1076  * We do something when a run of changed lines include a
1077  * line at one end and has an excluded, identical line at the other.
1078  * We are free to choose which identical line is included.
1079  * `compareseq' usually chooses the one at the beginning,
1080  * but usually it is cleaner to consider the following identical line
1081  * to be the "change".
1082  *
1083  * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
1084  */
1085  function _shift_boundaries ($lines, &$changed, $other_changed) {
1086  $fname = '_DiffEngine::_shift_boundaries';
1087  //wfProfileIn( $fname );
1088  $i = 0;
1089  $j = 0;
1090 
1091  USE_ASSERTS && assert('sizeof($lines) == sizeof($changed)');
1092  $len = sizeof($lines);
1093  $other_len = sizeof($other_changed);
1094 
1095  while (1) {
1096  /*
1097  * Scan forwards to find beginning of another run of changes.
1098  * Also keep track of the corresponding point in the other file.
1099  *
1100  * Throughout this code, $i and $j are adjusted together so that
1101  * the first $i elements of $changed and the first $j elements
1102  * of $other_changed both contain the same number of zeros
1103  * (unchanged lines).
1104  * Furthermore, $j is always kept so that $j == $other_len or
1105  * $other_changed[$j] == false.
1106  */
1107  while ($j < $other_len && $other_changed[$j])
1108  $j++;
1109 
1110  while ($i < $len && ! $changed[$i]) {
1111  USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1112  $i++; $j++;
1113  while ($j < $other_len && $other_changed[$j])
1114  $j++;
1115  }
1116 
1117  if ($i == $len)
1118  break;
1119 
1120  $start = $i;
1121 
1122  // Find the end of this run of changes.
1123  while (++$i < $len && $changed[$i])
1124  continue;
1125 
1126  do {
1127  /*
1128  * Record the length of this run of changes, so that
1129  * we can later determine whether the run has grown.
1130  */
1131  $runlength = $i - $start;
1132 
1133  /*
1134  * Move the changed region back, so long as the
1135  * previous unchanged line matches the last changed one.
1136  * This merges with previous changed regions.
1137  */
1138  while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1139  $changed[--$start] = 1;
1140  $changed[--$i] = false;
1141  while ($start > 0 && $changed[$start - 1])
1142  $start--;
1143  USE_ASSERTS && assert('$j > 0');
1144  while ($other_changed[--$j])
1145  continue;
1146  USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1147  }
1148 
1149  /*
1150  * Set CORRESPONDING to the end of the changed run, at the last
1151  * point where it corresponds to a changed run in the other file.
1152  * CORRESPONDING == LEN means no such point has been found.
1153  */
1154  $corresponding = $j < $other_len ? $i : $len;
1155 
1156  /*
1157  * Move the changed region forward, so long as the
1158  * first changed line matches the following unchanged one.
1159  * This merges with following changed regions.
1160  * Do this second, so that if there are no merges,
1161  * the changed region is moved forward as far as possible.
1162  */
1163  while ($i < $len && $lines[$start] == $lines[$i]) {
1164  $changed[$start++] = false;
1165  $changed[$i++] = 1;
1166  while ($i < $len && $changed[$i])
1167  $i++;
1168 
1169  USE_ASSERTS && assert('$j < $other_len && ! $other_changed[$j]');
1170  $j++;
1171  if ($j < $other_len && $other_changed[$j]) {
1172  $corresponding = $i;
1173  while ($j < $other_len && $other_changed[$j])
1174  $j++;
1175  }
1176  }
1177  } while ($runlength != $i - $start);
1178 
1179  /*
1180  * If possible, move the fully-merged run of changes
1181  * back to a corresponding run in the other file.
1182  */
1183  while ($corresponding < $i) {
1184  $changed[--$start] = 1;
1185  $changed[--$i] = 0;
1186  USE_ASSERTS && assert('$j > 0');
1187  while ($other_changed[--$j])
1188  continue;
1189  USE_ASSERTS && assert('$j >= 0 && !$other_changed[$j]');
1190  }
1191  }
1192  //wfProfileOut( $fname );
1193  }
1194 }
1195 
1202 class Diff
1203 {
1204  var $edits;
1205 
1214  function Diff($from_lines, $to_lines) {
1215  $eng = new _DiffEngine;
1216  $this->edits = $eng->diff($from_lines, $to_lines);
1217  //$this->_check($from_lines, $to_lines);
1218  }
1219 
1230  function reverse () {
1231  $rev = $this;
1232  $rev->edits = array();
1233  foreach ($this->edits as $edit) {
1234  $rev->edits[] = $edit->reverse();
1235  }
1236  return $rev;
1237  }
1238 
1244  function isEmpty () {
1245  foreach ($this->edits as $edit) {
1246  if ($edit->type != 'copy')
1247  return false;
1248  }
1249  return true;
1250  }
1251 
1259  function lcs () {
1260  $lcs = 0;
1261  foreach ($this->edits as $edit) {
1262  if ($edit->type == 'copy')
1263  $lcs += sizeof($edit->orig);
1264  }
1265  return $lcs;
1266  }
1267 
1276  function orig() {
1277  $lines = array();
1278 
1279  foreach ($this->edits as $edit) {
1280  if ($edit->orig)
1281  array_splice($lines, sizeof($lines), 0, $edit->orig);
1282  }
1283  return $lines;
1284  }
1285 
1294  function closing() {
1295  $lines = array();
1296 
1297  foreach ($this->edits as $edit) {
1298  if ($edit->closing)
1299  array_splice($lines, sizeof($lines), 0, $edit->closing);
1300  }
1301  return $lines;
1302  }
1303 
1309  function _check ($from_lines, $to_lines) {
1310  $fname = 'Diff::_check';
1311  //wfProfileIn( $fname );
1312  if (serialize($from_lines) != serialize($this->orig()))
1313  trigger_error("Reconstructed original doesn't match", E_USER_ERROR);
1314  if (serialize($to_lines) != serialize($this->closing()))
1315  trigger_error("Reconstructed closing doesn't match", E_USER_ERROR);
1316 
1317  $rev = $this->reverse();
1318  if (serialize($to_lines) != serialize($rev->orig()))
1319  trigger_error("Reversed original doesn't match", E_USER_ERROR);
1320  if (serialize($from_lines) != serialize($rev->closing()))
1321  trigger_error("Reversed closing doesn't match", E_USER_ERROR);
1322 
1323 
1324  $prevtype = 'none';
1325  foreach ($this->edits as $edit) {
1326  if ( $prevtype == $edit->type )
1327  trigger_error("Edit sequence is non-optimal", E_USER_ERROR);
1328  $prevtype = $edit->type;
1329  }
1330 
1331  $lcs = $this->lcs();
1332  trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE);
1333  //wfProfileOut( $fname );
1334  }
1335 }
1336 
1342 class MappedDiff extends Diff
1343 {
1367  function MappedDiff($from_lines, $to_lines,
1368  $mapped_from_lines, $mapped_to_lines) {
1369  $fname = 'MappedDiff::MappedDiff';
1370  //wfProfileIn( $fname );
1371 
1372  assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1373  assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1374 
1375  $this->Diff($mapped_from_lines, $mapped_to_lines);
1376 
1377  $xi = $yi = 0;
1378  for ($i = 0; $i < sizeof($this->edits); $i++) {
1379  $orig = &$this->edits[$i]->orig;
1380  if (is_array($orig)) {
1381  $orig = array_slice($from_lines, $xi, sizeof($orig));
1382  $xi += sizeof($orig);
1383  }
1384 
1385  $closing = &$this->edits[$i]->closing;
1386  if (is_array($closing)) {
1387  $closing = array_slice($to_lines, $yi, sizeof($closing));
1388  $yi += sizeof($closing);
1389  }
1390  }
1391  //wfProfileOut( $fname );
1392  }
1393 }
1394 
1405 class DiffFormatter
1406 {
1413  var $leading_context_lines = 0;
1414 
1421  var $trailing_context_lines = 0;
1422 
1429  function format($diff) {
1430  $fname = 'DiffFormatter::format';
1431  //wfProfileIn( $fname );
1432 
1433  $xi = $yi = 1;
1434  $block = false;
1435  $context = array();
1436 
1437  $nlead = $this->leading_context_lines;
1438  $ntrail = $this->trailing_context_lines;
1439 
1440  $this->_start_diff();
1441 
1442  foreach ($diff->edits as $edit) {
1443  if ($edit->type == 'copy') {
1444  if (is_array($block)) {
1445  if (sizeof($edit->orig) <= $nlead + $ntrail) {
1446  $block[] = $edit;
1447  }
1448  else{
1449  if ($ntrail) {
1450  $context = array_slice($edit->orig, 0, $ntrail);
1451  $block[] = new _DiffOp_Copy($context);
1452  }
1453  $this->_block($x0, $ntrail + $xi - $x0,
1454  $y0, $ntrail + $yi - $y0,
1455  $block);
1456  $block = false;
1457  }
1458  }
1459  $context = $edit->orig;
1460  }
1461  else {
1462  if (! is_array($block)) {
1463  $context = array_slice($context, sizeof($context) - $nlead);
1464  $x0 = $xi - sizeof($context);
1465  $y0 = $yi - sizeof($context);
1466  $block = array();
1467  if ($context)
1468  $block[] = new _DiffOp_Copy($context);
1469  }
1470  $block[] = $edit;
1471  }
1472 
1473  if ($edit->orig)
1474  $xi += sizeof($edit->orig);
1475  if ($edit->closing)
1476  $yi += sizeof($edit->closing);
1477  }
1478 
1479  if (is_array($block))
1480  $this->_block($x0, $xi - $x0,
1481  $y0, $yi - $y0,
1482  $block);
1483 
1484  $end = $this->_end_diff();
1485  //wfProfileOut( $fname );
1486  return $end;
1487  }
1488 
1489  function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1490  $fname = 'DiffFormatter::_block';
1491  //wfProfileIn( $fname );
1492  $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1493  foreach ($edits as $edit) {
1494  if ($edit->type == 'copy')
1495  $this->_context($edit->orig);
1496  elseif ($edit->type == 'add')
1497  $this->_added($edit->closing);
1498  elseif ($edit->type == 'delete')
1499  $this->_deleted($edit->orig);
1500  elseif ($edit->type == 'change')
1501  $this->_changed($edit->orig, $edit->closing);
1502  else
1503  trigger_error('Unknown edit type', E_USER_ERROR);
1504  }
1505  $this->_end_block();
1506  //wfProfileOut( $fname );
1507  }
1508 
1509  function _start_diff() {
1510  ob_start();
1511  }
1512 
1513  function _end_diff() {
1514  $val = ob_get_contents();
1515  ob_end_clean();
1516  return $val;
1517  }
1518 
1519  function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1520  if ($xlen > 1)
1521  $xbeg .= "," . ($xbeg + $xlen - 1);
1522  if ($ylen > 1)
1523  $ybeg .= "," . ($ybeg + $ylen - 1);
1524 
1525  return $xbeg . ($xlen ? ($ylen ? 'c' : 'd') : 'a') . $ybeg;
1526  }
1527 
1528  function _start_block($header) {
1529  echo $header;
1530  }
1531 
1532  function _end_block() {
1533  }
1534 
1535  function _lines($lines, $prefix = ' ') {
1536  foreach ($lines as $line)
1537  echo "$prefix $line\n";
1538  }
1539 
1540  function _context($lines) {
1541  $this->_lines($lines);
1542  }
1543 
1544  function _added($lines) {
1545  $this->_lines($lines, '>');
1546  }
1547  function _deleted($lines) {
1548  $this->_lines($lines, '<');
1549  }
1550 
1551  function _changed($orig, $closing) {
1552  $this->_deleted($orig);
1553  echo "---\n";
1554  $this->_added($closing);
1555  }
1556 }
1557 
1558 
1564 define('NBSP', '&#160;'); // iso-8859-x non-breaking space.
1565 
1571 class _HWLDF_WordAccumulator {
1572  function _HWLDF_WordAccumulator () {
1573  $this->_lines = array();
1574  $this->_line = '';
1575  $this->_group = '';
1576  $this->_tag = '';
1577  }
1578 
1579  function _flushGroup ($new_tag) {
1580  if ($this->_group !== '') {
1581  if ($this->_tag == 'ins')
1582  $this->_line .= '[ilDiffInsStart]' .
1583  htmlspecialchars ( $this->_group ) . '[ilDiffInsEnd]';
1584  elseif ($this->_tag == 'del')
1585  $this->_line .= '[ilDiffDelStart]' .
1586  htmlspecialchars ( $this->_group ) . '[ilDiffDelEnd]';
1587  else
1588  $this->_line .= htmlspecialchars ( $this->_group );
1589  }
1590  $this->_group = '';
1591  $this->_tag = $new_tag;
1592  }
1593 
1594  function _flushLine ($new_tag) {
1595  $this->_flushGroup($new_tag);
1596  if ($this->_line != '')
1597  array_push ( $this->_lines, $this->_line );
1598  else
1599  # make empty lines visible by inserting an NBSP
1600  array_push ( $this->_lines, NBSP );
1601  $this->_line = '';
1602  }
1603 
1604  function addWords ($words, $tag = '') {
1605  if ($tag != $this->_tag)
1606  $this->_flushGroup($tag);
1607 
1608  foreach ($words as $word) {
1609  // new-line should only come as first char of word.
1610  if ($word == '')
1611  continue;
1612  if ($word[0] == "\n") {
1613  $this->_flushLine($tag);
1614  $word = substr($word, 1);
1615  }
1616  assert(!strstr($word, "\n"));
1617  $this->_group .= $word;
1618  }
1619  }
1620 
1621  function getLines() {
1622  $this->_flushLine('~done');
1623  return $this->_lines;
1624  }
1625 }
1626 
1632 class WordLevelDiff extends MappedDiff
1633 {
1634  const MAX_LINE_LENGTH = 10000;
1635 
1636  function WordLevelDiff ($orig_lines, $closing_lines) {
1637  $fname = 'WordLevelDiff::WordLevelDiff';
1638  //wfProfileIn( $fname );
1639 
1640  list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1641  list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1642 
1643  $this->MappedDiff($orig_words, $closing_words,
1644  $orig_stripped, $closing_stripped);
1645  //wfProfileOut( $fname );
1646  }
1647 
1648  function _split($lines) {
1649  $fname = 'WordLevelDiff::_split';
1650  //wfProfileIn( $fname );
1651 
1652  $words = array();
1653  $stripped = array();
1654  $first = true;
1655  foreach ( $lines as $line ) {
1656  # If the line is too long, just pretend the entire line is one big word
1657  # This prevents resource exhaustion problems
1658  if ( $first ) {
1659  $first = false;
1660  } else {
1661  $words[] = "\n";
1662  $stripped[] = "\n";
1663  }
1664  if ( strlen( $line ) > self::MAX_LINE_LENGTH ) {
1665  $words[] = $line;
1666  $stripped[] = $line;
1667  } else {
1668  $m = array();
1669  if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1670  $line, $m))
1671  {
1672  $words = array_merge( $words, $m[0] );
1673  $stripped = array_merge( $stripped, $m[1] );
1674  }
1675  }
1676  }
1677  //wfProfileOut( $fname );
1678  return array($words, $stripped);
1679  }
1680 
1681  function orig () {
1682  $fname = 'WordLevelDiff::orig';
1683  //wfProfileIn( $fname );
1684  $orig = new _HWLDF_WordAccumulator;
1685 
1686  foreach ($this->edits as $edit) {
1687  if ($edit->type == 'copy')
1688  $orig->addWords($edit->orig);
1689  elseif ($edit->orig)
1690  $orig->addWords($edit->orig, 'del');
1691  }
1692  $lines = $orig->getLines();
1693  //wfProfileOut( $fname );
1694  return $lines;
1695  }
1696 
1697  function closing () {
1698  $fname = 'WordLevelDiff::closing';
1699  //wfProfileIn( $fname );
1700  $closing = new _HWLDF_WordAccumulator;
1701 
1702  foreach ($this->edits as $edit) {
1703  if ($edit->type == 'copy')
1704  $closing->addWords($edit->closing);
1705  elseif ($edit->closing)
1706  $closing->addWords($edit->closing, 'ins');
1707  }
1708  $lines = $closing->getLines();
1709  //wfProfileOut( $fname );
1710  return $lines;
1711  }
1712 }
1713 
1720 class TableDiffFormatter extends DiffFormatter
1721 {
1722  function TableDiffFormatter() {
1723  $this->leading_context_lines = 2;
1724  $this->trailing_context_lines = 2;
1725  }
1726 
1727  function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1728  $r = '<tr><td colspan="2" align="left"><strong><!--LINE '.$xbeg."--></strong></td>\n" .
1729  '<td colspan="2" align="left"><strong><!--LINE '.$ybeg."--></strong></td></tr>\n";
1730  return $r;
1731  }
1732 
1733  function _start_block( $header ) {
1734  echo $header;
1735  }
1736 
1737  function _end_block() {
1738  }
1739 
1740  function _lines( $lines, $prefix=' ', $color='white' ) {
1741  }
1742 
1743  # HTML-escape parameter before calling this
1744  function addedLine( $line ) {
1745  return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1746  }
1747 
1748  # HTML-escape parameter before calling this
1749  function deletedLine( $line ) {
1750  return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1751  }
1752 
1753  # HTML-escape parameter before calling this
1754  function contextLine( $line ) {
1755  return "<td> </td><td class='diff-context'>{$line}</td>";
1756  }
1757 
1758  function emptyLine() {
1759  return '<td colspan="2">&nbsp;</td>';
1760  }
1761 
1762  function _added( $lines ) {
1763  foreach ($lines as $line) {
1764  echo '<tr>' . $this->emptyLine() .
1765  $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1766  }
1767  }
1768 
1769  function _deleted($lines) {
1770  foreach ($lines as $line) {
1771  echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1772  $this->emptyLine() . "</tr>\n";
1773  }
1774  }
1775 
1776  function _context( $lines ) {
1777  foreach ($lines as $line) {
1778  echo '<tr>' .
1779  $this->contextLine( htmlspecialchars ( $line ) ) .
1780  $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1781  }
1782  }
1783 
1784  function _changed( $orig, $closing ) {
1785  $fname = 'TableDiffFormatter::_changed';
1786  //wfProfileIn( $fname );
1787 
1788  $diff = new WordLevelDiff( $orig, $closing );
1789  $del = $diff->orig();
1790  $add = $diff->closing();
1791 
1792  # Notice that WordLevelDiff returns HTML-escaped output.
1793  # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1794 
1795  while ( $line = array_shift( $del ) ) {
1796  $aline = array_shift( $add );
1797  echo '<tr>' . $this->deletedLine( $line ) .
1798  $this->addedLine( $aline ) . "</tr>\n";
1799  }
1800  foreach ($add as $line) { # If any leftovers
1801  echo '<tr>' . $this->emptyLine() .
1802  $this->addedLine( $line ) . "</tr>\n";
1803  }
1804  //wfProfileOut( $fname );
1805  }
1806 }
1807 
1808 ?>