Merge branch 'ew/instaweb'
[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"
427dcb4b 7
25d5ea41
JH
8/* Table of rename/copy destinations */
9
10static struct diff_rename_dst {
11 struct diff_filespec *two;
12 struct diff_filepair *pair;
13} *rename_dst;
14static int rename_dst_nr, rename_dst_alloc;
427dcb4b 15
25d5ea41
JH
16static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two,
17 int insert_ok)
427dcb4b 18{
25d5ea41
JH
19 int first, last;
20
21 first = 0;
22 last = rename_dst_nr;
23 while (last > first) {
24 int next = (last + first) >> 1;
25 struct diff_rename_dst *dst = &(rename_dst[next]);
26 int cmp = strcmp(two->path, dst->two->path);
27 if (!cmp)
28 return dst;
29 if (cmp < 0) {
30 last = next;
31 continue;
32 }
33 first = next+1;
34 }
35 /* not found */
36 if (!insert_ok)
37 return NULL;
38 /* insert to make it at "first" */
39 if (rename_dst_alloc <= rename_dst_nr) {
40 rename_dst_alloc = alloc_nr(rename_dst_alloc);
41 rename_dst = xrealloc(rename_dst,
42 rename_dst_alloc * sizeof(*rename_dst));
43 }
44 rename_dst_nr++;
45 if (first < rename_dst_nr)
46 memmove(rename_dst + first + 1, rename_dst + first,
47 (rename_dst_nr - first - 1) * sizeof(*rename_dst));
5098bafb
JH
48 rename_dst[first].two = alloc_filespec(two->path);
49 fill_filespec(rename_dst[first].two, two->sha1, two->mode);
25d5ea41
JH
50 rename_dst[first].pair = NULL;
51 return &(rename_dst[first]);
427dcb4b
JH
52}
53
15d061b4 54/* Table of rename/copy src files */
25d5ea41
JH
55static struct diff_rename_src {
56 struct diff_filespec *one;
fc580719 57 unsigned short score; /* to remember the break score */
6bac10d8 58 unsigned src_path_left : 1;
25d5ea41
JH
59} *rename_src;
60static int rename_src_nr, rename_src_alloc;
427dcb4b 61
15d061b4 62static struct diff_rename_src *register_rename_src(struct diff_filespec *one,
fc580719
JH
63 int src_path_left,
64 unsigned short score)
25d5ea41
JH
65{
66 int first, last;
67
68 first = 0;
69 last = rename_src_nr;
70 while (last > first) {
71 int next = (last + first) >> 1;
72 struct diff_rename_src *src = &(rename_src[next]);
73 int cmp = strcmp(one->path, src->one->path);
74 if (!cmp)
75 return src;
76 if (cmp < 0) {
77 last = next;
78 continue;
79 }
80 first = next+1;
81 }
15d061b4 82
25d5ea41
JH
83 /* insert to make it at "first" */
84 if (rename_src_alloc <= rename_src_nr) {
85 rename_src_alloc = alloc_nr(rename_src_alloc);
86 rename_src = xrealloc(rename_src,
87 rename_src_alloc * sizeof(*rename_src));
427dcb4b 88 }
25d5ea41
JH
89 rename_src_nr++;
90 if (first < rename_src_nr)
91 memmove(rename_src + first + 1, rename_src + first,
92 (rename_src_nr - first - 1) * sizeof(*rename_src));
93 rename_src[first].one = one;
fc580719 94 rename_src[first].score = score;
6bac10d8 95 rename_src[first].src_path_left = src_path_left;
25d5ea41 96 return &(rename_src[first]);
427dcb4b
JH
97}
98
99static int is_exact_match(struct diff_filespec *src, struct diff_filespec *dst)
100{
101 if (src->sha1_valid && dst->sha1_valid &&
102 !memcmp(src->sha1, dst->sha1, 20))
103 return 1;
f0c6b2a2
JH
104 if (diff_populate_filespec(src, 1) || diff_populate_filespec(dst, 1))
105 return 0;
106 if (src->size != dst->size)
107 return 0;
108 if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
427dcb4b
JH
109 return 0;
110 if (src->size == dst->size &&
111 !memcmp(src->data, dst->data, src->size))
112 return 1;
113 return 0;
114}
115
116struct diff_score {
25d5ea41
JH
117 int src; /* index in rename_src */
118 int dst; /* index in rename_dst */
427dcb4b 119 int score;
427dcb4b
JH
120};
121
122static int estimate_similarity(struct diff_filespec *src,
123 struct diff_filespec *dst,
124 int minimum_score)
125{
126 /* src points at a file that existed in the original tree (or
127 * optionally a file in the destination tree) and dst points
128 * at a newly created file. They may be quite similar, in which
129 * case we want to say src is renamed to dst or src is copied into
130 * dst, and then some edit has been applied to dst.
131 *
132 * Compare them and return how similar they are, representing
f0c6b2a2
JH
133 * the score as an integer between 0 and MAX_SCORE.
134 *
135 * When there is an exact match, it is considered a better
136 * match than anything else; the destination does not even
137 * call into this function in that case.
427dcb4b 138 */
90bd932c 139 unsigned long max_size, delta_size, base_size, src_copied, literal_added;
75c660ac 140 unsigned long delta_limit;
427dcb4b
JH
141 int score;
142
60896c7b
JH
143 /* We deal only with regular files. Symlink renames are handled
144 * only when they are exact matches --- in other words, no edits
145 * after renaming.
146 */
147 if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
148 return 0;
149
90bd932c 150 max_size = ((src->size > dst->size) ? src->size : dst->size);
58b103f5 151 base_size = ((src->size < dst->size) ? src->size : dst->size);
90bd932c 152 delta_size = max_size - base_size;
427dcb4b 153
58b103f5
JH
154 /* We would not consider edits that change the file size so
155 * drastically. delta_size must be smaller than
cd1870ed 156 * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
f0c6b2a2 157 *
58b103f5
JH
158 * Note that base_size == 0 case is handled here already
159 * and the final score computation below would not have a
160 * divide-by-zero issue.
427dcb4b 161 */
cd1870ed 162 if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
427dcb4b
JH
163 return 0;
164
f0c6b2a2
JH
165 if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
166 return 0; /* error but caught downstream */
167
85976974 168
65416758
JH
169 delta_limit = base_size * (MAX_SCORE-minimum_score) / MAX_SCORE;
170 if (diffcore_count_changes(src->data, src->size,
171 dst->data, dst->size,
c06c7966 172 &src->cnt_data, &dst->cnt_data,
65416758
JH
173 delta_limit,
174 &src_copied, &literal_added))
85976974 175 return 0;
355e76a4 176
1706306a
JH
177 /* How similar are they?
178 * what percentage of material in dst are from source?
427dcb4b 179 */
90bd932c 180 if (!dst->size)
1706306a
JH
181 score = 0; /* should not happen */
182 else
90bd932c 183 score = src_copied * MAX_SCORE / max_size;
427dcb4b
JH
184 return score;
185}
186
5098bafb 187static void record_rename_pair(int dst_index, int src_index, int score)
427dcb4b 188{
25d5ea41
JH
189 struct diff_filespec *one, *two, *src, *dst;
190 struct diff_filepair *dp;
f7c1512a 191
25d5ea41
JH
192 if (rename_dst[dst_index].pair)
193 die("internal error: dst already matched.");
427dcb4b 194
25d5ea41
JH
195 src = rename_src[src_index].one;
196 one = alloc_filespec(src->path);
197 fill_filespec(one, src->sha1, src->mode);
427dcb4b 198
25d5ea41
JH
199 dst = rename_dst[dst_index].two;
200 two = alloc_filespec(dst->path);
201 fill_filespec(two, dst->sha1, dst->mode);
427dcb4b 202
5098bafb 203 dp = diff_queue(NULL, one, two);
fc580719
JH
204 if (!strcmp(src->path, dst->path))
205 dp->score = rename_src[src_index].score;
206 else
207 dp->score = score;
6bac10d8 208 dp->source_stays = rename_src[src_index].src_path_left;
25d5ea41 209 rename_dst[dst_index].pair = dp;
427dcb4b
JH
210}
211
212/*
213 * We sort the rename similarity matrix with the score, in descending
15d061b4 214 * order (the most similar first).
427dcb4b
JH
215 */
216static int score_compare(const void *a_, const void *b_)
217{
218 const struct diff_score *a = a_, *b = b_;
219 return b->score - a->score;
220}
221
6bac10d8
JH
222static int compute_stays(struct diff_queue_struct *q,
223 struct diff_filespec *one)
224{
225 int i;
226 for (i = 0; i < q->nr; i++) {
227 struct diff_filepair *p = q->queue[i];
228 if (strcmp(one->path, p->two->path))
229 continue;
230 if (DIFF_PAIR_RENAME(p)) {
231 return 0; /* something else is renamed into this */
232 }
233 }
234 return 1;
235}
236
8082d8d3 237void diffcore_rename(struct diff_options *options)
427dcb4b 238{
8082d8d3
JH
239 int detect_rename = options->detect_rename;
240 int minimum_score = options->rename_score;
241 int rename_limit = options->rename_limit;
38c6f780 242 struct diff_queue_struct *q = &diff_queued_diff;
5098bafb 243 struct diff_queue_struct outq;
427dcb4b 244 struct diff_score *mx;
5098bafb 245 int i, j, rename_count;
25d5ea41 246 int num_create, num_src, dst_cnt;
427dcb4b 247
26dee0ad 248 if (!minimum_score)
f345b0a0 249 minimum_score = DEFAULT_RENAME_SCORE;
5098bafb 250 rename_count = 0;
427dcb4b 251
427dcb4b 252 for (i = 0; i < q->nr; i++) {
52e95789 253 struct diff_filepair *p = q->queue[i];
81e50eab
JH
254 if (!DIFF_FILE_VALID(p->one))
255 if (!DIFF_FILE_VALID(p->two))
60896c7b 256 continue; /* unmerged */
427dcb4b 257 else
25d5ea41 258 locate_rename_dst(p->two, 1);
2210100a
JH
259 else if (!DIFF_FILE_VALID(p->two)) {
260 /* If the source is a broken "delete", and
261 * they did not really want to get broken,
262 * that means the source actually stays.
263 */
264 int stays = (p->broken_pair && !p->score);
fc580719 265 register_rename_src(p->one, stays, p->score);
2210100a 266 }
15d061b4 267 else if (detect_rename == DIFF_DETECT_COPY)
fc580719 268 register_rename_src(p->one, 1, p->score);
427dcb4b 269 }
7d6fb370 270 if (rename_dst_nr == 0 || rename_src_nr == 0 ||
3299c6f6 271 (0 < rename_limit && rename_limit < rename_dst_nr))
427dcb4b
JH
272 goto cleanup; /* nothing to do */
273
274 /* We really want to cull the candidates list early
275 * with cheap tests in order to avoid doing deltas.
427dcb4b 276 */
25d5ea41
JH
277 for (i = 0; i < rename_dst_nr; i++) {
278 struct diff_filespec *two = rename_dst[i].two;
279 for (j = 0; j < rename_src_nr; j++) {
280 struct diff_filespec *one = rename_src[j].one;
281 if (!is_exact_match(one, two))
282 continue;
5098bafb
JH
283 record_rename_pair(i, j, MAX_SCORE);
284 rename_count++;
25d5ea41 285 break; /* we are done with this entry */
427dcb4b
JH
286 }
287 }
427dcb4b
JH
288
289 /* Have we run out the created file pool? If so we can avoid
290 * doing the delta matrix altogether.
291 */
5098bafb 292 if (rename_count == rename_dst_nr)
15d061b4 293 goto cleanup;
427dcb4b 294
9f70b806
JH
295 if (minimum_score == MAX_SCORE)
296 goto cleanup;
297
5098bafb 298 num_create = (rename_dst_nr - rename_count);
25d5ea41 299 num_src = rename_src_nr;
427dcb4b 300 mx = xmalloc(sizeof(*mx) * num_create * num_src);
25d5ea41 301 for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
427dcb4b 302 int base = dst_cnt * num_src;
25d5ea41
JH
303 struct diff_filespec *two = rename_dst[i].two;
304 if (rename_dst[i].pair)
427dcb4b 305 continue; /* dealt with exact match already. */
25d5ea41
JH
306 for (j = 0; j < rename_src_nr; j++) {
307 struct diff_filespec *one = rename_src[j].one;
308 struct diff_score *m = &mx[base+j];
309 m->src = j;
310 m->dst = i;
311 m->score = estimate_similarity(one, two,
312 minimum_score);
427dcb4b 313 }
2821104d
JH
314 /* We do not need the text anymore */
315 diff_free_filespec_data(two);
427dcb4b
JH
316 dst_cnt++;
317 }
318 /* cost matrix sorted by most to least similar pair */
319 qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
320 for (i = 0; i < num_create * num_src; i++) {
25d5ea41
JH
321 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
322 if (dst->pair)
323 continue; /* already done, either exact or fuzzy. */
427dcb4b 324 if (mx[i].score < minimum_score)
15d061b4 325 break; /* there is no more usable pair. */
5098bafb
JH
326 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
327 rename_count++;
427dcb4b
JH
328 }
329 free(mx);
427dcb4b 330
15d061b4 331 cleanup:
427dcb4b 332 /* At this point, we have found some renames and copies and they
5098bafb 333 * are recorded in rename_dst. The original list is still in *q.
427dcb4b 334 */
25d5ea41
JH
335 outq.queue = NULL;
336 outq.nr = outq.alloc = 0;
427dcb4b 337 for (i = 0; i < q->nr; i++) {
25d5ea41 338 struct diff_filepair *p = q->queue[i];
25d5ea41
JH
339 struct diff_filepair *pair_to_free = NULL;
340
2cd68882
JH
341 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
342 /*
343 * Creation
344 *
345 * We would output this create record if it has
346 * not been turned into a rename/copy already.
347 */
348 struct diff_rename_dst *dst =
349 locate_rename_dst(p->two, 0);
350 if (dst && dst->pair) {
25d5ea41
JH
351 diff_q(&outq, dst->pair);
352 pair_to_free = p;
353 }
354 else
2cd68882
JH
355 /* no matching rename/copy source, so
356 * record this as a creation.
25d5ea41
JH
357 */
358 diff_q(&outq, p);
427dcb4b 359 }
2cd68882
JH
360 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
361 /*
362 * Deletion
363 *
f345b0a0
JH
364 * We would output this delete record if:
365 *
366 * (1) this is a broken delete and the counterpart
367 * broken create remains in the output; or
5098bafb
JH
368 * (2) this is not a broken delete, and rename_dst
369 * does not have a rename/copy to move p->one->path
370 * out of existence.
f345b0a0
JH
371 *
372 * Otherwise, the counterpart broken create
373 * has been turned into a rename-edit; or
374 * delete did not have a matching create to
375 * begin with.
2cd68882 376 */
f345b0a0
JH
377 if (DIFF_PAIR_BROKEN(p)) {
378 /* broken delete */
379 struct diff_rename_dst *dst =
380 locate_rename_dst(p->one, 0);
381 if (dst && dst->pair)
382 /* counterpart is now rename/copy */
383 pair_to_free = p;
384 }
385 else {
5098bafb
JH
386 for (j = 0; j < rename_dst_nr; j++) {
387 if (!rename_dst[j].pair)
388 continue;
389 if (strcmp(rename_dst[j].pair->
390 one->path,
391 p->one->path))
392 continue;
393 break;
394 }
395 if (j < rename_dst_nr)
f345b0a0
JH
396 /* this path remains */
397 pair_to_free = p;
398 }
2cd68882
JH
399
400 if (pair_to_free)
401 ;
402 else
403 diff_q(&outq, p);
404 }
25d5ea41 405 else if (!diff_unmodified_pair(p))
15d061b4 406 /* all the usual ones need to be kept */
25d5ea41 407 diff_q(&outq, p);
15d061b4
JH
408 else
409 /* no need to keep unmodified pairs */
410 pair_to_free = p;
411
226406f6
JH
412 if (pair_to_free)
413 diff_free_filepair(pair_to_free);
427dcb4b 414 }
25d5ea41 415 diff_debug_queue("done copying original", &outq);
427dcb4b 416
25d5ea41
JH
417 free(q->queue);
418 *q = outq;
419 diff_debug_queue("done collapsing", q);
427dcb4b 420
6bac10d8
JH
421 /* We need to see which rename source really stays here;
422 * earlier we only checked if the path is left in the result,
423 * but even if a path remains in the result, if that is coming
424 * from copying something else on top of it, then the original
425 * source is lost and does not stay.
426 */
427 for (i = 0; i < q->nr; i++) {
428 struct diff_filepair *p = q->queue[i];
429 if (DIFF_PAIR_RENAME(p) && p->source_stays) {
430 /* If one appears as the target of a rename-copy,
431 * then mark p->source_stays = 0; otherwise
432 * leave it as is.
433 */
434 p->source_stays = compute_stays(q, p->one);
435 }
436 }
437
5098bafb
JH
438 for (i = 0; i < rename_dst_nr; i++) {
439 diff_free_filespec_data(rename_dst[i].two);
440 free(rename_dst[i].two);
441 }
442
25d5ea41
JH
443 free(rename_dst);
444 rename_dst = NULL;
445 rename_dst_nr = rename_dst_alloc = 0;
446 free(rename_src);
447 rename_src = NULL;
448 rename_src_nr = rename_src_alloc = 0;
427dcb4b
JH
449 return;
450}