5 * @subpackage DifferenceEngine
9 require_once( 'Revision.php' );
11 define( 'MAX_DIFF_LINE', 10000 );
12 define( 'MAX_DIFF_XREF_LENGTH', 10000 );
18 * @subpackage DifferenceEngine
20 class DifferenceEngine
{
24 var $mOldid, $mNewid, $mTitle;
25 var $mOldtitle, $mNewtitle, $mPagetitle;
26 var $mOldtext, $mNewtext;
27 var $mOldUser, $mNewUser;
28 var $mOldComment, $mNewComment;
29 var $mOldPage, $mNewPage;
30 var $mRcidMarkPatrolled;
31 var $mOldRev, $mNewRev;
32 var $mRevisionsLoaded = false; // Have the revisions been loaded
33 var $mTextLoaded = 0; // How many text blobs have been loaded, 0, 1 or 2?
38 * @param Title $titleObj Title object that the diff is associated with
39 * @param integer $old Old ID we want to show and diff with.
40 * @param string $new Either 'prev' or 'next'.
41 * @param integer $rcid ??? (default 0)
43 function DifferenceEngine( $titleObj = null, $old = 0, $new = 0, $rcid = 0 ) {
44 $this->mTitle
= $titleObj;
45 wfDebug("DifferenceEngine old '$old' new '$new' rcid '$rcid'\n");
47 if ( 'prev' == $new ) {
48 # Show diff between revision $old and the previous one.
49 # Get previous one from DB.
51 $this->mNewid
= intval($old);
53 $this->mOldid
= $this->mTitle
->getPreviousRevisionID( $this->mNewid
);
55 } elseif ( 'next' == $new ) {
56 # Show diff between revision $old and the previous one.
57 # Get previous one from DB.
59 $this->mOldid
= intval($old);
60 $this->mNewid
= $this->mTitle
->getNextRevisionID( $this->mOldid
);
61 if ( false === $this->mNewid
) {
62 # if no result, NewId points to the newest old revision. The only newer
63 # revision is cur, which is "0".
68 $this->mOldid
= intval($old);
69 $this->mNewid
= intval($new);
71 $this->mRcidMarkPatrolled
= intval($rcid); # force it to be an integer
74 function showDiffPage() {
75 global $wgUser, $wgOut, $wgContLang, $wgOnlySysopsCanPatrol,
76 $wgUseExternalEditor, $wgUseRCPatrol;
77 $fname = 'DifferenceEngine::showDiffPage';
78 wfProfileIn( $fname );
80 # If external diffs are enabled both globally and for the user,
81 # we'll use the application/x-external-editor interface to call
82 # an external diff tool like kompare, kdiff3, etc.
83 if($wgUseExternalEditor && $wgUser->getOption('externaldiff')) {
84 global $wgInputEncoding,$wgServer,$wgScript,$wgLang;
86 header ( "Content-type: application/x-external-editor; charset=".$wgInputEncoding );
87 $url1=$this->mTitle
->getFullURL("action=raw&oldid=".$this->mOldid
);
88 $url2=$this->mTitle
->getFullURL("action=raw&oldid=".$this->mNewid
);
89 $special=$wgLang->getNsText(NS_SPECIAL
);
94 Script={$wgServer}{$wgScript}
95 Special namespace={$special}
109 $t = $this->mTitle
->getPrefixedText() . " (Diff: {$this->mOldid}, " .
111 $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
113 $wgOut->setArticleFlag( false );
114 if ( ! $this->loadRevisionData() ) {
115 $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
116 $wgOut->addWikitext( $mtext );
117 wfProfileOut( $fname );
120 if ( $this->mNewRev
->isCurrent() ) {
121 $wgOut->setArticleFlag( true );
124 # mOldid is false if the difference engine is called with a "vague" query for
125 # a diff between a version V and its previous version V' AND the version V
126 # is the first version of that article. In that case, V' does not exist.
127 if ( $this->mOldid
=== false ) {
128 $this->showFirstRevision();
129 wfProfileOut( $fname );
133 $wgOut->suppressQuickbar();
135 $oldTitle = $this->mOldPage
->getPrefixedText();
136 $newTitle = $this->mNewPage
->getPrefixedText();
137 if( $oldTitle == $newTitle ) {
138 $wgOut->setPageTitle( $newTitle );
140 $wgOut->setPageTitle( $oldTitle . ', ' . $newTitle );
142 $wgOut->setSubtitle( wfMsg( 'difference' ) );
143 $wgOut->setRobotpolicy( 'noindex,follow' );
145 if ( !( $this->mOldPage
->userCanRead() && $this->mNewPage
->userCanRead() ) ) {
146 $wgOut->loginToUse();
148 wfProfileOut( $fname );
152 $sk = $wgUser->getSkin();
153 $talk = $wgContLang->getNsText( NS_TALK
);
154 $contribs = wfMsg( 'contribslink' );
156 $this->mOldComment
= $sk->formatComment($this->mOldComment
);
157 $this->mNewComment
= $sk->formatComment($this->mNewComment
);
159 $oldUserLink = $sk->makeLinkObj( Title
::makeTitleSafe( NS_USER
, $this->mOldUser
), $this->mOldUser
);
160 $newUserLink = $sk->makeLinkObj( Title
::makeTitleSafe( NS_USER
, $this->mNewUser
), $this->mNewUser
);
161 $oldUTLink = $sk->makeLinkObj( Title
::makeTitleSafe( NS_USER_TALK
, $this->mOldUser
), $talk );
162 $newUTLink = $sk->makeLinkObj( Title
::makeTitleSafe( NS_USER_TALK
, $this->mNewUser
), $talk );
163 $oldContribs = $sk->makeKnownLinkObj( Title
::makeTitle( NS_SPECIAL
, 'Contributions' ), $contribs,
164 'target=' . urlencode($this->mOldUser
) );
165 $newContribs = $sk->makeKnownLinkObj( Title
::makeTitle( NS_SPECIAL
, 'Contributions' ), $contribs,
166 'target=' . urlencode($this->mNewUser
) );
167 if ( $this->mNewRev
->isCurrent() && $wgUser->isAllowed('rollback') ) {
168 $rollback = ' <strong>[' . $sk->makeKnownLinkObj( $this->mTitle
, wfMsg( 'rollbacklink' ),
169 'action=rollback&from=' . urlencode($this->mNewUser
) .
170 '&token=' . urlencode( $wgUser->editToken( array( $this->mTitle
->getPrefixedText(), $this->mNewUser
) ) ) ) .
175 if ( $wgUseRCPatrol && $this->mRcidMarkPatrolled
!= 0 && $wgUser->isLoggedIn() &&
176 ( $wgUser->isAllowed('rollback') ||
!$wgOnlySysopsCanPatrol ) )
178 $patrol = ' [' . $sk->makeKnownLinkObj( $this->mTitle
, wfMsg( 'markaspatrolleddiff' ),
179 "action=markpatrolled&rcid={$this->mRcidMarkPatrolled}" ) . ']';
184 $prevlink = $sk->makeKnownLinkObj( $this->mTitle
, wfMsgHtml( 'previousdiff' ),
185 'diff=prev&oldid='.$this->mOldid
, '', '', 'id="differences-prevlink"' );
186 if ( $this->mNewRev
->isCurrent() ) {
189 $nextlink = $sk->makeKnownLinkObj( $this->mTitle
, wfMsgHtml( 'nextdiff' ),
190 'diff=next&oldid='.$this->mNewid
, '', '', 'id="differences-nextlink"' );
193 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink " .
194 "($oldUTLink | $oldContribs)<br />" . $this->mOldComment
.
195 '<br />' . $prevlink;
196 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink " .
197 "($newUTLink | $newContribs) $rollback<br />" . $this->mNewComment
.
198 '<br />' . $nextlink . $patrol;
200 $this->showDiff( $oldHeader, $newHeader );
201 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
203 if( !$this->mNewRev
->isCurrent() ) {
204 $oldEditSectionSetting = $wgOut->mParserOptions
->setEditSection( false );
207 $this->loadNewText();
208 if( is_object( $this->mNewRev
) ) {
209 $wgOut->setRevisionId( $this->mNewRev
->getId() );
211 $wgOut->addSecondaryWikiText( $this->mNewtext
);
213 if( !$this->mNewRev
->isCurrent() ) {
214 $wgOut->mParserOptions
->setEditSection( $oldEditSectionSetting );
217 wfProfileOut( $fname );
221 * Show the first revision of an article. Uses normal diff headers in
222 * contrast to normal "old revision" display style.
224 function showFirstRevision() {
225 global $wgOut, $wgUser, $wgLang;
227 $fname = 'DifferenceEngine::showFirstRevision';
228 wfProfileIn( $fname );
230 # Get article text from the DB
232 if ( ! $this->loadNewText() ) {
233 $t = $this->mTitle
->getPrefixedText() . " (Diff: {$this->mOldid}, " .
235 $mtext = wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" );
236 $wgOut->setPagetitle( wfMsg( 'errorpagetitle' ) );
237 $wgOut->addWikitext( $mtext );
238 wfProfileOut( $fname );
241 if ( $this->mNewRev
->isCurrent() ) {
242 $wgOut->setArticleFlag( true );
245 # Check if user is allowed to look at this page. If not, bail out.
247 if ( !( $this->mTitle
->userCanRead() ) ) {
248 $wgOut->loginToUse();
250 wfProfileOut( $fname );
254 # Prepare the header box
256 $sk = $wgUser->getSkin();
258 $uTLink = $sk->makeLinkObj( Title
::makeTitleSafe( NS_USER_TALK
, $this->mOldUser
), $wgLang->getNsText( NS_TALK
) );
259 $userLink = $sk->makeLinkObj( Title
::makeTitleSafe( NS_USER
, $this->mOldUser
), $this->mOldUser
);
260 $contribs = $sk->makeKnownLinkObj( Title
::makeTitle( NS_SPECIAL
, 'Contributions' ), wfMsg( 'contribslink' ),
261 'target=' . urlencode($this->mOldUser
) );
262 $nextlink = $sk->makeKnownLinkObj( $this->mTitle
, wfMsgHtml( 'nextdiff' ), 'diff=next&oldid='.$this->mNewid
, '', '', 'id="differences-nextlink"' );
263 $header = "<div class=\"firstrevisionheader\" style=\"text-align: center\"><strong>{$this->mOldtitle}</strong><br />$userLink " .
264 "($uTLink | $contribs)<br />" . $this->mOldComment
.
265 '<br />' . $nextlink. "</div>\n";
267 $wgOut->addHTML( $header );
269 $wgOut->setSubtitle( wfMsg( 'difference' ) );
270 $wgOut->setRobotpolicy( 'noindex,follow' );
273 # Show current revision
275 $wgOut->addHTML( "<hr /><h2>{$this->mPagetitle}</h2>\n" );
276 if( is_object( $this->mNewRev
) ) {
277 $wgOut->setRevisionId( $this->mNewRev
->getId() );
279 $wgOut->addSecondaryWikiText( $this->mNewtext
);
281 wfProfileOut( $fname );
285 * Get the diff text, send it to $wgOut
286 * Returns false if the diff could not be generated, otherwise returns true
288 function showDiff( $otitle, $ntitle ) {
290 $diff = $this->getDiff( $otitle, $ntitle );
291 if ( $diff === false ) {
292 $wgOut->addWikitext( wfMsg( 'missingarticle', "<nowiki>$t</nowiki>" ) );
295 $wgOut->addHTML( $diff );
301 * Get diff table, including header
302 * Note that the interface has changed, it's no longer static.
303 * Returns false on error
305 function getDiff( $otitle, $ntitle ) {
306 $body = $this->getDiffBody();
307 if ( $body === false ) {
310 return $this->addHeader( $body, $otitle, $ntitle );
315 * Get the diff table body, without header
316 * Returns false on error
318 function getDiffBody() {
319 global $wgUseExternalDiffEngine, $wgContLang, $wgMemc, $wgDBname;
323 if ( $this->mOldid
&& $this->mNewid
) {
325 $key = "$wgDBname:diff:oldid:{$this->mOldid}:newid:{$this->mNewid}";
326 $difftext = $wgMemc->get( $key );
328 wfIncrStats( 'diff_cache_hit' );
329 $difftext .= "\n<!-- diff cache key $key -->\n";
334 if ( !$this->loadText() ) {
338 $otext = $wgContLang->segmentForDiff($this->mOldtext
);
339 $ntext = $wgContLang->segmentForDiff($this->mNewtext
);
340 if ( $wgUseExternalDiffEngine ) {
341 # For historical reasons, external diff engine expects
342 # input text to be HTML-escaped already
343 $otext = str_replace( "\r\n", "\n", htmlspecialchars ( $otext ) );
344 $ntext = str_replace( "\r\n", "\n", htmlspecialchars ( $ntext ) );
345 if( !function_exists( 'wikidiff_do_diff' ) ) {
346 dl('php_wikidiff.so');
348 $difftext = wikidiff_do_diff( $otext, $ntext, 2 );
350 $ota = explode( "\n", str_replace( "\r\n", "\n", $otext ) );
351 $nta = explode( "\n", str_replace( "\r\n", "\n", $ntext ) );
352 $diffs =& new Diff( $ota, $nta );
353 $formatter =& new TableDiffFormatter();
354 $difftext = $formatter->format( $diffs );
356 $difftext = $wgContLang->unsegmentForDiff($difftext);
358 // Save to cache for 7 days
359 if ( $key !== false ) {
360 wfIncrStats( 'diff_cache_miss' );
361 $wgMemc->set( $key, $difftext, 7*86400 );
363 wfIncrStats( 'diff_uncacheable' );
369 * Add the header to a diff body
371 function addHeader( $diff, $otitle, $ntitle ) {
373 <table border='0' width='98%' cellpadding='0' cellspacing='4' class='diff'>
375 <td colspan='2' width='50%' align='center' class='diff-otitle'>{$otitle}</td>
376 <td colspan='2' width='50%' align='center' class='diff-ntitle'>{$ntitle}</td>
385 * Use specified text instead of loading from the database
387 function setText( $oldText, $newText ) {
388 $this->mOldtext
= $oldText;
389 $this->mNewtext
= $newText;
390 $this->mTextLoaded
= 2;
394 * Load revision metadata for the specified articles. If newid is 0, then compare
395 * the old article in oldid to the current article; if oldid is 0, then
396 * compare the current article to the immediately previous one (ignoring the
399 * If oldid is false, leave the corresponding revision object set
400 * to false. This is impossible via ordinary user input, and is provided for
403 function loadRevisionData() {
405 if ( $this->mRevisionsLoaded
) {
408 // Whether it succeeds or fails, we don't want to try again
409 $this->mRevisionsLoaded
= true;
412 // Load the new revision object
413 if( $this->mNewid
) {
414 $this->mNewRev
= Revision
::newFromId( $this->mNewid
);
416 $this->mNewRev
= Revision
::newFromTitle( $this->mTitle
);
419 if( is_null( $this->mNewRev
) ) {
423 // Set assorted variables
424 if( $this->mNewRev
->isCurrent() ) {
425 $this->mPagetitle
= htmlspecialchars( wfMsg( 'currentrev' ) );
426 $this->mNewPage
= $this->mTitle
;
427 $newLink = $this->mNewPage
->escapeLocalUrl();
428 $this->mNewtitle
= "<a href='$newLink'>{$this->mPagetitle}</a>";
430 $this->mNewPage
= $this->mNewRev
->getTitle();
431 $newLink = $this->mNewPage
->escapeLocalUrl ('oldid=' . $this->mNewid
);
432 $t = $wgLang->timeanddate( $this->mNewRev
->getTimestamp(), true );
433 $this->mPagetitle
= htmlspecialchars( wfMsg( 'revisionasof', $t ) );
434 $this->mNewtitle
= "<a href='$newLink'>{$this->mPagetitle}</a>";
437 $this->mNewUser
= $this->mNewRev
->getUserText();
438 $this->mNewComment
= $this->mNewRev
->getComment();
440 // Load the old revision object
441 $this->mOldRev
= false;
442 if( $this->mOldid
) {
443 $this->mOldRev
= Revision
::newFromId( $this->mOldid
);
444 } elseif ( $this->mOldid
=== 0 ) {
445 $rev = $this->mNewRev
->getPrevious();
447 $this->mOldid
= $rev->getId();
448 $this->mOldRev
= $rev;
450 // No previous revision; mark to show as first-version only.
451 $this->mOldid
= false;
452 $this->mOldRev
= false;
454 }/* elseif ( $this->mOldid === false ) leave mOldRev false; */
456 if( is_null( $this->mOldRev
) ) {
460 if ( $this->mOldRev
) {
461 $this->mOldPage
= $this->mOldRev
->getTitle();
463 $t = $wgLang->timeanddate( $this->mOldRev
->getTimestamp(), true );
464 $oldLink = $this->mOldPage
->escapeLocalUrl( 'oldid=' . $this->mOldid
);
465 $this->mOldtitle
= "<a href='$oldLink'>" . htmlspecialchars( wfMsg( 'revisionasof', $t ) ) . '</a>';
468 $this->mOldUser
= $this->mOldRev
->getUserText();
469 $this->mOldComment
= $this->mOldRev
->getComment();
476 * Load the text of the revisions, as well as revision data.
478 function loadText() {
479 if ( $this->mTextLoaded
== 2 ) {
482 // Whether it succeeds or fails, we don't want to try again
483 $this->mTextLoaded
= 2;
486 if ( !$this->loadRevisionData() ) {
489 if ( $this->mOldRev
) {
490 $this->mOldtext
= $this->mOldRev
->getText();
491 if ( $this->mOldtext
=== false ) {
495 if ( $this->mNewRev
) {
496 $this->mNewtext
= $this->mNewRev
->getText();
497 if ( $this->mNewtext
=== false ) {
505 * Load the text of the new revision, not the old one
507 function loadNewText() {
508 if ( $this->mTextLoaded
>= 1 ) {
511 $this->mTextLoaded
= 1;
513 if ( !$this->loadRevisionData() ) {
516 $this->mNewtext
= $this->mNewRev
->getText();
523 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
525 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
526 // You may copy this code freely under the conditions of the GPL.
529 define('USE_ASSERTS', function_exists('assert'));
535 * @subpackage DifferenceEngine
543 trigger_error('pure virtual', E_USER_ERROR
);
547 return $this->orig ?
sizeof($this->orig
) : 0;
550 function nclosing() {
551 return $this->closing ?
sizeof($this->closing
) : 0;
559 * @subpackage DifferenceEngine
561 class _DiffOp_Copy
extends _DiffOp
{
564 function _DiffOp_Copy ($orig, $closing = false) {
565 if (!is_array($closing))
568 $this->closing
= $closing;
572 return new _DiffOp_Copy($this->closing
, $this->orig
);
580 * @subpackage DifferenceEngine
582 class _DiffOp_Delete
extends _DiffOp
{
583 var $type = 'delete';
585 function _DiffOp_Delete ($lines) {
586 $this->orig
= $lines;
587 $this->closing
= false;
591 return new _DiffOp_Add($this->orig
);
599 * @subpackage DifferenceEngine
601 class _DiffOp_Add
extends _DiffOp
{
604 function _DiffOp_Add ($lines) {
605 $this->closing
= $lines;
610 return new _DiffOp_Delete($this->closing
);
618 * @subpackage DifferenceEngine
620 class _DiffOp_Change
extends _DiffOp
{
621 var $type = 'change';
623 function _DiffOp_Change ($orig, $closing) {
625 $this->closing
= $closing;
629 return new _DiffOp_Change($this->closing
, $this->orig
);
635 * Class used internally by Diff to actually compute the diffs.
637 * The algorithm used here is mostly lifted from the perl module
638 * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
639 * http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
641 * More ideas are taken from:
642 * http://www.ics.uci.edu/~eppstein/161/960229.html
644 * Some ideas are (and a bit of code) are from from analyze.c, from GNU
645 * diffutils-2.7, which can be found at:
646 * ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
648 * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
651 * Line length limits for robustness added by Tim Starling, 2005-08-31
653 * @author Geoffrey T. Dairiki, Tim Starling
656 * @subpackage DifferenceEngine
660 function diff ($from_lines, $to_lines) {
661 $fname = '_DiffEngine::diff';
662 wfProfileIn( $fname );
664 $n_from = sizeof($from_lines);
665 $n_to = sizeof($to_lines);
667 $this->xchanged
= $this->ychanged
= array();
668 $this->xv
= $this->yv
= array();
669 $this->xind
= $this->yind
= array();
671 unset($this->in_seq
);
674 // Skip leading common lines.
675 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++
) {
676 if ($from_lines[$skip] !== $to_lines[$skip])
678 $this->xchanged
[$skip] = $this->ychanged
[$skip] = false;
680 // Skip trailing common lines.
681 $xi = $n_from; $yi = $n_to;
682 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++
) {
683 if ($from_lines[$xi] !== $to_lines[$yi])
685 $this->xchanged
[$xi] = $this->ychanged
[$yi] = false;
688 // Ignore lines which do not exist in both files.
689 for ($xi = $skip; $xi < $n_from - $endskip; $xi++
) {
690 $xhash[$this->_line_hash($from_lines[$xi])] = 1;
693 for ($yi = $skip; $yi < $n_to - $endskip; $yi++
) {
694 $line = $to_lines[$yi];
695 if ( ($this->ychanged
[$yi] = empty($xhash[$this->_line_hash($line)])) )
697 $yhash[$this->_line_hash($line)] = 1;
701 for ($xi = $skip; $xi < $n_from - $endskip; $xi++
) {
702 $line = $from_lines[$xi];
703 if ( ($this->xchanged
[$xi] = empty($yhash[$this->_line_hash($line)])) )
710 $this->_compareseq(0, sizeof($this->xv
), 0, sizeof($this->yv
));
712 // Merge edits when possible
713 $this->_shift_boundaries($from_lines, $this->xchanged
, $this->ychanged
);
714 $this->_shift_boundaries($to_lines, $this->ychanged
, $this->xchanged
);
716 // Compute the edit operations.
719 while ($xi < $n_from ||
$yi < $n_to) {
720 USE_ASSERTS
&& assert($yi < $n_to ||
$this->xchanged
[$xi]);
721 USE_ASSERTS
&& assert($xi < $n_from ||
$this->ychanged
[$yi]);
723 // Skip matching "snake".
725 while ( $xi < $n_from && $yi < $n_to
726 && !$this->xchanged
[$xi] && !$this->ychanged
[$yi]) {
727 $copy[] = $from_lines[$xi++
];
731 $edits[] = new _DiffOp_Copy($copy);
733 // Find deletes & adds.
735 while ($xi < $n_from && $this->xchanged
[$xi])
736 $delete[] = $from_lines[$xi++
];
739 while ($yi < $n_to && $this->ychanged
[$yi])
740 $add[] = $to_lines[$yi++
];
743 $edits[] = new _DiffOp_Change($delete, $add);
745 $edits[] = new _DiffOp_Delete($delete);
747 $edits[] = new _DiffOp_Add($add);
749 wfProfileOut( $fname );
754 * Returns the whole line if it's small enough, or the MD5 hash otherwise
756 function _line_hash( $line ) {
757 if ( strlen( $line ) > MAX_DIFF_XREF_LENGTH
) {
765 /* Divide the Largest Common Subsequence (LCS) of the sequences
766 * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
769 * Returns (LCS, PTS). LCS is the length of the LCS. PTS is an
770 * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
771 * sub sequences. The first sub-sequence is contained in [X0, X1),
772 * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on. Note
773 * that (X0, Y0) == (XOFF, YOFF) and
774 * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
776 * This function assumes that the first lines of the specified portions
777 * of the two files do not match, and likewise that the last lines do not
778 * match. The caller must trim matching lines from the beginning and end
779 * of the portions it is going to specify.
781 function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
782 $fname = '_DiffEngine::_diag';
783 wfProfileIn( $fname );
786 if ($xlim - $xoff > $ylim - $yoff) {
787 // Things seems faster (I'm not sure I understand why)
788 // when the shortest sequence in X.
790 list ($xoff, $xlim, $yoff, $ylim)
791 = array( $yoff, $ylim, $xoff, $xlim);
795 for ($i = $ylim - 1; $i >= $yoff; $i--)
796 $ymatches[$this->xv
[$i]][] = $i;
798 for ($i = $ylim - 1; $i >= $yoff; $i--)
799 $ymatches[$this->yv
[$i]][] = $i;
802 $this->seq
[0]= $yoff - 1;
803 $this->in_seq
= array();
806 $numer = $xlim - $xoff +
$nchunks - 1;
808 for ($chunk = 0; $chunk < $nchunks; $chunk++
) {
809 wfProfileIn( "$fname-chunk" );
811 for ($i = 0; $i <= $this->lcs
; $i++
)
812 $ymids[$i][$chunk-1] = $this->seq
[$i];
814 $x1 = $xoff +
(int)(($numer +
($xlim-$xoff)*$chunk) / $nchunks);
815 for ( ; $x < $x1; $x++
) {
816 $line = $flip ?
$this->yv
[$x] : $this->xv
[$x];
817 if (empty($ymatches[$line]))
819 $matches = $ymatches[$line];
821 while (list ($junk, $y) = each($matches))
822 if (empty($this->in_seq
[$y])) {
823 $k = $this->_lcs_pos($y);
824 USE_ASSERTS
&& assert($k > 0);
825 $ymids[$k] = $ymids[$k-1];
828 while (list ($junk, $y) = each($matches)) {
829 if ($y > $this->seq
[$k-1]) {
830 USE_ASSERTS
&& assert($y < $this->seq
[$k]);
831 // Optimization: this is a common case:
832 // next match is just replacing previous match.
833 $this->in_seq
[$this->seq
[$k]] = false;
835 $this->in_seq
[$y] = 1;
836 } else if (empty($this->in_seq
[$y])) {
837 $k = $this->_lcs_pos($y);
838 USE_ASSERTS
&& assert($k > 0);
839 $ymids[$k] = $ymids[$k-1];
843 wfProfileOut( "$fname-chunk" );
846 $seps[] = $flip ?
array($yoff, $xoff) : array($xoff, $yoff);
847 $ymid = $ymids[$this->lcs
];
848 for ($n = 0; $n < $nchunks - 1; $n++
) {
849 $x1 = $xoff +
(int)(($numer +
($xlim - $xoff) * $n) / $nchunks);
851 $seps[] = $flip ?
array($y1, $x1) : array($x1, $y1);
853 $seps[] = $flip ?
array($ylim, $xlim) : array($xlim, $ylim);
855 wfProfileOut( $fname );
856 return array($this->lcs
, $seps);
859 function _lcs_pos ($ypos) {
860 $fname = '_DiffEngine::_lcs_pos';
861 wfProfileIn( $fname );
864 if ($end == 0 ||
$ypos > $this->seq
[$end]) {
865 $this->seq
[++
$this->lcs
] = $ypos;
866 $this->in_seq
[$ypos] = 1;
867 wfProfileOut( $fname );
872 while ($beg < $end) {
873 $mid = (int)(($beg +
$end) / 2);
874 if ( $ypos > $this->seq
[$mid] )
880 USE_ASSERTS
&& assert($ypos != $this->seq
[$end]);
882 $this->in_seq
[$this->seq
[$end]] = false;
883 $this->seq
[$end] = $ypos;
884 $this->in_seq
[$ypos] = 1;
885 wfProfileOut( $fname );
889 /* Find LCS of two sequences.
891 * The results are recorded in the vectors $this->{x,y}changed[], by
892 * storing a 1 in the element for each line that is an insertion
893 * or deletion (ie. is not in the LCS).
895 * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
897 * Note that XLIM, YLIM are exclusive bounds.
898 * All line numbers are origin-0 and discarded lines are not counted.
900 function _compareseq ($xoff, $xlim, $yoff, $ylim) {
901 $fname = '_DiffEngine::_compareseq';
902 wfProfileIn( $fname );
904 // Slide down the bottom initial diagonal.
905 while ($xoff < $xlim && $yoff < $ylim
906 && $this->xv
[$xoff] == $this->yv
[$yoff]) {
911 // Slide up the top initial diagonal.
912 while ($xlim > $xoff && $ylim > $yoff
913 && $this->xv
[$xlim - 1] == $this->yv
[$ylim - 1]) {
918 if ($xoff == $xlim ||
$yoff == $ylim)
921 // This is ad hoc but seems to work well.
922 //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
923 //$nchunks = max(2,min(8,(int)$nchunks));
924 $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) +
1;
926 = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
930 // X and Y sequences have no common subsequence:
932 while ($yoff < $ylim)
933 $this->ychanged
[$this->yind
[$yoff++
]] = 1;
934 while ($xoff < $xlim)
935 $this->xchanged
[$this->xind
[$xoff++
]] = 1;
937 // Use the partitions to split this problem into subproblems.
940 while ($pt2 = next($seps)) {
941 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
945 wfProfileOut( $fname );
948 /* Adjust inserts/deletes of identical lines to join changes
949 * as much as possible.
951 * We do something when a run of changed lines include a
952 * line at one end and has an excluded, identical line at the other.
953 * We are free to choose which identical line is included.
954 * `compareseq' usually chooses the one at the beginning,
955 * but usually it is cleaner to consider the following identical line
956 * to be the "change".
958 * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
960 function _shift_boundaries ($lines, &$changed, $other_changed) {
961 $fname = '_DiffEngine::_shift_boundaries';
962 wfProfileIn( $fname );
966 USE_ASSERTS
&& assert('sizeof($lines) == sizeof($changed)');
967 $len = sizeof($lines);
968 $other_len = sizeof($other_changed);
972 * Scan forwards to find beginning of another run of changes.
973 * Also keep track of the corresponding point in the other file.
975 * Throughout this code, $i and $j are adjusted together so that
976 * the first $i elements of $changed and the first $j elements
977 * of $other_changed both contain the same number of zeros
979 * Furthermore, $j is always kept so that $j == $other_len or
980 * $other_changed[$j] == false.
982 while ($j < $other_len && $other_changed[$j])
985 while ($i < $len && ! $changed[$i]) {
986 USE_ASSERTS
&& assert('$j < $other_len && ! $other_changed[$j]');
988 while ($j < $other_len && $other_changed[$j])
997 // Find the end of this run of changes.
998 while (++
$i < $len && $changed[$i])
1003 * Record the length of this run of changes, so that
1004 * we can later determine whether the run has grown.
1006 $runlength = $i - $start;
1009 * Move the changed region back, so long as the
1010 * previous unchanged line matches the last changed one.
1011 * This merges with previous changed regions.
1013 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
1014 $changed[--$start] = 1;
1015 $changed[--$i] = false;
1016 while ($start > 0 && $changed[$start - 1])
1018 USE_ASSERTS
&& assert('$j > 0');
1019 while ($other_changed[--$j])
1021 USE_ASSERTS
&& assert('$j >= 0 && !$other_changed[$j]');
1025 * Set CORRESPONDING to the end of the changed run, at the last
1026 * point where it corresponds to a changed run in the other file.
1027 * CORRESPONDING == LEN means no such point has been found.
1029 $corresponding = $j < $other_len ?
$i : $len;
1032 * Move the changed region forward, so long as the
1033 * first changed line matches the following unchanged one.
1034 * This merges with following changed regions.
1035 * Do this second, so that if there are no merges,
1036 * the changed region is moved forward as far as possible.
1038 while ($i < $len && $lines[$start] == $lines[$i]) {
1039 $changed[$start++
] = false;
1041 while ($i < $len && $changed[$i])
1044 USE_ASSERTS
&& assert('$j < $other_len && ! $other_changed[$j]');
1046 if ($j < $other_len && $other_changed[$j]) {
1047 $corresponding = $i;
1048 while ($j < $other_len && $other_changed[$j])
1052 } while ($runlength != $i - $start);
1055 * If possible, move the fully-merged run of changes
1056 * back to a corresponding run in the other file.
1058 while ($corresponding < $i) {
1059 $changed[--$start] = 1;
1061 USE_ASSERTS
&& assert('$j > 0');
1062 while ($other_changed[--$j])
1064 USE_ASSERTS
&& assert('$j >= 0 && !$other_changed[$j]');
1067 wfProfileOut( $fname );
1072 * Class representing a 'diff' between two sequences of strings.
1075 * @package MediaWiki
1076 * @subpackage DifferenceEngine
1084 * Computes diff between sequences of strings.
1086 * @param $from_lines array An array of strings.
1087 * (Typically these are lines from a file.)
1088 * @param $to_lines array An array of strings.
1090 function Diff($from_lines, $to_lines) {
1091 $eng = new _DiffEngine
;
1092 $this->edits
= $eng->diff($from_lines, $to_lines);
1093 //$this->_check($from_lines, $to_lines);
1097 * Compute reversed Diff.
1101 * $diff = new Diff($lines1, $lines2);
1102 * $rev = $diff->reverse();
1103 * @return object A Diff object representing the inverse of the
1106 function reverse () {
1108 $rev->edits
= array();
1109 foreach ($this->edits
as $edit) {
1110 $rev->edits
[] = $edit->reverse();
1116 * Check for empty diff.
1118 * @return bool True iff two sequences were identical.
1120 function isEmpty () {
1121 foreach ($this->edits
as $edit) {
1122 if ($edit->type
!= 'copy')
1129 * Compute the length of the Longest Common Subsequence (LCS).
1131 * This is mostly for diagnostic purposed.
1133 * @return int The length of the LCS.
1137 foreach ($this->edits
as $edit) {
1138 if ($edit->type
== 'copy')
1139 $lcs +
= sizeof($edit->orig
);
1145 * Get the original set of lines.
1147 * This reconstructs the $from_lines parameter passed to the
1150 * @return array The original sequence of strings.
1155 foreach ($this->edits
as $edit) {
1157 array_splice($lines, sizeof($lines), 0, $edit->orig
);
1163 * Get the closing set of lines.
1165 * This reconstructs the $to_lines parameter passed to the
1168 * @return array The sequence of strings.
1170 function closing() {
1173 foreach ($this->edits
as $edit) {
1175 array_splice($lines, sizeof($lines), 0, $edit->closing
);
1181 * Check a Diff for validity.
1183 * This is here only for debugging purposes.
1185 function _check ($from_lines, $to_lines) {
1186 $fname = 'Diff::_check';
1187 wfProfileIn( $fname );
1188 if (serialize($from_lines) != serialize($this->orig()))
1189 trigger_error("Reconstructed original doesn't match", E_USER_ERROR
);
1190 if (serialize($to_lines) != serialize($this->closing()))
1191 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR
);
1193 $rev = $this->reverse();
1194 if (serialize($to_lines) != serialize($rev->orig()))
1195 trigger_error("Reversed original doesn't match", E_USER_ERROR
);
1196 if (serialize($from_lines) != serialize($rev->closing()))
1197 trigger_error("Reversed closing doesn't match", E_USER_ERROR
);
1201 foreach ($this->edits
as $edit) {
1202 if ( $prevtype == $edit->type
)
1203 trigger_error("Edit sequence is non-optimal", E_USER_ERROR
);
1204 $prevtype = $edit->type
;
1207 $lcs = $this->lcs();
1208 trigger_error('Diff okay: LCS = '.$lcs, E_USER_NOTICE
);
1209 wfProfileOut( $fname );
1217 * @package MediaWiki
1218 * @subpackage DifferenceEngine
1220 class MappedDiff
extends Diff
1225 * Computes diff between sequences of strings.
1227 * This can be used to compute things like
1228 * case-insensitve diffs, or diffs which ignore
1229 * changes in white-space.
1231 * @param $from_lines array An array of strings.
1232 * (Typically these are lines from a file.)
1234 * @param $to_lines array An array of strings.
1236 * @param $mapped_from_lines array This array should
1237 * have the same size number of elements as $from_lines.
1238 * The elements in $mapped_from_lines and
1239 * $mapped_to_lines are what is actually compared
1240 * when computing the diff.
1242 * @param $mapped_to_lines array This array should
1243 * have the same number of elements as $to_lines.
1245 function MappedDiff($from_lines, $to_lines,
1246 $mapped_from_lines, $mapped_to_lines) {
1247 $fname = 'MappedDiff::MappedDiff';
1248 wfProfileIn( $fname );
1250 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
1251 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
1253 $this->Diff($mapped_from_lines, $mapped_to_lines);
1256 for ($i = 0; $i < sizeof($this->edits
); $i++
) {
1257 $orig = &$this->edits
[$i]->orig
;
1258 if (is_array($orig)) {
1259 $orig = array_slice($from_lines, $xi, sizeof($orig));
1260 $xi +
= sizeof($orig);
1263 $closing = &$this->edits
[$i]->closing
;
1264 if (is_array($closing)) {
1265 $closing = array_slice($to_lines, $yi, sizeof($closing));
1266 $yi +
= sizeof($closing);
1269 wfProfileOut( $fname );
1274 * A class to format Diffs
1276 * This class formats the diff in classic diff format.
1277 * It is intended that this class be customized via inheritance,
1278 * to obtain fancier outputs.
1281 * @package MediaWiki
1282 * @subpackage DifferenceEngine
1287 * Number of leading context "lines" to preserve.
1289 * This should be left at zero for this class, but subclasses
1290 * may want to set this to other values.
1292 var $leading_context_lines = 0;
1295 * Number of trailing context "lines" to preserve.
1297 * This should be left at zero for this class, but subclasses
1298 * may want to set this to other values.
1300 var $trailing_context_lines = 0;
1305 * @param $diff object A Diff object.
1306 * @return string The formatted output.
1308 function format($diff) {
1309 $fname = 'DiffFormatter::format';
1310 wfProfileIn( $fname );
1316 $nlead = $this->leading_context_lines
;
1317 $ntrail = $this->trailing_context_lines
;
1319 $this->_start_diff();
1321 foreach ($diff->edits
as $edit) {
1322 if ($edit->type
== 'copy') {
1323 if (is_array($block)) {
1324 if (sizeof($edit->orig
) <= $nlead +
$ntrail) {
1329 $context = array_slice($edit->orig
, 0, $ntrail);
1330 $block[] = new _DiffOp_Copy($context);
1332 $this->_block($x0, $ntrail +
$xi - $x0,
1333 $y0, $ntrail +
$yi - $y0,
1338 $context = $edit->orig
;
1341 if (! is_array($block)) {
1342 $context = array_slice($context, sizeof($context) - $nlead);
1343 $x0 = $xi - sizeof($context);
1344 $y0 = $yi - sizeof($context);
1347 $block[] = new _DiffOp_Copy($context);
1353 $xi +
= sizeof($edit->orig
);
1355 $yi +
= sizeof($edit->closing
);
1358 if (is_array($block))
1359 $this->_block($x0, $xi - $x0,
1363 $end = $this->_end_diff();
1364 wfProfileOut( $fname );
1368 function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
1369 $fname = 'DiffFormatter::_block';
1370 wfProfileIn( $fname );
1371 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
1372 foreach ($edits as $edit) {
1373 if ($edit->type
== 'copy')
1374 $this->_context($edit->orig
);
1375 elseif ($edit->type
== 'add')
1376 $this->_added($edit->closing
);
1377 elseif ($edit->type
== 'delete')
1378 $this->_deleted($edit->orig
);
1379 elseif ($edit->type
== 'change')
1380 $this->_changed($edit->orig
, $edit->closing
);
1382 trigger_error('Unknown edit type', E_USER_ERROR
);
1384 $this->_end_block();
1385 wfProfileOut( $fname );
1388 function _start_diff() {
1392 function _end_diff() {
1393 $val = ob_get_contents();
1398 function _block_header($xbeg, $xlen, $ybeg, $ylen) {
1400 $xbeg .= "," . ($xbeg +
$xlen - 1);
1402 $ybeg .= "," . ($ybeg +
$ylen - 1);
1404 return $xbeg . ($xlen ?
($ylen ?
'c' : 'd') : 'a') . $ybeg;
1407 function _start_block($header) {
1411 function _end_block() {
1414 function _lines($lines, $prefix = ' ') {
1415 foreach ($lines as $line)
1416 echo "$prefix $line\n";
1419 function _context($lines) {
1420 $this->_lines($lines);
1423 function _added($lines) {
1424 $this->_lines($lines, '>');
1426 function _deleted($lines) {
1427 $this->_lines($lines, '<');
1430 function _changed($orig, $closing) {
1431 $this->_deleted($orig);
1433 $this->_added($closing);
1439 * Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1443 define('NBSP', ' '); // iso-8859-x non-breaking space.
1448 * @package MediaWiki
1449 * @subpackage DifferenceEngine
1451 class _HWLDF_WordAccumulator
{
1452 function _HWLDF_WordAccumulator () {
1453 $this->_lines
= array();
1459 function _flushGroup ($new_tag) {
1460 if ($this->_group
!== '') {
1461 if ($this->_tag
== 'mark')
1462 $this->_line
.= '<span class="diffchange">' .
1463 htmlspecialchars ( $this->_group
) . '</span>';
1465 $this->_line
.= htmlspecialchars ( $this->_group
);
1468 $this->_tag
= $new_tag;
1471 function _flushLine ($new_tag) {
1472 $this->_flushGroup($new_tag);
1473 if ($this->_line
!= '')
1474 array_push ( $this->_lines
, $this->_line
);
1476 # make empty lines visible by inserting an NBSP
1477 array_push ( $this->_lines
, NBSP
);
1481 function addWords ($words, $tag = '') {
1482 if ($tag != $this->_tag
)
1483 $this->_flushGroup($tag);
1485 foreach ($words as $word) {
1486 // new-line should only come as first char of word.
1489 if ($word[0] == "\n") {
1490 $this->_flushLine($tag);
1491 $word = substr($word, 1);
1493 assert(!strstr($word, "\n"));
1494 $this->_group
.= $word;
1498 function getLines() {
1499 $this->_flushLine('~done');
1500 return $this->_lines
;
1507 * @package MediaWiki
1508 * @subpackage DifferenceEngine
1510 class WordLevelDiff
extends MappedDiff
1512 function WordLevelDiff ($orig_lines, $closing_lines) {
1513 $fname = 'WordLevelDiff::WordLevelDiff';
1514 wfProfileIn( $fname );
1516 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1517 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1519 $this->MappedDiff($orig_words, $closing_words,
1520 $orig_stripped, $closing_stripped);
1521 wfProfileOut( $fname );
1524 function _split($lines) {
1525 $fname = 'WordLevelDiff::_split';
1526 wfProfileIn( $fname );
1529 $stripped = array();
1531 foreach ( $lines as $line ) {
1532 # If the line is too long, just pretend the entire line is one big word
1533 # This prevents resource exhaustion problems
1540 if ( strlen( $line ) > MAX_DIFF_LINE
) {
1542 $stripped[] = $line;
1544 if (preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1547 $words = array_merge( $words, $m[0] );
1548 $stripped = array_merge( $stripped, $m[1] );
1552 wfProfileOut( $fname );
1553 return array($words, $stripped);
1557 $fname = 'WordLevelDiff::orig';
1558 wfProfileIn( $fname );
1559 $orig = new _HWLDF_WordAccumulator
;
1561 foreach ($this->edits
as $edit) {
1562 if ($edit->type
== 'copy')
1563 $orig->addWords($edit->orig
);
1564 elseif ($edit->orig
)
1565 $orig->addWords($edit->orig
, 'mark');
1567 $lines = $orig->getLines();
1568 wfProfileOut( $fname );
1572 function closing () {
1573 $fname = 'WordLevelDiff::closing';
1574 wfProfileIn( $fname );
1575 $closing = new _HWLDF_WordAccumulator
;
1577 foreach ($this->edits
as $edit) {
1578 if ($edit->type
== 'copy')
1579 $closing->addWords($edit->closing
);
1580 elseif ($edit->closing
)
1581 $closing->addWords($edit->closing
, 'mark');
1583 $lines = $closing->getLines();
1584 wfProfileOut( $fname );
1590 * Wikipedia Table style diff formatter.
1593 * @package MediaWiki
1594 * @subpackage DifferenceEngine
1596 class TableDiffFormatter
extends DiffFormatter
1598 function TableDiffFormatter() {
1599 $this->leading_context_lines
= 2;
1600 $this->trailing_context_lines
= 2;
1603 function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1604 $l1 = wfMsg( 'lineno', $xbeg );
1605 $l2 = wfMsg( 'lineno', $ybeg );
1607 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1608 '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1612 function _start_block( $header ) {
1617 function _end_block() {
1620 function _lines( $lines, $prefix=' ', $color='white' ) {
1623 # HTML-escape parameter before calling this
1624 function addedLine( $line ) {
1625 return "<td>+</td><td class='diff-addedline'>{$line}</td>";
1628 # HTML-escape parameter before calling this
1629 function deletedLine( $line ) {
1630 return "<td>-</td><td class='diff-deletedline'>{$line}</td>";
1633 # HTML-escape parameter before calling this
1634 function contextLine( $line ) {
1635 return "<td> </td><td class='diff-context'>{$line}</td>";
1638 function emptyLine() {
1639 return '<td colspan="2"> </td>';
1642 function _added( $lines ) {
1643 foreach ($lines as $line) {
1644 echo '<tr>' . $this->emptyLine() .
1645 $this->addedLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1649 function _deleted($lines) {
1650 foreach ($lines as $line) {
1651 echo '<tr>' . $this->deletedLine( htmlspecialchars ( $line ) ) .
1652 $this->emptyLine() . "</tr>\n";
1656 function _context( $lines ) {
1657 foreach ($lines as $line) {
1659 $this->contextLine( htmlspecialchars ( $line ) ) .
1660 $this->contextLine( htmlspecialchars ( $line ) ) . "</tr>\n";
1664 function _changed( $orig, $closing ) {
1665 $fname = 'TableDiffFormatter::_changed';
1666 wfProfileIn( $fname );
1668 $diff = new WordLevelDiff( $orig, $closing );
1669 $del = $diff->orig();
1670 $add = $diff->closing();
1672 # Notice that WordLevelDiff returns HTML-escaped output.
1673 # Hence, we will be calling addedLine/deletedLine without HTML-escaping.
1675 while ( $line = array_shift( $del ) ) {
1676 $aline = array_shift( $add );
1677 echo '<tr>' . $this->deletedLine( $line ) .
1678 $this->addedLine( $aline ) . "</tr>\n";
1680 foreach ($add as $line) { # If any leftovers
1681 echo '<tr>' . $this->emptyLine() .
1682 $this->addedLine( $line ) . "</tr>\n";
1684 wfProfileOut( $fname );