3 * $Horde: framework/Text_Diff/Diff/Engine/native.php,v 1.10 2008/01/04 10:27:53 jan Exp $
5 * Class used internally by Text_Diff to actually compute the diffs. This
6 * class is implemented using native PHP code.
8 * The algorithm used here is mostly lifted from the perl module
9 * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
10 * http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
12 * More ideas are taken from: http://www.ics.uci.edu/~eppstein/161/960229.html
14 * Some ideas (and a bit of code) are taken from analyze.c, of GNU
15 * diffutils-2.7, which can be found at:
16 * ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
18 * Some ideas (subdivision by NCHUNKS > 2, and some optimizations) are from
19 * Geoffrey T. Dairiki <dairiki@dairiki.org>. The original PHP version of this
20 * code was written by him, and is used/adapted with his permission.
22 * Copyright 2004-2008 The Horde Project (http://www.horde.org/)
24 * See the enclosed file COPYING for license information (LGPL). If you did
25 * not receive this file, see http://opensource.org/licenses/lgpl-license.php.
27 * @author Geoffrey T. Dairiki <dairiki@dairiki.org>
30 class Text_Diff_Engine_native {
32 function diff($from_lines, $to_lines)
34 array_walk($from_lines, array('Text_Diff', 'trimNewlines'));
35 array_walk($to_lines, array('Text_Diff', 'trimNewlines'));
37 $n_from = count($from_lines);
38 $n_to = count($to_lines);
40 $this->xchanged = $this->ychanged = array();
41 $this->xv = $this->yv = array();
42 $this->xind = $this->yind = array();
47 // Skip leading common lines.
48 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
49 if ($from_lines[$skip] !== $to_lines[$skip]) {
52 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
55 // Skip trailing common lines.
56 $xi = $n_from; $yi = $n_to;
57 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
58 if ($from_lines[$xi] !== $to_lines[$yi]) {
61 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
64 // Ignore lines which do not exist in both files.
65 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
66 $xhash[$from_lines[$xi]] = 1;
68 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
69 $line = $to_lines[$yi];
70 if (($this->ychanged[$yi] = empty($xhash[$line]))) {
77 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
78 $line = $from_lines[$xi];
79 if (($this->xchanged[$xi] = empty($yhash[$line]))) {
87 $this->_compareseq(0, count($this->xv), 0, count($this->yv));
89 // Merge edits when possible.
90 $this->_shiftBoundaries($from_lines, $this->xchanged, $this->ychanged);
91 $this->_shiftBoundaries($to_lines, $this->ychanged, $this->xchanged);
93 // Compute the edit operations.
96 while ($xi < $n_from || $yi < $n_to) {
97 assert($yi < $n_to || $this->xchanged[$xi]);
98 assert($xi < $n_from || $this->ychanged[$yi]);
100 // Skip matching "snake".
102 while ($xi < $n_from && $yi < $n_to
103 && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
104 $copy[] = $from_lines[$xi++];
108 $edits[] = &new Text_Diff_Op_copy($copy);
111 // Find deletes & adds.
113 while ($xi < $n_from && $this->xchanged[$xi]) {
114 $delete[] = $from_lines[$xi++];
118 while ($yi < $n_to && $this->ychanged[$yi]) {
119 $add[] = $to_lines[$yi++];
122 if ($delete && $add) {
123 $edits[] = &new Text_Diff_Op_change($delete, $add);
125 $edits[] = &new Text_Diff_Op_delete($delete);
127 $edits[] = &new Text_Diff_Op_add($add);
135 * Divides the Largest Common Subsequence (LCS) of the sequences (XOFF,
136 * XLIM) and (YOFF, YLIM) into NCHUNKS approximately equally sized
139 * Returns (LCS, PTS). LCS is the length of the LCS. PTS is an array of
140 * NCHUNKS+1 (X, Y) indexes giving the diving points between sub
141 * sequences. The first sub-sequence is contained in (X0, X1), (Y0, Y1),
142 * the second in (X1, X2), (Y1, Y2) and so on. Note that (X0, Y0) ==
143 * (XOFF, YOFF) and (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
145 * This function assumes that the first lines of the specified portions of
146 * the two files do not match, and likewise that the last lines do not
147 * match. The caller must trim matching lines from the beginning and end
148 * of the portions it is going to specify.
150 function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks)
154 if ($xlim - $xoff > $ylim - $yoff) {
155 /* Things seems faster (I'm not sure I understand why) when the
156 * shortest sequence is in X. */
158 list ($xoff, $xlim, $yoff, $ylim)
159 = array($yoff, $ylim, $xoff, $xlim);
163 for ($i = $ylim - 1; $i >= $yoff; $i--) {
164 $ymatches[$this->xv[$i]][] = $i;
167 for ($i = $ylim - 1; $i >= $yoff; $i--) {
168 $ymatches[$this->yv[$i]][] = $i;
173 $this->seq[0]= $yoff - 1;
174 $this->in_seq = array();
177 $numer = $xlim - $xoff + $nchunks - 1;
179 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
181 for ($i = 0; $i <= $this->lcs; $i++) {
182 $ymids[$i][$chunk - 1] = $this->seq[$i];
186 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $chunk) / $nchunks);
187 for (; $x < $x1; $x++) {
188 $line = $flip ? $this->yv[$x] : $this->xv[$x];
189 if (empty($ymatches[$line])) {
192 $matches = $ymatches[$line];
194 while (list(, $y) = each($matches)) {
195 if (empty($this->in_seq[$y])) {
196 $k = $this->_lcsPos($y);
198 $ymids[$k] = $ymids[$k - 1];
202 while (list(, $y) = each($matches)) {
203 if ($y > $this->seq[$k - 1]) {
204 assert($y <= $this->seq[$k]);
205 /* Optimization: this is a common case: next match is
206 * just replacing previous match. */
207 $this->in_seq[$this->seq[$k]] = false;
209 $this->in_seq[$y] = 1;
210 } elseif (empty($this->in_seq[$y])) {
211 $k = $this->_lcsPos($y);
213 $ymids[$k] = $ymids[$k - 1];
219 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
220 $ymid = $ymids[$this->lcs];
221 for ($n = 0; $n < $nchunks - 1; $n++) {
222 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
224 $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
226 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
228 return array($this->lcs, $seps);
231 function _lcsPos($ypos)
234 if ($end == 0 || $ypos > $this->seq[$end]) {
235 $this->seq[++$this->lcs] = $ypos;
236 $this->in_seq[$ypos] = 1;
241 while ($beg < $end) {
242 $mid = (int)(($beg + $end) / 2);
243 if ($ypos > $this->seq[$mid]) {
250 assert($ypos != $this->seq[$end]);
252 $this->in_seq[$this->seq[$end]] = false;
253 $this->seq[$end] = $ypos;
254 $this->in_seq[$ypos] = 1;
259 * Finds LCS of two sequences.
261 * The results are recorded in the vectors $this->{x,y}changed[], by
262 * storing a 1 in the element for each line that is an insertion or
263 * deletion (ie. is not in the LCS).
265 * The subsequence of file 0 is (XOFF, XLIM) and likewise for file 1.
267 * Note that XLIM, YLIM are exclusive bounds. All line numbers are
268 * origin-0 and discarded lines are not counted.
270 function _compareseq ($xoff, $xlim, $yoff, $ylim)
272 /* Slide down the bottom initial diagonal. */
273 while ($xoff < $xlim && $yoff < $ylim
274 && $this->xv[$xoff] == $this->yv[$yoff]) {
279 /* Slide up the top initial diagonal. */
280 while ($xlim > $xoff && $ylim > $yoff
281 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
286 if ($xoff == $xlim || $yoff == $ylim) {
289 /* This is ad hoc but seems to work well. $nchunks =
290 * sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5); $nchunks =
291 * max(2,min(8,(int)$nchunks)); */
292 $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
294 = $this->_diag($xoff, $xlim, $yoff, $ylim, $nchunks);
298 /* X and Y sequences have no common subsequence: mark all
300 while ($yoff < $ylim) {
301 $this->ychanged[$this->yind[$yoff++]] = 1;
303 while ($xoff < $xlim) {
304 $this->xchanged[$this->xind[$xoff++]] = 1;
307 /* Use the partitions to split this problem into subproblems. */
310 while ($pt2 = next($seps)) {
311 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
318 * Adjusts inserts/deletes of identical lines to join changes as much as
321 * We do something when a run of changed lines include a line at one end
322 * and has an excluded, identical line at the other. We are free to
323 * choose which identical line is included. `compareseq' usually chooses
324 * the one at the beginning, but usually it is cleaner to consider the
325 * following identical line to be the "change".
327 * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
329 function _shiftBoundaries($lines, &$changed, $other_changed)
334 assert('count($lines) == count($changed)');
335 $len = count($lines);
336 $other_len = count($other_changed);
339 /* Scan forward to find the beginning of another run of
340 * changes. Also keep track of the corresponding point in the
343 * Throughout this code, $i and $j are adjusted together so that
344 * the first $i elements of $changed and the first $j elements of
345 * $other_changed both contain the same number of zeros (unchanged
348 * Furthermore, $j is always kept so that $j == $other_len or
349 * $other_changed[$j] == false. */
350 while ($j < $other_len && $other_changed[$j]) {
354 while ($i < $len && ! $changed[$i]) {
355 assert('$j < $other_len && ! $other_changed[$j]');
357 while ($j < $other_len && $other_changed[$j]) {
368 /* Find the end of this run of changes. */
369 while (++$i < $len && $changed[$i]) {
374 /* Record the length of this run of changes, so that we can
375 * later determine whether the run has grown. */
376 $runlength = $i - $start;
378 /* Move the changed region back, so long as the previous
379 * unchanged line matches the last changed one. This merges
380 * with previous changed regions. */
381 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
382 $changed[--$start] = 1;
383 $changed[--$i] = false;
384 while ($start > 0 && $changed[$start - 1]) {
388 while ($other_changed[--$j]) {
391 assert('$j >= 0 && !$other_changed[$j]');
394 /* Set CORRESPONDING to the end of the changed run, at the
395 * last point where it corresponds to a changed run in the
396 * other file. CORRESPONDING == LEN means no such point has
398 $corresponding = $j < $other_len ? $i : $len;
400 /* Move the changed region forward, so long as the first
401 * changed line matches the following unchanged one. This
402 * merges with following changed regions. Do this second, so
403 * that if there are no merges, the changed region is moved
404 * forward as far as possible. */
405 while ($i < $len && $lines[$start] == $lines[$i]) {
406 $changed[$start++] = false;
408 while ($i < $len && $changed[$i]) {
412 assert('$j < $other_len && ! $other_changed[$j]');
414 if ($j < $other_len && $other_changed[$j]) {
416 while ($j < $other_len && $other_changed[$j]) {
421 } while ($runlength != $i - $start);
423 /* If possible, move the fully-merged run of changes back to a
424 * corresponding run in the other file. */
425 while ($corresponding < $i) {
426 $changed[--$start] = 1;
429 while ($other_changed[--$j]) {
432 assert('$j >= 0 && !$other_changed[$j]');