4 class DifferenceEngine
{
5 /* private */ var $mOldid, $mNewid;
6 /* private */ var $mOldtitle, $mNewtitle;
7 /* private */ var $mOldtext, $mNewtext;
8 /* private */ var $mOldUser, $mNewUser;
9 /* private */ var $mOldComment, $mNewComment;
10 /* private */ var $mOldPage, $mNewPage;
12 function DifferenceEngine( $old, $new )
18 function showDiffPage()
20 global $wgUser, $wgTitle, $wgOut, $wgLang;
21 $fname = "DifferenceEngine::showDiffPage";
22 wfProfileIn( $fname );
24 $t = $wgTitle->getPrefixedText() . " (Diff: {$this->mOldid}, " .
26 $mtext = wfMsg( "missingarticle", $t );
28 $wgOut->setArticleFlag( false );
29 if ( ! $this->loadText() ) {
30 $wgOut->setPagetitle( wfMsg( "errorpagetitle" ) );
31 $wgOut->addHTML( $mtext );
32 wfProfileOut( $fname );
35 $wgOut->suppressQuickbar();
37 $oldTitle = $this->mOldPage
->getPrefixedText();
38 $newTitle = $this->mNewPage
->getPrefixedText();
39 if( $oldTitle == $newTitle ) {
40 $wgOut->setPageTitle( $newTitle );
42 $wgOut->setPageTitle( $oldTitle . ", " . $newTitle );
44 $wgOut->setSubtitle( wfMsg( "difference" ) );
45 $wgOut->setRobotpolicy( "noindex,follow" );
47 if ( !( $this->mOldPage
->userCanRead() && $this->mNewPage
->userCanRead() ) ) {
50 wfProfileOut( $fname );
54 $sk = $wgUser->getSkin();
55 $talk = $wgLang->getNsText( NS_TALK
);
56 $contribs = wfMsg( "contribslink" );
59 $this->mOldComment
= $sk->formatComment($this->mOldComment
);
60 $this->mNewComment
= $sk->formatComment($this->mNewComment
);
63 $oldUserLink = $sk->makeLinkObj( Title
::makeTitle( NS_USER
, $this->mOldUser
), $this->mOldUser
);
64 $newUserLink = $sk->makeLinkObj( Title
::makeTitle( NS_USER
, $this->mNewUser
), $this->mNewUser
);
65 $oldUTLink = $sk->makeLinkObj( Title
::makeTitle( NS_USER_TALK
, $this->mOldUser
), $talk );
66 $newUTLink = $sk->makeLinkObj( Title
::makeTitle( NS_USER_TALK
, $this->mNewUser
), $talk );
67 $oldContribs = $sk->makeKnownLinkObj( Title
::makeTitle( NS_SPECIAL
, "Contributions" ), $contribs,
68 "target=" . urlencode($this->mOldUser
) );
69 $newContribs = $sk->makeKnownLinkObj( Title
::makeTitle( NS_SPECIAL
, "Contributions" ), $contribs,
70 "target=" . urlencode($this->mNewUser
) );
71 if ( !$this->mNewid
&& $wgUser->isSysop() ) {
72 $rollback = " <strong>[" . $sk->makeKnownLinkObj( $wgTitle, wfMsg( "rollbacklink" ),
73 "action=rollback&from=" . urlencode($this->mNewUser
) ) . "]</strong>";
78 $oldHeader = "<strong>{$this->mOldtitle}</strong><br />$oldUserLink ($oldUTLink | $oldContribs)<br />".
80 $newHeader = "<strong>{$this->mNewtitle}</strong><br />$newUserLink ($newUTLink | $newContribs) $rollback<br />".
83 DifferenceEngine
::showDiff( $this->mOldtext
, $this->mNewtext
,
84 $oldHeader, $newHeader );
85 $wgOut->addHTML( "<hr /><h2>{$this->mNewtitle}</h2>\n" );
86 $wgOut->addWikiText( $this->mNewtext
);
88 wfProfileOut( $fname );
91 function showDiff( $otext, $ntext, $otitle, $ntitle )
95 $ota = explode( "\n", str_replace( "\r\n", "\n",
96 htmlspecialchars( $otext ) ) );
97 $nta = explode( "\n", str_replace( "\r\n", "\n",
98 htmlspecialchars( $ntext ) ) );
100 $wgOut->addHTML( "<table border='0' width='98%'
101 cellpadding='0' cellspacing='4px' class='diff'><tr>
102 <td colspan='2' width='50%' align='center' class='diff-otitle'>
104 <td colspan='2' width='50%' align='center' class='diff-ntitle'>
108 $diffs = new Diff( $ota, $nta );
109 $formatter = new TableDiffFormatter();
110 $formatter->format( $diffs );
111 $wgOut->addHTML( "</table>\n" );
114 # Load the text of the articles to compare. If newid is 0, then compare
115 # the old article in oldid to the current article; if oldid is 0, then
116 # compare the current article to the immediately previous one (ignoring
117 # the value of newid).
121 global $wgTitle, $wgOut, $wgLang;
122 $fname = "DifferenceEngine::loadText";
124 $dbr =& wfGetDB( DB_SLAVE
);
125 if ( 0 == $this->mNewid ||
0 == $this->mOldid
) {
126 $wgOut->setArticleFlag( true );
127 $this->mNewtitle
= wfMsg( "currentrev" );
128 $id = $wgTitle->getArticleID();
130 $s = $dbr->getArray( 'cur', array( 'cur_text', 'cur_user_text', 'cur_comment' ),
131 array( 'cur_id' => $id ), $fname );
132 if ( $s === false ) {
136 $this->mNewPage
= &$wgTitle;
137 $this->mNewtext
= $s->cur_text
;
138 $this->mNewUser
= $s->cur_user_text
;
139 $this->mNewComment
= $s->cur_comment
;
141 $s = $dbr->getArray( 'old', array( 'old_namespace','old_title','old_timestamp', 'old_text',
142 'old_flags','old_user_text','old_comment' ), array( 'old_id' => $this->mNewid
), $fname );
144 if ( $s === false ) {
148 $this->mNewtext
= Article
::getRevisionText( $s );
150 $t = $wgLang->timeanddate( $s->old_timestamp
, true );
151 $this->mNewPage
= Title
::MakeTitle( $s->old_namespace
, $s->old_title
);
152 $this->mNewtitle
= wfMsg( "revisionasof", $t );
153 $this->mNewUser
= $s->old_user_text
;
154 $this->mNewComment
= $s->old_comment
;
156 if ( 0 == $this->mOldid
) {
157 $s = $dbr->getArray( 'old',
158 array( 'old_namespace','old_title','old_timestamp','old_text', 'old_flags','old_user_text','old_comment' ),
160 'old_namespace' => $this->mNewPage
->getNamespace(),
161 'old_title' => $this->mNewPage
->getDBkey()
162 ), $fname, array( 'ORDER BY' => 'inverse_timestamp', 'USE INDEX' => 'name_title_timestamp' )
165 $s = $dbr->getArray( 'old',
166 array( 'old_namespace','old_title','old_timestamp','old_text','old_flags','old_user_text','old_comment'),
167 array( 'old_id' => $this->mOldid
),
171 if ( $s === false ) {
174 $this->mOldPage
= Title
::MakeTitle( $s->old_namespace
, $s->old_title
);
175 $this->mOldtext
= Article
::getRevisionText( $s );
177 $t = $wgLang->timeanddate( $s->old_timestamp
, true );
178 $this->mOldtitle
= wfMsg( "revisionasof", $t );
179 $this->mOldUser
= $s->old_user_text
;
180 $this->mOldComment
= $s->old_comment
;
186 // A PHP diff engine for phpwiki. (Taken from phpwiki-1.3.3)
188 // Copyright (C) 2000, 2001 Geoffrey T. Dairiki <dairiki@dairiki.org>
189 // You may copy this code freely under the conditions of the GPL.
192 define('USE_ASSERTS', function_exists('assert'));
200 trigger_error("pure virtual", E_USER_ERROR
);
204 return $this->orig ?
sizeof($this->orig
) : 0;
207 function nclosing() {
208 return $this->closing ?
sizeof($this->closing
) : 0;
212 class _DiffOp_Copy
extends _DiffOp
{
215 function _DiffOp_Copy ($orig, $closing = false) {
216 if (!is_array($closing))
219 $this->closing
= $closing;
223 return new _DiffOp_Copy($this->closing
, $this->orig
);
227 class _DiffOp_Delete
extends _DiffOp
{
228 var $type = 'delete';
230 function _DiffOp_Delete ($lines) {
231 $this->orig
= $lines;
232 $this->closing
= false;
236 return new _DiffOp_Add($this->orig
);
240 class _DiffOp_Add
extends _DiffOp
{
243 function _DiffOp_Add ($lines) {
244 $this->closing
= $lines;
249 return new _DiffOp_Delete($this->closing
);
253 class _DiffOp_Change
extends _DiffOp
{
254 var $type = 'change';
256 function _DiffOp_Change ($orig, $closing) {
258 $this->closing
= $closing;
262 return new _DiffOp_Change($this->closing
, $this->orig
);
268 * Class used internally by Diff to actually compute the diffs.
270 * The algorithm used here is mostly lifted from the perl module
271 * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
272 * http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
274 * More ideas are taken from:
275 * http://www.ics.uci.edu/~eppstein/161/960229.html
277 * Some ideas are (and a bit of code) are from from analyze.c, from GNU
278 * diffutils-2.7, which can be found at:
279 * ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
281 * closingly, some ideas (subdivision by NCHUNKS > 2, and some optimizations)
284 * @author Geoffrey T. Dairiki
289 function diff ($from_lines, $to_lines) {
290 $n_from = sizeof($from_lines);
291 $n_to = sizeof($to_lines);
293 $this->xchanged
= $this->ychanged
= array();
294 $this->xv
= $this->yv
= array();
295 $this->xind
= $this->yind
= array();
297 unset($this->in_seq
);
300 // Skip leading common lines.
301 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++
) {
302 if ($from_lines[$skip] != $to_lines[$skip])
304 $this->xchanged
[$skip] = $this->ychanged
[$skip] = false;
306 // Skip trailing common lines.
307 $xi = $n_from; $yi = $n_to;
308 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++
) {
309 if ($from_lines[$xi] != $to_lines[$yi])
311 $this->xchanged
[$xi] = $this->ychanged
[$yi] = false;
314 // Ignore lines which do not exist in both files.
315 for ($xi = $skip; $xi < $n_from - $endskip; $xi++
)
316 $xhash[$from_lines[$xi]] = 1;
317 for ($yi = $skip; $yi < $n_to - $endskip; $yi++
) {
318 $line = $to_lines[$yi];
319 if ( ($this->ychanged
[$yi] = empty($xhash[$line])) )
325 for ($xi = $skip; $xi < $n_from - $endskip; $xi++
) {
326 $line = $from_lines[$xi];
327 if ( ($this->xchanged
[$xi] = empty($yhash[$line])) )
334 $this->_compareseq(0, sizeof($this->xv
), 0, sizeof($this->yv
));
336 // Merge edits when possible
337 $this->_shift_boundaries($from_lines, $this->xchanged
, $this->ychanged
);
338 $this->_shift_boundaries($to_lines, $this->ychanged
, $this->xchanged
);
340 // Compute the edit operations.
343 while ($xi < $n_from ||
$yi < $n_to) {
344 USE_ASSERTS
&& assert($yi < $n_to ||
$this->xchanged
[$xi]);
345 USE_ASSERTS
&& assert($xi < $n_from ||
$this->ychanged
[$yi]);
347 // Skip matching "snake".
349 while ( $xi < $n_from && $yi < $n_to
350 && !$this->xchanged
[$xi] && !$this->ychanged
[$yi]) {
351 $copy[] = $from_lines[$xi++
];
355 $edits[] = new _DiffOp_Copy($copy);
357 // Find deletes & adds.
359 while ($xi < $n_from && $this->xchanged
[$xi])
360 $delete[] = $from_lines[$xi++
];
363 while ($yi < $n_to && $this->ychanged
[$yi])
364 $add[] = $to_lines[$yi++
];
367 $edits[] = new _DiffOp_Change($delete, $add);
369 $edits[] = new _DiffOp_Delete($delete);
371 $edits[] = new _DiffOp_Add($add);
377 /* Divide the Largest Common Subsequence (LCS) of the sequences
378 * [XOFF, XLIM) and [YOFF, YLIM) into NCHUNKS approximately equally
381 * Returns (LCS, PTS). LCS is the length of the LCS. PTS is an
382 * array of NCHUNKS+1 (X, Y) indexes giving the diving points between
383 * sub sequences. The first sub-sequence is contained in [X0, X1),
384 * [Y0, Y1), the second in [X1, X2), [Y1, Y2) and so on. Note
385 * that (X0, Y0) == (XOFF, YOFF) and
386 * (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
388 * This function assumes that the first lines of the specified portions
389 * of the two files do not match, and likewise that the last lines do not
390 * match. The caller must trim matching lines from the beginning and end
391 * of the portions it is going to specify.
393 function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks) {
396 if ($xlim - $xoff > $ylim - $yoff) {
397 // Things seems faster (I'm not sure I understand why)
398 // when the shortest sequence in X.
400 list ($xoff, $xlim, $yoff, $ylim)
401 = array( $yoff, $ylim, $xoff, $xlim);
405 for ($i = $ylim - 1; $i >= $yoff; $i--)
406 $ymatches[$this->xv
[$i]][] = $i;
408 for ($i = $ylim - 1; $i >= $yoff; $i--)
409 $ymatches[$this->yv
[$i]][] = $i;
412 $this->seq
[0]= $yoff - 1;
413 $this->in_seq
= array();
416 $numer = $xlim - $xoff +
$nchunks - 1;
418 for ($chunk = 0; $chunk < $nchunks; $chunk++
) {
420 for ($i = 0; $i <= $this->lcs
; $i++
)
421 $ymids[$i][$chunk-1] = $this->seq
[$i];
423 $x1 = $xoff +
(int)(($numer +
($xlim-$xoff)*$chunk) / $nchunks);
424 for ( ; $x < $x1; $x++
) {
425 $line = $flip ?
$this->yv
[$x] : $this->xv
[$x];
426 if (empty($ymatches[$line]))
428 $matches = $ymatches[$line];
430 while (list ($junk, $y) = each($matches))
431 if (empty($this->in_seq
[$y])) {
432 $k = $this->_lcs_pos($y);
433 USE_ASSERTS
&& assert($k > 0);
434 $ymids[$k] = $ymids[$k-1];
437 while (list ($junk, $y) = each($matches)) {
438 if ($y > $this->seq
[$k-1]) {
439 USE_ASSERTS
&& assert($y < $this->seq
[$k]);
440 // Optimization: this is a common case:
441 // next match is just replacing previous match.
442 $this->in_seq
[$this->seq
[$k]] = false;
444 $this->in_seq
[$y] = 1;
446 else if (empty($this->in_seq
[$y])) {
447 $k = $this->_lcs_pos($y);
448 USE_ASSERTS
&& assert($k > 0);
449 $ymids[$k] = $ymids[$k-1];
455 $seps[] = $flip ?
array($yoff, $xoff) : array($xoff, $yoff);
456 $ymid = $ymids[$this->lcs
];
457 for ($n = 0; $n < $nchunks - 1; $n++
) {
458 $x1 = $xoff +
(int)(($numer +
($xlim - $xoff) * $n) / $nchunks);
460 $seps[] = $flip ?
array($y1, $x1) : array($x1, $y1);
462 $seps[] = $flip ?
array($ylim, $xlim) : array($xlim, $ylim);
464 return array($this->lcs
, $seps);
467 function _lcs_pos ($ypos) {
469 if ($end == 0 ||
$ypos > $this->seq
[$end]) {
470 $this->seq
[++
$this->lcs
] = $ypos;
471 $this->in_seq
[$ypos] = 1;
476 while ($beg < $end) {
477 $mid = (int)(($beg +
$end) / 2);
478 if ( $ypos > $this->seq
[$mid] )
484 USE_ASSERTS
&& assert($ypos != $this->seq
[$end]);
486 $this->in_seq
[$this->seq
[$end]] = false;
487 $this->seq
[$end] = $ypos;
488 $this->in_seq
[$ypos] = 1;
492 /* Find LCS of two sequences.
494 * The results are recorded in the vectors $this->{x,y}changed[], by
495 * storing a 1 in the element for each line that is an insertion
496 * or deletion (ie. is not in the LCS).
498 * The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
500 * Note that XLIM, YLIM are exclusive bounds.
501 * All line numbers are origin-0 and discarded lines are not counted.
503 function _compareseq ($xoff, $xlim, $yoff, $ylim) {
504 // Slide down the bottom initial diagonal.
505 while ($xoff < $xlim && $yoff < $ylim
506 && $this->xv
[$xoff] == $this->yv
[$yoff]) {
511 // Slide up the top initial diagonal.
512 while ($xlim > $xoff && $ylim > $yoff
513 && $this->xv
[$xlim - 1] == $this->yv
[$ylim - 1]) {
518 if ($xoff == $xlim ||
$yoff == $ylim)
521 // This is ad hoc but seems to work well.
522 //$nchunks = sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5);
523 //$nchunks = max(2,min(8,(int)$nchunks));
524 $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) +
1;
526 = $this->_diag($xoff,$xlim,$yoff, $ylim,$nchunks);
530 // X and Y sequences have no common subsequence:
532 while ($yoff < $ylim)
533 $this->ychanged
[$this->yind
[$yoff++
]] = 1;
534 while ($xoff < $xlim)
535 $this->xchanged
[$this->xind
[$xoff++
]] = 1;
538 // Use the partitions to split this problem into subproblems.
541 while ($pt2 = next($seps)) {
542 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
548 /* Adjust inserts/deletes of identical lines to join changes
549 * as much as possible.
551 * We do something when a run of changed lines include a
552 * line at one end and has an excluded, identical line at the other.
553 * We are free to choose which identical line is included.
554 * `compareseq' usually chooses the one at the beginning,
555 * but usually it is cleaner to consider the following identical line
556 * to be the "change".
558 * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
560 function _shift_boundaries ($lines, &$changed, $other_changed) {
564 USE_ASSERTS
&& assert('sizeof($lines) == sizeof($changed)');
565 $len = sizeof($lines);
566 $other_len = sizeof($other_changed);
570 * Scan forwards to find beginning of another run of changes.
571 * Also keep track of the corresponding point in the other file.
573 * Throughout this code, $i and $j are adjusted together so that
574 * the first $i elements of $changed and the first $j elements
575 * of $other_changed both contain the same number of zeros
577 * Furthermore, $j is always kept so that $j == $other_len or
578 * $other_changed[$j] == false.
580 while ($j < $other_len && $other_changed[$j])
583 while ($i < $len && ! $changed[$i]) {
584 USE_ASSERTS
&& assert('$j < $other_len && ! $other_changed[$j]');
586 while ($j < $other_len && $other_changed[$j])
595 // Find the end of this run of changes.
596 while (++
$i < $len && $changed[$i])
601 * Record the length of this run of changes, so that
602 * we can later determine whether the run has grown.
604 $runlength = $i - $start;
607 * Move the changed region back, so long as the
608 * previous unchanged line matches the last changed one.
609 * This merges with previous changed regions.
611 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
612 $changed[--$start] = 1;
613 $changed[--$i] = false;
614 while ($start > 0 && $changed[$start - 1])
616 USE_ASSERTS
&& assert('$j > 0');
617 while ($other_changed[--$j])
619 USE_ASSERTS
&& assert('$j >= 0 && !$other_changed[$j]');
623 * Set CORRESPONDING to the end of the changed run, at the last
624 * point where it corresponds to a changed run in the other file.
625 * CORRESPONDING == LEN means no such point has been found.
627 $corresponding = $j < $other_len ?
$i : $len;
630 * Move the changed region forward, so long as the
631 * first changed line matches the following unchanged one.
632 * This merges with following changed regions.
633 * Do this second, so that if there are no merges,
634 * the changed region is moved forward as far as possible.
636 while ($i < $len && $lines[$start] == $lines[$i]) {
637 $changed[$start++
] = false;
639 while ($i < $len && $changed[$i])
642 USE_ASSERTS
&& assert('$j < $other_len && ! $other_changed[$j]');
644 if ($j < $other_len && $other_changed[$j]) {
646 while ($j < $other_len && $other_changed[$j])
650 } while ($runlength != $i - $start);
653 * If possible, move the fully-merged run of changes
654 * back to a corresponding run in the other file.
656 while ($corresponding < $i) {
657 $changed[--$start] = 1;
659 USE_ASSERTS
&& assert('$j > 0');
660 while ($other_changed[--$j])
662 USE_ASSERTS
&& assert('$j >= 0 && !$other_changed[$j]');
669 * Class representing a 'diff' between two sequences of strings.
677 * Computes diff between sequences of strings.
679 * @param $from_lines array An array of strings.
680 * (Typically these are lines from a file.)
681 * @param $to_lines array An array of strings.
683 function Diff($from_lines, $to_lines) {
684 $eng = new _DiffEngine
;
685 $this->edits
= $eng->diff($from_lines, $to_lines);
686 //$this->_check($from_lines, $to_lines);
690 * Compute reversed Diff.
694 * $diff = new Diff($lines1, $lines2);
695 * $rev = $diff->reverse();
696 * @return object A Diff object representing the inverse of the
699 function reverse () {
701 $rev->edits
= array();
702 foreach ($this->edits
as $edit) {
703 $rev->edits
[] = $edit->reverse();
709 * Check for empty diff.
711 * @return bool True iff two sequences were identical.
713 function isEmpty () {
714 foreach ($this->edits
as $edit) {
715 if ($edit->type
!= 'copy')
722 * Compute the length of the Longest Common Subsequence (LCS).
724 * This is mostly for diagnostic purposed.
726 * @return int The length of the LCS.
730 foreach ($this->edits
as $edit) {
731 if ($edit->type
== 'copy')
732 $lcs +
= sizeof($edit->orig
);
738 * Get the original set of lines.
740 * This reconstructs the $from_lines parameter passed to the
743 * @return array The original sequence of strings.
748 foreach ($this->edits
as $edit) {
750 array_splice($lines, sizeof($lines), 0, $edit->orig
);
756 * Get the closing set of lines.
758 * This reconstructs the $to_lines parameter passed to the
761 * @return array The sequence of strings.
766 foreach ($this->edits
as $edit) {
768 array_splice($lines, sizeof($lines), 0, $edit->closing
);
774 * Check a Diff for validity.
776 * This is here only for debugging purposes.
778 function _check ($from_lines, $to_lines) {
779 if (serialize($from_lines) != serialize($this->orig()))
780 trigger_error("Reconstructed original doesn't match", E_USER_ERROR
);
781 if (serialize($to_lines) != serialize($this->closing()))
782 trigger_error("Reconstructed closing doesn't match", E_USER_ERROR
);
784 $rev = $this->reverse();
785 if (serialize($to_lines) != serialize($rev->orig()))
786 trigger_error("Reversed original doesn't match", E_USER_ERROR
);
787 if (serialize($from_lines) != serialize($rev->closing()))
788 trigger_error("Reversed closing doesn't match", E_USER_ERROR
);
792 foreach ($this->edits
as $edit) {
793 if ( $prevtype == $edit->type
)
794 trigger_error("Edit sequence is non-optimal", E_USER_ERROR
);
795 $prevtype = $edit->type
;
799 trigger_error("Diff okay: LCS = $lcs", E_USER_NOTICE
);
812 * Computes diff between sequences of strings.
814 * This can be used to compute things like
815 * case-insensitve diffs, or diffs which ignore
816 * changes in white-space.
818 * @param $from_lines array An array of strings.
819 * (Typically these are lines from a file.)
821 * @param $to_lines array An array of strings.
823 * @param $mapped_from_lines array This array should
824 * have the same size number of elements as $from_lines.
825 * The elements in $mapped_from_lines and
826 * $mapped_to_lines are what is actually compared
827 * when computing the diff.
829 * @param $mapped_to_lines array This array should
830 * have the same number of elements as $to_lines.
832 function MappedDiff($from_lines, $to_lines,
833 $mapped_from_lines, $mapped_to_lines) {
835 assert(sizeof($from_lines) == sizeof($mapped_from_lines));
836 assert(sizeof($to_lines) == sizeof($mapped_to_lines));
838 $this->Diff($mapped_from_lines, $mapped_to_lines);
841 for ($i = 0; $i < sizeof($this->edits
); $i++
) {
842 $orig = &$this->edits
[$i]->orig
;
843 if (is_array($orig)) {
844 $orig = array_slice($from_lines, $xi, sizeof($orig));
845 $xi +
= sizeof($orig);
848 $closing = &$this->edits
[$i]->closing
;
849 if (is_array($closing)) {
850 $closing = array_slice($to_lines, $yi, sizeof($closing));
851 $yi +
= sizeof($closing);
858 * A class to format Diffs
860 * This class formats the diff in classic diff format.
861 * It is intended that this class be customized via inheritance,
862 * to obtain fancier outputs.
867 * Number of leading context "lines" to preserve.
869 * This should be left at zero for this class, but subclasses
870 * may want to set this to other values.
872 var $leading_context_lines = 0;
875 * Number of trailing context "lines" to preserve.
877 * This should be left at zero for this class, but subclasses
878 * may want to set this to other values.
880 var $trailing_context_lines = 0;
885 * @param $diff object A Diff object.
886 * @return string The formatted output.
888 function format($diff) {
894 $nlead = $this->leading_context_lines
;
895 $ntrail = $this->trailing_context_lines
;
897 $this->_start_diff();
899 foreach ($diff->edits
as $edit) {
900 if ($edit->type
== 'copy') {
901 if (is_array($block)) {
902 if (sizeof($edit->orig
) <= $nlead +
$ntrail) {
907 $context = array_slice($edit->orig
, 0, $ntrail);
908 $block[] = new _DiffOp_Copy($context);
910 $this->_block($x0, $ntrail +
$xi - $x0,
911 $y0, $ntrail +
$yi - $y0,
916 $context = $edit->orig
;
919 if (! is_array($block)) {
920 $context = array_slice($context, sizeof($context) - $nlead);
921 $x0 = $xi - sizeof($context);
922 $y0 = $yi - sizeof($context);
925 $block[] = new _DiffOp_Copy($context);
931 $xi +
= sizeof($edit->orig
);
933 $yi +
= sizeof($edit->closing
);
936 if (is_array($block))
937 $this->_block($x0, $xi - $x0,
941 return $this->_end_diff();
944 function _block($xbeg, $xlen, $ybeg, $ylen, &$edits) {
945 $this->_start_block($this->_block_header($xbeg, $xlen, $ybeg, $ylen));
946 foreach ($edits as $edit) {
947 if ($edit->type
== 'copy')
948 $this->_context($edit->orig
);
949 elseif ($edit->type
== 'add')
950 $this->_added($edit->closing
);
951 elseif ($edit->type
== 'delete')
952 $this->_deleted($edit->orig
);
953 elseif ($edit->type
== 'change')
954 $this->_changed($edit->orig
, $edit->closing
);
956 trigger_error("Unknown edit type", E_USER_ERROR
);
961 function _start_diff() {
965 function _end_diff() {
966 $val = ob_get_contents();
971 function _block_header($xbeg, $xlen, $ybeg, $ylen) {
973 $xbeg .= "," . ($xbeg +
$xlen - 1);
975 $ybeg .= "," . ($ybeg +
$ylen - 1);
977 return $xbeg . ($xlen ?
($ylen ?
'c' : 'd') : 'a') . $ybeg;
980 function _start_block($header) {
984 function _end_block() {
987 function _lines($lines, $prefix = ' ') {
988 foreach ($lines as $line)
989 echo "$prefix $line\n";
992 function _context($lines) {
993 $this->_lines($lines);
996 function _added($lines) {
997 $this->_lines($lines, ">");
999 function _deleted($lines) {
1000 $this->_lines($lines, "<");
1003 function _changed($orig, $closing) {
1004 $this->_deleted($orig);
1006 $this->_added($closing);
1012 * Additions by Axel Boldt follow, partly taken from diff.php, phpwiki-1.3.3
1016 define('NBSP', ' '); // iso-8859-x non-breaking space.
1018 class _HWLDF_WordAccumulator
{
1019 function _HWLDF_WordAccumulator () {
1020 $this->_lines
= array();
1026 function _flushGroup ($new_tag) {
1027 if ($this->_group
!== '') {
1028 if ($this->_tag
== 'mark')
1029 $this->_line
.= '<span class="diffchange">'.$this->_group
.'</span>';
1031 $this->_line
.= $this->_group
;
1034 $this->_tag
= $new_tag;
1037 function _flushLine ($new_tag) {
1038 $this->_flushGroup($new_tag);
1039 if ($this->_line
!= '')
1040 $this->_lines
[] = $this->_line
;
1044 function addWords ($words, $tag = '') {
1045 if ($tag != $this->_tag
)
1046 $this->_flushGroup($tag);
1048 foreach ($words as $word) {
1049 // new-line should only come as first char of word.
1052 if ($word[0] == "\n") {
1053 $this->_group
.= NBSP
;
1054 $this->_flushLine($tag);
1055 $word = substr($word, 1);
1057 assert(!strstr($word, "\n"));
1058 $this->_group
.= $word;
1062 function getLines() {
1063 $this->_flushLine('~done');
1064 return $this->_lines
;
1068 class WordLevelDiff
extends MappedDiff
1070 function WordLevelDiff ($orig_lines, $closing_lines) {
1071 list ($orig_words, $orig_stripped) = $this->_split($orig_lines);
1072 list ($closing_words, $closing_stripped) = $this->_split($closing_lines);
1075 $this->MappedDiff($orig_words, $closing_words,
1076 $orig_stripped, $closing_stripped);
1079 function _split($lines) {
1080 // FIXME: fix POSIX char class.
1081 # if (!preg_match_all('/ ( [^\S\n]+ | [[:alnum:]]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1082 if (!preg_match_all('/ ( [^\S\n]+ | [0-9_A-Za-z\x80-\xff]+ | . ) (?: (?!< \n) [^\S\n])? /xs',
1083 implode("\n", $lines),
1085 return array(array(''), array(''));
1087 return array($m[0], $m[1]);
1091 $orig = new _HWLDF_WordAccumulator
;
1093 foreach ($this->edits
as $edit) {
1094 if ($edit->type
== 'copy')
1095 $orig->addWords($edit->orig
);
1096 elseif ($edit->orig
)
1097 $orig->addWords($edit->orig
, 'mark');
1099 return $orig->getLines();
1102 function closing () {
1103 $closing = new _HWLDF_WordAccumulator
;
1105 foreach ($this->edits
as $edit) {
1106 if ($edit->type
== 'copy')
1107 $closing->addWords($edit->closing
);
1108 elseif ($edit->closing
)
1109 $closing->addWords($edit->closing
, 'mark');
1111 return $closing->getLines();
1116 * Wikipedia Table style diff formatter.
1119 class TableDiffFormatter
extends DiffFormatter
1121 function TableDiffFormatter() {
1122 $this->leading_context_lines
= 2;
1123 $this->trailing_context_lines
= 2;
1126 function _block_header( $xbeg, $xlen, $ybeg, $ylen ) {
1127 $l1 = wfMsg( "lineno", $xbeg );
1128 $l2 = wfMsg( "lineno", $ybeg );
1130 $r = '<tr><td colspan="2" align="left"><strong>'.$l1."</strong></td>\n" .
1131 '<td colspan="2" align="left"><strong>'.$l2."</strong></td></tr>\n";
1135 function _start_block( $header ) {
1137 $wgOut->addHTML( $header );
1140 function _end_block() {
1143 function _lines( $lines, $prefix=' ', $color="white" ) {
1146 function addedLine( $line ) {
1147 return '<td>+</td><td class="diff-addedline">' .
1151 function deletedLine( $line ) {
1152 return '<td>-</td><td class="diff-deletedline">' .
1156 function emptyLine() {
1157 return '<td colspan="2"> </td>';
1160 function contextLine( $line ) {
1161 return '<td> </td><td class="diff-context">'.$line.'</td>';
1164 function _added($lines) {
1166 foreach ($lines as $line) {
1167 $wgOut->addHTML( '<tr>' . $this->emptyLine() .
1168 $this->addedLine( $line ) . "</tr>\n" );
1172 function _deleted($lines) {
1174 foreach ($lines as $line) {
1175 $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1176 $this->emptyLine() . "</tr>\n" );
1180 function _context( $lines ) {
1182 foreach ($lines as $line) {
1183 $wgOut->addHTML( '<tr>' . $this->contextLine( $line ) .
1184 $this->contextLine( $line ) . "</tr>\n" );
1188 function _changed( $orig, $closing ) {
1190 $diff = new WordLevelDiff( $orig, $closing );
1191 $del = $diff->orig();
1192 $add = $diff->closing();
1194 while ( $line = array_shift( $del ) ) {
1195 $aline = array_shift( $add );
1196 $wgOut->addHTML( '<tr>' . $this->deletedLine( $line ) .
1197 $this->addedLine( $aline ) . "</tr>\n" );
1199 $this->_added( $add ); # If any leftovers