[PATCH] archimport - add merge detection
[git/git.git] / diffcore-rename.c
CommitLineData
427dcb4b
JH
1/*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4#include "cache.h"
5#include "diff.h"
6#include "diffcore.h"
7#include "delta.h"
85976974 8#include "count-delta.h"
427dcb4b 9
25d5ea41
JH
10/* Table of rename/copy destinations */
11
12static struct diff_rename_dst {
13 struct diff_filespec *two;
14 struct diff_filepair *pair;
15} *rename_dst;
16static int rename_dst_nr, rename_dst_alloc;
427dcb4b 17
25d5ea41
JH
18static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two,
19 int insert_ok)
427dcb4b 20{
25d5ea41
JH
21 int first, last;
22
23 first = 0;
24 last = rename_dst_nr;
25 while (last > first) {
26 int next = (last + first) >> 1;
27 struct diff_rename_dst *dst = &(rename_dst[next]);
28 int cmp = strcmp(two->path, dst->two->path);
29 if (!cmp)
30 return dst;
31 if (cmp < 0) {
32 last = next;
33 continue;
34 }
35 first = next+1;
36 }
37 /* not found */
38 if (!insert_ok)
39 return NULL;
40 /* insert to make it at "first" */
41 if (rename_dst_alloc <= rename_dst_nr) {
42 rename_dst_alloc = alloc_nr(rename_dst_alloc);
43 rename_dst = xrealloc(rename_dst,
44 rename_dst_alloc * sizeof(*rename_dst));
45 }
46 rename_dst_nr++;
47 if (first < rename_dst_nr)
48 memmove(rename_dst + first + 1, rename_dst + first,
49 (rename_dst_nr - first - 1) * sizeof(*rename_dst));
50 rename_dst[first].two = two;
51 rename_dst[first].pair = NULL;
52 return &(rename_dst[first]);
427dcb4b
JH
53}
54
15d061b4 55/* Table of rename/copy src files */
25d5ea41
JH
56static struct diff_rename_src {
57 struct diff_filespec *one;
15d061b4 58 unsigned src_stays : 1;
25d5ea41
JH
59} *rename_src;
60static int rename_src_nr, rename_src_alloc;
427dcb4b 61
15d061b4
JH
62static struct diff_rename_src *register_rename_src(struct diff_filespec *one,
63 int src_stays)
25d5ea41
JH
64{
65 int first, last;
66
67 first = 0;
68 last = rename_src_nr;
69 while (last > first) {
70 int next = (last + first) >> 1;
71 struct diff_rename_src *src = &(rename_src[next]);
72 int cmp = strcmp(one->path, src->one->path);
73 if (!cmp)
74 return src;
75 if (cmp < 0) {
76 last = next;
77 continue;
78 }
79 first = next+1;
80 }
15d061b4 81
25d5ea41
JH
82 /* insert to make it at "first" */
83 if (rename_src_alloc <= rename_src_nr) {
84 rename_src_alloc = alloc_nr(rename_src_alloc);
85 rename_src = xrealloc(rename_src,
86 rename_src_alloc * sizeof(*rename_src));
427dcb4b 87 }
25d5ea41
JH
88 rename_src_nr++;
89 if (first < rename_src_nr)
90 memmove(rename_src + first + 1, rename_src + first,
91 (rename_src_nr - first - 1) * sizeof(*rename_src));
92 rename_src[first].one = one;
15d061b4 93 rename_src[first].src_stays = src_stays;
25d5ea41 94 return &(rename_src[first]);
427dcb4b
JH
95}
96
97static int is_exact_match(struct diff_filespec *src, struct diff_filespec *dst)
98{
99 if (src->sha1_valid && dst->sha1_valid &&
100 !memcmp(src->sha1, dst->sha1, 20))
101 return 1;
f0c6b2a2
JH
102 if (diff_populate_filespec(src, 1) || diff_populate_filespec(dst, 1))
103 return 0;
104 if (src->size != dst->size)
105 return 0;
106 if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
427dcb4b
JH
107 return 0;
108 if (src->size == dst->size &&
109 !memcmp(src->data, dst->data, src->size))
110 return 1;
111 return 0;
112}
113
114struct diff_score {
25d5ea41
JH
115 int src; /* index in rename_src */
116 int dst; /* index in rename_dst */
427dcb4b 117 int score;
427dcb4b
JH
118};
119
120static int estimate_similarity(struct diff_filespec *src,
121 struct diff_filespec *dst,
122 int minimum_score)
123{
124 /* src points at a file that existed in the original tree (or
125 * optionally a file in the destination tree) and dst points
126 * at a newly created file. They may be quite similar, in which
127 * case we want to say src is renamed to dst or src is copied into
128 * dst, and then some edit has been applied to dst.
129 *
130 * Compare them and return how similar they are, representing
f0c6b2a2
JH
131 * the score as an integer between 0 and MAX_SCORE.
132 *
133 * When there is an exact match, it is considered a better
134 * match than anything else; the destination does not even
135 * call into this function in that case.
427dcb4b
JH
136 */
137 void *delta;
355e76a4 138 unsigned long delta_size, base_size, src_copied, literal_added;
75c660ac 139 unsigned long delta_limit;
427dcb4b
JH
140 int score;
141
60896c7b
JH
142 /* We deal only with regular files. Symlink renames are handled
143 * only when they are exact matches --- in other words, no edits
144 * after renaming.
145 */
146 if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
147 return 0;
148
427dcb4b
JH
149 delta_size = ((src->size < dst->size) ?
150 (dst->size - src->size) : (src->size - dst->size));
58b103f5 151 base_size = ((src->size < dst->size) ? src->size : dst->size);
427dcb4b 152
58b103f5
JH
153 /* We would not consider edits that change the file size so
154 * drastically. delta_size must be smaller than
cd1870ed 155 * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
f0c6b2a2 156 *
58b103f5
JH
157 * Note that base_size == 0 case is handled here already
158 * and the final score computation below would not have a
159 * divide-by-zero issue.
427dcb4b 160 */
cd1870ed 161 if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
427dcb4b
JH
162 return 0;
163
f0c6b2a2
JH
164 if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
165 return 0; /* error but caught downstream */
166
75c660ac 167 delta_limit = base_size * (MAX_SCORE-minimum_score) / MAX_SCORE;
427dcb4b
JH
168 delta = diff_delta(src->data, src->size,
169 dst->data, dst->size,
75c660ac
JH
170 &delta_size, delta_limit);
171 if (!delta)
172 /* If delta_limit is exceeded, we have too much differences */
173 return 0;
85976974
JH
174
175 /* A delta that has a lot of literal additions would have
176 * big delta_size no matter what else it does.
58b103f5 177 */
a00d7d10 178 if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
85976974
JH
179 return 0;
180
181 /* Estimate the edit size by interpreting delta. */
355e76a4
JH
182 if (count_delta(delta, delta_size, &src_copied, &literal_added)) {
183 free(delta);
85976974 184 return 0;
355e76a4
JH
185 }
186 free(delta);
187
188 /* Extent of damage */
189 if (src->size + literal_added < src_copied)
190 delta_size = 0;
191 else
192 delta_size = (src->size - src_copied) + literal_added;
427dcb4b 193
58b103f5
JH
194 /*
195 * Now we will give some score to it. 100% edit gets 0 points
196 * and 0% edit gets MAX_SCORE points.
427dcb4b 197 */
58b103f5 198 score = MAX_SCORE - (MAX_SCORE * delta_size / base_size);
427dcb4b
JH
199 if (score < 0) return 0;
200 if (MAX_SCORE < score) return MAX_SCORE;
201 return score;
202}
203
25d5ea41
JH
204static void record_rename_pair(struct diff_queue_struct *renq,
205 int dst_index, int src_index, int score)
427dcb4b 206{
25d5ea41
JH
207 struct diff_filespec *one, *two, *src, *dst;
208 struct diff_filepair *dp;
f7c1512a 209
25d5ea41
JH
210 if (rename_dst[dst_index].pair)
211 die("internal error: dst already matched.");
427dcb4b 212
25d5ea41
JH
213 src = rename_src[src_index].one;
214 one = alloc_filespec(src->path);
215 fill_filespec(one, src->sha1, src->mode);
427dcb4b 216
25d5ea41
JH
217 dst = rename_dst[dst_index].two;
218 two = alloc_filespec(dst->path);
219 fill_filespec(two, dst->sha1, dst->mode);
427dcb4b 220
25d5ea41 221 dp = diff_queue(renq, one, two);
01c4e70f 222 dp->score = score;
15d061b4 223 dp->source_stays = rename_src[src_index].src_stays;
25d5ea41 224 rename_dst[dst_index].pair = dp;
427dcb4b
JH
225}
226
227/*
228 * We sort the rename similarity matrix with the score, in descending
15d061b4 229 * order (the most similar first).
427dcb4b
JH
230 */
231static int score_compare(const void *a_, const void *b_)
232{
233 const struct diff_score *a = a_, *b = b_;
234 return b->score - a->score;
235}
236
6b14d7fa 237void diffcore_rename(int detect_rename, int minimum_score)
427dcb4b 238{
38c6f780 239 struct diff_queue_struct *q = &diff_queued_diff;
25d5ea41 240 struct diff_queue_struct renq, outq;
427dcb4b 241 struct diff_score *mx;
25d5ea41
JH
242 int i, j;
243 int num_create, num_src, dst_cnt;
427dcb4b 244
26dee0ad 245 if (!minimum_score)
f345b0a0 246 minimum_score = DEFAULT_RENAME_SCORE;
25d5ea41
JH
247 renq.queue = NULL;
248 renq.nr = renq.alloc = 0;
427dcb4b 249
427dcb4b 250 for (i = 0; i < q->nr; i++) {
52e95789 251 struct diff_filepair *p = q->queue[i];
81e50eab
JH
252 if (!DIFF_FILE_VALID(p->one))
253 if (!DIFF_FILE_VALID(p->two))
60896c7b 254 continue; /* unmerged */
427dcb4b 255 else
25d5ea41 256 locate_rename_dst(p->two, 1);
2210100a
JH
257 else if (!DIFF_FILE_VALID(p->two)) {
258 /* If the source is a broken "delete", and
259 * they did not really want to get broken,
260 * that means the source actually stays.
261 */
262 int stays = (p->broken_pair && !p->score);
263 register_rename_src(p->one, stays);
264 }
15d061b4
JH
265 else if (detect_rename == DIFF_DETECT_COPY)
266 register_rename_src(p->one, 1);
427dcb4b 267 }
25d5ea41 268 if (rename_dst_nr == 0)
427dcb4b
JH
269 goto cleanup; /* nothing to do */
270
271 /* We really want to cull the candidates list early
272 * with cheap tests in order to avoid doing deltas.
427dcb4b 273 */
25d5ea41
JH
274 for (i = 0; i < rename_dst_nr; i++) {
275 struct diff_filespec *two = rename_dst[i].two;
276 for (j = 0; j < rename_src_nr; j++) {
277 struct diff_filespec *one = rename_src[j].one;
278 if (!is_exact_match(one, two))
279 continue;
280 record_rename_pair(&renq, i, j, MAX_SCORE);
281 break; /* we are done with this entry */
427dcb4b
JH
282 }
283 }
25d5ea41 284 diff_debug_queue("done detecting exact", &renq);
427dcb4b
JH
285
286 /* Have we run out the created file pool? If so we can avoid
287 * doing the delta matrix altogether.
288 */
25d5ea41 289 if (renq.nr == rename_dst_nr)
15d061b4 290 goto cleanup;
427dcb4b 291
25d5ea41
JH
292 num_create = (rename_dst_nr - renq.nr);
293 num_src = rename_src_nr;
427dcb4b 294 mx = xmalloc(sizeof(*mx) * num_create * num_src);
25d5ea41 295 for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
427dcb4b 296 int base = dst_cnt * num_src;
25d5ea41
JH
297 struct diff_filespec *two = rename_dst[i].two;
298 if (rename_dst[i].pair)
427dcb4b 299 continue; /* dealt with exact match already. */
25d5ea41
JH
300 for (j = 0; j < rename_src_nr; j++) {
301 struct diff_filespec *one = rename_src[j].one;
302 struct diff_score *m = &mx[base+j];
303 m->src = j;
304 m->dst = i;
305 m->score = estimate_similarity(one, two,
306 minimum_score);
427dcb4b
JH
307 }
308 dst_cnt++;
309 }
310 /* cost matrix sorted by most to least similar pair */
311 qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
312 for (i = 0; i < num_create * num_src; i++) {
25d5ea41
JH
313 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
314 if (dst->pair)
315 continue; /* already done, either exact or fuzzy. */
427dcb4b 316 if (mx[i].score < minimum_score)
15d061b4 317 break; /* there is no more usable pair. */
25d5ea41 318 record_rename_pair(&renq, mx[i].dst, mx[i].src, mx[i].score);
427dcb4b
JH
319 }
320 free(mx);
25d5ea41 321 diff_debug_queue("done detecting fuzzy", &renq);
427dcb4b 322
15d061b4 323 cleanup:
427dcb4b 324 /* At this point, we have found some renames and copies and they
25d5ea41 325 * are kept in renq. The original list is still in *q.
427dcb4b 326 */
25d5ea41
JH
327 outq.queue = NULL;
328 outq.nr = outq.alloc = 0;
427dcb4b 329 for (i = 0; i < q->nr; i++) {
25d5ea41 330 struct diff_filepair *p = q->queue[i];
25d5ea41
JH
331 struct diff_filepair *pair_to_free = NULL;
332
2cd68882
JH
333 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
334 /*
335 * Creation
336 *
337 * We would output this create record if it has
338 * not been turned into a rename/copy already.
339 */
340 struct diff_rename_dst *dst =
341 locate_rename_dst(p->two, 0);
342 if (dst && dst->pair) {
25d5ea41
JH
343 diff_q(&outq, dst->pair);
344 pair_to_free = p;
345 }
346 else
2cd68882
JH
347 /* no matching rename/copy source, so
348 * record this as a creation.
25d5ea41
JH
349 */
350 diff_q(&outq, p);
427dcb4b 351 }
2cd68882
JH
352 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
353 /*
354 * Deletion
355 *
f345b0a0
JH
356 * We would output this delete record if:
357 *
358 * (1) this is a broken delete and the counterpart
359 * broken create remains in the output; or
360 * (2) this is not a broken delete, and renq does
361 * not have a rename/copy to move p->one->path
362 * out.
363 *
364 * Otherwise, the counterpart broken create
365 * has been turned into a rename-edit; or
366 * delete did not have a matching create to
367 * begin with.
2cd68882 368 */
f345b0a0
JH
369 if (DIFF_PAIR_BROKEN(p)) {
370 /* broken delete */
371 struct diff_rename_dst *dst =
372 locate_rename_dst(p->one, 0);
373 if (dst && dst->pair)
374 /* counterpart is now rename/copy */
375 pair_to_free = p;
376 }
377 else {
378 for (j = 0; j < renq.nr; j++)
379 if (!strcmp(renq.queue[j]->one->path,
380 p->one->path))
381 break;
382 if (j < renq.nr)
383 /* this path remains */
384 pair_to_free = p;
385 }
2cd68882
JH
386
387 if (pair_to_free)
388 ;
389 else
390 diff_q(&outq, p);
391 }
25d5ea41 392 else if (!diff_unmodified_pair(p))
15d061b4 393 /* all the usual ones need to be kept */
25d5ea41 394 diff_q(&outq, p);
15d061b4
JH
395 else
396 /* no need to keep unmodified pairs */
397 pair_to_free = p;
398
226406f6
JH
399 if (pair_to_free)
400 diff_free_filepair(pair_to_free);
427dcb4b 401 }
25d5ea41 402 diff_debug_queue("done copying original", &outq);
427dcb4b 403
25d5ea41
JH
404 free(renq.queue);
405 free(q->queue);
406 *q = outq;
407 diff_debug_queue("done collapsing", q);
427dcb4b 408
25d5ea41
JH
409 free(rename_dst);
410 rename_dst = NULL;
411 rename_dst_nr = rename_dst_alloc = 0;
412 free(rename_src);
413 rename_src = NULL;
414 rename_src_nr = rename_src_alloc = 0;
427dcb4b
JH
415 return;
416}