jablonka.czprosek.czf

websvn

Subversion Repositories:
[/] [lib/] [pear/] [Text/] [Diff/] [Engine/] [native.php] - Blame information for rev 5

 

Line No. Rev Author Line
15simandl<?php
2/**
3 * Class used internally by Text_Diff to actually compute the diffs.
4 *
5 * This class is implemented using native PHP code.
6 *
7 * The algorithm used here is mostly lifted from the perl module
8 * Algorithm::Diff (version 1.06) by Ned Konz, which is available at:
9 * http://www.perl.com/CPAN/authors/id/N/NE/NEDKONZ/Algorithm-Diff-1.06.zip
10 *
11 * More ideas are taken from: http://www.ics.uci.edu/~eppstein/161/960229.html
12 *
13 * Some ideas (and a bit of code) are taken from analyze.c, of GNU
14 * diffutils-2.7, which can be found at:
15 * ftp://gnudist.gnu.org/pub/gnu/diffutils/diffutils-2.7.tar.gz
16 *
17 * Some ideas (subdivision by NCHUNKS > 2, and some optimizations) are from
18 * Geoffrey T. Dairiki <dairiki@dairiki.org>. The original PHP version of this
19 * code was written by him, and is used/adapted with his permission.
20 *
21 * $Horde: framework/Text_Diff/Diff/Engine/native.php,v 1.7.2.4 2008/01/04 10:38:10 jan Exp $
22 *
23 * Copyright 2004-2008 The Horde Project (http://www.horde.org/)
24 *
25 * See the enclosed file COPYING for license information (LGPL). If you did
26 * not receive this file, see http://opensource.org/licenses/lgpl-license.php.
27 *
28 * @author Geoffrey T. Dairiki <dairiki@dairiki.org>
29 * @package Text_Diff
30 */
31class Text_Diff_Engine_native {
32 
33 function diff($from_lines, $to_lines)
34 {
35 array_walk($from_lines, array('Text_Diff', 'trimNewlines'));
36 array_walk($to_lines, array('Text_Diff', 'trimNewlines'));
37 
38 $n_from = count($from_lines);
39 $n_to = count($to_lines);
40 
41 $this->xchanged = $this->ychanged = array();
42 $this->xv = $this->yv = array();
43 $this->xind = $this->yind = array();
44 unset($this->seq);
45 unset($this->in_seq);
46 unset($this->lcs);
47 
48 // Skip leading common lines.
49 for ($skip = 0; $skip < $n_from && $skip < $n_to; $skip++) {
50 if ($from_lines[$skip] !== $to_lines[$skip]) {
51 break;
52 }
53 $this->xchanged[$skip] = $this->ychanged[$skip] = false;
54 }
55 
56 // Skip trailing common lines.
57 $xi = $n_from; $yi = $n_to;
58 for ($endskip = 0; --$xi > $skip && --$yi > $skip; $endskip++) {
59 if ($from_lines[$xi] !== $to_lines[$yi]) {
60 break;
61 }
62 $this->xchanged[$xi] = $this->ychanged[$yi] = false;
63 }
64 
65 // Ignore lines which do not exist in both files.
66 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
67 $xhash[$from_lines[$xi]] = 1;
68 }
69 for ($yi = $skip; $yi < $n_to - $endskip; $yi++) {
70 $line = $to_lines[$yi];
71 if (($this->ychanged[$yi] = empty($xhash[$line]))) {
72 continue;
73 }
74 $yhash[$line] = 1;
75 $this->yv[] = $line;
76 $this->yind[] = $yi;
77 }
78 for ($xi = $skip; $xi < $n_from - $endskip; $xi++) {
79 $line = $from_lines[$xi];
80 if (($this->xchanged[$xi] = empty($yhash[$line]))) {
81 continue;
82 }
83 $this->xv[] = $line;
84 $this->xind[] = $xi;
85 }
86 
87 // Find the LCS.
88 $this->_compareseq(0, count($this->xv), 0, count($this->yv));
89 
90 // Merge edits when possible.
91 $this->_shiftBoundaries($from_lines, $this->xchanged, $this->ychanged);
92 $this->_shiftBoundaries($to_lines, $this->ychanged, $this->xchanged);
93 
94 // Compute the edit operations.
95 $edits = array();
96 $xi = $yi = 0;
97 while ($xi < $n_from || $yi < $n_to) {
98 assert($yi < $n_to || $this->xchanged[$xi]);
99 assert($xi < $n_from || $this->ychanged[$yi]);
100 
101 // Skip matching "snake".
102 $copy = array();
103 while ($xi < $n_from && $yi < $n_to
104 && !$this->xchanged[$xi] && !$this->ychanged[$yi]) {
105 $copy[] = $from_lines[$xi++];
106 ++$yi;
107 }
108 if ($copy) {
109 $edits[] = &new Text_Diff_Op_copy($copy);
110 }
111 
112 // Find deletes & adds.
113 $delete = array();
114 while ($xi < $n_from && $this->xchanged[$xi]) {
115 $delete[] = $from_lines[$xi++];
116 }
117 
118 $add = array();
119 while ($yi < $n_to && $this->ychanged[$yi]) {
120 $add[] = $to_lines[$yi++];
121 }
122 
123 if ($delete && $add) {
124 $edits[] = &new Text_Diff_Op_change($delete, $add);
125 } elseif ($delete) {
126 $edits[] = &new Text_Diff_Op_delete($delete);
127 } elseif ($add) {
128 $edits[] = &new Text_Diff_Op_add($add);
129 }
130 }
131 
132 return $edits;
133 }
134 
135 /**
136 * Divides the Largest Common Subsequence (LCS) of the sequences (XOFF,
137 * XLIM) and (YOFF, YLIM) into NCHUNKS approximately equally sized
138 * segments.
139 *
140 * Returns (LCS, PTS). LCS is the length of the LCS. PTS is an array of
141 * NCHUNKS+1 (X, Y) indexes giving the diving points between sub
142 * sequences. The first sub-sequence is contained in (X0, X1), (Y0, Y1),
143 * the second in (X1, X2), (Y1, Y2) and so on. Note that (X0, Y0) ==
144 * (XOFF, YOFF) and (X[NCHUNKS], Y[NCHUNKS]) == (XLIM, YLIM).
145 *
146 * This function assumes that the first lines of the specified portions of
147 * the two files do not match, and likewise that the last lines do not
148 * match. The caller must trim matching lines from the beginning and end
149 * of the portions it is going to specify.
150 */
151 function _diag ($xoff, $xlim, $yoff, $ylim, $nchunks)
152 {
153 $flip = false;
154 
155 if ($xlim - $xoff > $ylim - $yoff) {
156 /* Things seems faster (I'm not sure I understand why) when the
157 * shortest sequence is in X. */
158 $flip = true;
159 list ($xoff, $xlim, $yoff, $ylim)
160 = array($yoff, $ylim, $xoff, $xlim);
161 }
162 
163 if ($flip) {
164 for ($i = $ylim - 1; $i >= $yoff; $i--) {
165 $ymatches[$this->xv[$i]][] = $i;
166 }
167 } else {
168 for ($i = $ylim - 1; $i >= $yoff; $i--) {
169 $ymatches[$this->yv[$i]][] = $i;
170 }
171 }
172 
173 $this->lcs = 0;
174 $this->seq[0]= $yoff - 1;
175 $this->in_seq = array();
176 $ymids[0] = array();
177 
178 $numer = $xlim - $xoff + $nchunks - 1;
179 $x = $xoff;
180 for ($chunk = 0; $chunk < $nchunks; $chunk++) {
181 if ($chunk > 0) {
182 for ($i = 0; $i <= $this->lcs; $i++) {
183 $ymids[$i][$chunk - 1] = $this->seq[$i];
184 }
185 }
186 
187 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $chunk) / $nchunks);
188 for (; $x < $x1; $x++) {
189 $line = $flip ? $this->yv[$x] : $this->xv[$x];
190 if (empty($ymatches[$line])) {
191 continue;
192 }
193 $matches = $ymatches[$line];
194 reset($matches);
195 while (list(, $y) = each($matches)) {
196 if (empty($this->in_seq[$y])) {
197 $k = $this->_lcsPos($y);
198 assert($k > 0);
199 $ymids[$k] = $ymids[$k - 1];
200 break;
201 }
202 }
203 while (list(, $y) = each($matches)) {
204 if ($y > $this->seq[$k - 1]) {
205 assert($y <= $this->seq[$k]);
206 /* Optimization: this is a common case: next match is
207 * just replacing previous match. */
208 $this->in_seq[$this->seq[$k]] = false;
209 $this->seq[$k] = $y;
210 $this->in_seq[$y] = 1;
211 } elseif (empty($this->in_seq[$y])) {
212 $k = $this->_lcsPos($y);
213 assert($k > 0);
214 $ymids[$k] = $ymids[$k - 1];
215 }
216 }
217 }
218 }
219 
220 $seps[] = $flip ? array($yoff, $xoff) : array($xoff, $yoff);
221 $ymid = $ymids[$this->lcs];
222 for ($n = 0; $n < $nchunks - 1; $n++) {
223 $x1 = $xoff + (int)(($numer + ($xlim - $xoff) * $n) / $nchunks);
224 $y1 = $ymid[$n] + 1;
225 $seps[] = $flip ? array($y1, $x1) : array($x1, $y1);
226 }
227 $seps[] = $flip ? array($ylim, $xlim) : array($xlim, $ylim);
228 
229 return array($this->lcs, $seps);
230 }
231 
232 function _lcsPos($ypos)
233 {
234 $end = $this->lcs;
235 if ($end == 0 || $ypos > $this->seq[$end]) {
236 $this->seq[++$this->lcs] = $ypos;
237 $this->in_seq[$ypos] = 1;
238 return $this->lcs;
239 }
240 
241 $beg = 1;
242 while ($beg < $end) {
243 $mid = (int)(($beg + $end) / 2);
244 if ($ypos > $this->seq[$mid]) {
245 $beg = $mid + 1;
246 } else {
247 $end = $mid;
248 }
249 }
250 
251 assert($ypos != $this->seq[$end]);
252 
253 $this->in_seq[$this->seq[$end]] = false;
254 $this->seq[$end] = $ypos;
255 $this->in_seq[$ypos] = 1;
256 return $end;
257 }
258 
259 /**
260 * Finds LCS of two sequences.
261 *
262 * The results are recorded in the vectors $this->{x,y}changed[], by
263 * storing a 1 in the element for each line that is an insertion or
264 * deletion (ie. is not in the LCS).
265 *
266 * The subsequence of file 0 is (XOFF, XLIM) and likewise for file 1.
267 *
268 * Note that XLIM, YLIM are exclusive bounds. All line numbers are
269 * origin-0 and discarded lines are not counted.
270 */
271 function _compareseq ($xoff, $xlim, $yoff, $ylim)
272 {
273 /* Slide down the bottom initial diagonal. */
274 while ($xoff < $xlim && $yoff < $ylim
275 && $this->xv[$xoff] == $this->yv[$yoff]) {
276 ++$xoff;
277 ++$yoff;
278 }
279 
280 /* Slide up the top initial diagonal. */
281 while ($xlim > $xoff && $ylim > $yoff
282 && $this->xv[$xlim - 1] == $this->yv[$ylim - 1]) {
283 --$xlim;
284 --$ylim;
285 }
286 
287 if ($xoff == $xlim || $yoff == $ylim) {
288 $lcs = 0;
289 } else {
290 /* This is ad hoc but seems to work well. $nchunks =
291 * sqrt(min($xlim - $xoff, $ylim - $yoff) / 2.5); $nchunks =
292 * max(2,min(8,(int)$nchunks)); */
293 $nchunks = min(7, $xlim - $xoff, $ylim - $yoff) + 1;
294 list($lcs, $seps)
295 = $this->_diag($xoff, $xlim, $yoff, $ylim, $nchunks);
296 }
297 
298 if ($lcs == 0) {
299 /* X and Y sequences have no common subsequence: mark all
300 * changed. */
301 while ($yoff < $ylim) {
302 $this->ychanged[$this->yind[$yoff++]] = 1;
303 }
304 while ($xoff < $xlim) {
305 $this->xchanged[$this->xind[$xoff++]] = 1;
306 }
307 } else {
308 /* Use the partitions to split this problem into subproblems. */
309 reset($seps);
310 $pt1 = $seps[0];
311 while ($pt2 = next($seps)) {
312 $this->_compareseq ($pt1[0], $pt2[0], $pt1[1], $pt2[1]);
313 $pt1 = $pt2;
314 }
315 }
316 }
317 
318 /**
319 * Adjusts inserts/deletes of identical lines to join changes as much as
320 * possible.
321 *
322 * We do something when a run of changed lines include a line at one end
323 * and has an excluded, identical line at the other. We are free to
324 * choose which identical line is included. `compareseq' usually chooses
325 * the one at the beginning, but usually it is cleaner to consider the
326 * following identical line to be the "change".
327 *
328 * This is extracted verbatim from analyze.c (GNU diffutils-2.7).
329 */
330 function _shiftBoundaries($lines, &$changed, $other_changed)
331 {
332 $i = 0;
333 $j = 0;
334 
335 assert('count($lines) == count($changed)');
336 $len = count($lines);
337 $other_len = count($other_changed);
338 
339 while (1) {
340 /* Scan forward to find the beginning of another run of
341 * changes. Also keep track of the corresponding point in the
342 * other file.
343 *
344 * Throughout this code, $i and $j are adjusted together so that
345 * the first $i elements of $changed and the first $j elements of
346 * $other_changed both contain the same number of zeros (unchanged
347 * lines).
348 *
349 * Furthermore, $j is always kept so that $j == $other_len or
350 * $other_changed[$j] == false. */
351 while ($j < $other_len && $other_changed[$j]) {
352 $j++;
353 }
354 
355 while ($i < $len && ! $changed[$i]) {
356 assert('$j < $other_len && ! $other_changed[$j]');
357 $i++; $j++;
358 while ($j < $other_len && $other_changed[$j]) {
359 $j++;
360 }
361 }
362 
363 if ($i == $len) {
364 break;
365 }
366 
367 $start = $i;
368 
369 /* Find the end of this run of changes. */
370 while (++$i < $len && $changed[$i]) {
371 continue;
372 }
373 
374 do {
375 /* Record the length of this run of changes, so that we can
376 * later determine whether the run has grown. */
377 $runlength = $i - $start;
378 
379 /* Move the changed region back, so long as the previous
380 * unchanged line matches the last changed one. This merges
381 * with previous changed regions. */
382 while ($start > 0 && $lines[$start - 1] == $lines[$i - 1]) {
383 $changed[--$start] = 1;
384 $changed[--$i] = false;
385 while ($start > 0 && $changed[$start - 1]) {
386 $start--;
387 }
388 assert('$j > 0');
389 while ($other_changed[--$j]) {
390 continue;
391 }
392 assert('$j >= 0 && !$other_changed[$j]');
393 }
394 
395 /* Set CORRESPONDING to the end of the changed run, at the
396 * last point where it corresponds to a changed run in the
397 * other file. CORRESPONDING == LEN means no such point has
398 * been found. */
399 $corresponding = $j < $other_len ? $i : $len;
400 
401 /* Move the changed region forward, so long as the first
402 * changed line matches the following unchanged one. This
403 * merges with following changed regions. Do this second, so
404 * that if there are no merges, the changed region is moved
405 * forward as far as possible. */
406 while ($i < $len && $lines[$start] == $lines[$i]) {
407 $changed[$start++] = false;
408 $changed[$i++] = 1;
409 while ($i < $len && $changed[$i]) {
410 $i++;
411 }
412 
413 assert('$j < $other_len && ! $other_changed[$j]');
414 $j++;
415 if ($j < $other_len && $other_changed[$j]) {
416 $corresponding = $i;
417 while ($j < $other_len && $other_changed[$j]) {
418 $j++;
419 }
420 }
421 }
422 } while ($runlength != $i - $start);
423 
424 /* If possible, move the fully-merged run of changes back to a
425 * corresponding run in the other file. */
426 while ($corresponding < $i) {
427 $changed[--$start] = 1;
428 $changed[--$i] = 0;
429 assert('$j > 0');
430 while ($other_changed[--$j]) {
431 continue;
432 }
433 assert('$j >= 0 && !$other_changed[$j]');
434 }
435 }
436 }
437 
438}

Powered by WebSVN 2.2.1