]>
Commit | Line | Data |
---|---|---|
427dcb4b JH |
1 | /* |
2 | * Copyright (C) 2005 Junio C Hamano | |
3 | */ | |
4 | #include "cache.h" | |
5 | #include "diff.h" | |
6 | #include "diffcore.h" | |
cbd53a21 | 7 | #include "object-store.h" |
f79d9c58 | 8 | #include "hashmap.h" |
3ac942d4 | 9 | #include "progress.h" |
427dcb4b | 10 | |
25d5ea41 JH |
11 | /* Table of rename/copy destinations */ |
12 | ||
13 | static struct diff_rename_dst { | |
14 | struct diff_filespec *two; | |
15 | struct diff_filepair *pair; | |
16 | } *rename_dst; | |
17 | static int rename_dst_nr, rename_dst_alloc; | |
427dcb4b | 18 | |
f98c2f7e | 19 | static int find_rename_dst(struct diff_filespec *two) |
427dcb4b | 20 | { |
25d5ea41 JH |
21 | int first, last; |
22 | ||
23 | first = 0; | |
24 | last = rename_dst_nr; | |
25 | while (last > first) { | |
568a05c5 | 26 | int next = first + ((last - first) >> 1); |
25d5ea41 JH |
27 | struct diff_rename_dst *dst = &(rename_dst[next]); |
28 | int cmp = strcmp(two->path, dst->two->path); | |
29 | if (!cmp) | |
f98c2f7e | 30 | return next; |
25d5ea41 JH |
31 | if (cmp < 0) { |
32 | last = next; | |
33 | continue; | |
34 | } | |
35 | first = next+1; | |
36 | } | |
f98c2f7e JK |
37 | return -first - 1; |
38 | } | |
39 | ||
40 | static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two) | |
41 | { | |
42 | int ofs = find_rename_dst(two); | |
43 | return ofs < 0 ? NULL : &rename_dst[ofs]; | |
44 | } | |
45 | ||
46 | /* | |
47 | * Returns 0 on success, -1 if we found a duplicate. | |
48 | */ | |
49 | static int add_rename_dst(struct diff_filespec *two) | |
50 | { | |
51 | int first = find_rename_dst(two); | |
52 | ||
53 | if (first >= 0) | |
54 | return -1; | |
55 | first = -first - 1; | |
56 | ||
25d5ea41 | 57 | /* insert to make it at "first" */ |
337ce247 | 58 | ALLOC_GROW(rename_dst, rename_dst_nr + 1, rename_dst_alloc); |
25d5ea41 JH |
59 | rename_dst_nr++; |
60 | if (first < rename_dst_nr) | |
f919ffeb SG |
61 | MOVE_ARRAY(rename_dst + first + 1, rename_dst + first, |
62 | rename_dst_nr - first - 1); | |
5098bafb | 63 | rename_dst[first].two = alloc_filespec(two->path); |
f9704c2d | 64 | fill_filespec(rename_dst[first].two, &two->oid, two->oid_valid, |
a0d12c44 | 65 | two->mode); |
25d5ea41 | 66 | rename_dst[first].pair = NULL; |
f98c2f7e | 67 | return 0; |
427dcb4b JH |
68 | } |
69 | ||
15d061b4 | 70 | /* Table of rename/copy src files */ |
25d5ea41 | 71 | static struct diff_rename_src { |
e88d6bc6 | 72 | struct diff_filepair *p; |
fc580719 | 73 | unsigned short score; /* to remember the break score */ |
25d5ea41 JH |
74 | } *rename_src; |
75 | static int rename_src_nr, rename_src_alloc; | |
427dcb4b | 76 | |
e88d6bc6 | 77 | static struct diff_rename_src *register_rename_src(struct diff_filepair *p) |
25d5ea41 JH |
78 | { |
79 | int first, last; | |
e88d6bc6 JH |
80 | struct diff_filespec *one = p->one; |
81 | unsigned short score = p->score; | |
25d5ea41 JH |
82 | |
83 | first = 0; | |
84 | last = rename_src_nr; | |
85 | while (last > first) { | |
568a05c5 | 86 | int next = first + ((last - first) >> 1); |
25d5ea41 | 87 | struct diff_rename_src *src = &(rename_src[next]); |
e88d6bc6 | 88 | int cmp = strcmp(one->path, src->p->one->path); |
25d5ea41 JH |
89 | if (!cmp) |
90 | return src; | |
91 | if (cmp < 0) { | |
92 | last = next; | |
93 | continue; | |
94 | } | |
95 | first = next+1; | |
96 | } | |
15d061b4 | 97 | |
25d5ea41 | 98 | /* insert to make it at "first" */ |
337ce247 | 99 | ALLOC_GROW(rename_src, rename_src_nr + 1, rename_src_alloc); |
25d5ea41 JH |
100 | rename_src_nr++; |
101 | if (first < rename_src_nr) | |
f919ffeb SG |
102 | MOVE_ARRAY(rename_src + first + 1, rename_src + first, |
103 | rename_src_nr - first - 1); | |
e88d6bc6 | 104 | rename_src[first].p = p; |
fc580719 | 105 | rename_src[first].score = score; |
25d5ea41 | 106 | return &(rename_src[first]); |
427dcb4b JH |
107 | } |
108 | ||
0ce39643 JS |
109 | static int basename_same(struct diff_filespec *src, struct diff_filespec *dst) |
110 | { | |
111 | int src_len = strlen(src->path), dst_len = strlen(dst->path); | |
112 | while (src_len && dst_len) { | |
113 | char c1 = src->path[--src_len]; | |
114 | char c2 = dst->path[--dst_len]; | |
115 | if (c1 != c2) | |
116 | return 0; | |
117 | if (c1 == '/') | |
118 | return 1; | |
119 | } | |
120 | return (!src_len || src->path[src_len - 1] == '/') && | |
121 | (!dst_len || dst->path[dst_len - 1] == '/'); | |
122 | } | |
123 | ||
427dcb4b | 124 | struct diff_score { |
25d5ea41 JH |
125 | int src; /* index in rename_src */ |
126 | int dst; /* index in rename_dst */ | |
6d24ad97 JH |
127 | unsigned short score; |
128 | short name_score; | |
427dcb4b JH |
129 | }; |
130 | ||
b78ea5fc NTND |
131 | static int estimate_similarity(struct repository *r, |
132 | struct diff_filespec *src, | |
427dcb4b JH |
133 | struct diff_filespec *dst, |
134 | int minimum_score) | |
135 | { | |
136 | /* src points at a file that existed in the original tree (or | |
137 | * optionally a file in the destination tree) and dst points | |
138 | * at a newly created file. They may be quite similar, in which | |
139 | * case we want to say src is renamed to dst or src is copied into | |
140 | * dst, and then some edit has been applied to dst. | |
141 | * | |
142 | * Compare them and return how similar they are, representing | |
f0c6b2a2 JH |
143 | * the score as an integer between 0 and MAX_SCORE. |
144 | * | |
145 | * When there is an exact match, it is considered a better | |
146 | * match than anything else; the destination does not even | |
147 | * call into this function in that case. | |
427dcb4b | 148 | */ |
90bd932c | 149 | unsigned long max_size, delta_size, base_size, src_copied, literal_added; |
427dcb4b JH |
150 | int score; |
151 | ||
60896c7b JH |
152 | /* We deal only with regular files. Symlink renames are handled |
153 | * only when they are exact matches --- in other words, no edits | |
154 | * after renaming. | |
155 | */ | |
156 | if (!S_ISREG(src->mode) || !S_ISREG(dst->mode)) | |
157 | return 0; | |
158 | ||
81ac051d LT |
159 | /* |
160 | * Need to check that source and destination sizes are | |
161 | * filled in before comparing them. | |
162 | * | |
163 | * If we already have "cnt_data" filled in, we know it's | |
164 | * all good (avoid checking the size for zero, as that | |
165 | * is a possible size - we really should have a flag to | |
166 | * say whether the size is valid or not!) | |
167 | */ | |
8e5dd3d6 | 168 | if (!src->cnt_data && |
b78ea5fc | 169 | diff_populate_filespec(r, src, CHECK_SIZE_ONLY)) |
81ac051d | 170 | return 0; |
8e5dd3d6 | 171 | if (!dst->cnt_data && |
b78ea5fc | 172 | diff_populate_filespec(r, dst, CHECK_SIZE_ONLY)) |
81ac051d LT |
173 | return 0; |
174 | ||
90bd932c | 175 | max_size = ((src->size > dst->size) ? src->size : dst->size); |
58b103f5 | 176 | base_size = ((src->size < dst->size) ? src->size : dst->size); |
90bd932c | 177 | delta_size = max_size - base_size; |
427dcb4b | 178 | |
58b103f5 JH |
179 | /* We would not consider edits that change the file size so |
180 | * drastically. delta_size must be smaller than | |
cd1870ed | 181 | * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size). |
f0c6b2a2 | 182 | * |
58b103f5 JH |
183 | * Note that base_size == 0 case is handled here already |
184 | * and the final score computation below would not have a | |
185 | * divide-by-zero issue. | |
427dcb4b | 186 | */ |
3a4d6769 | 187 | if (max_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE) |
427dcb4b JH |
188 | return 0; |
189 | ||
b78ea5fc | 190 | if (!src->cnt_data && diff_populate_filespec(r, src, 0)) |
885c716f | 191 | return 0; |
b78ea5fc | 192 | if (!dst->cnt_data && diff_populate_filespec(r, dst, 0)) |
885c716f BS |
193 | return 0; |
194 | ||
b78ea5fc | 195 | if (diffcore_count_changes(r, src, dst, |
c06c7966 | 196 | &src->cnt_data, &dst->cnt_data, |
65416758 | 197 | &src_copied, &literal_added)) |
85976974 | 198 | return 0; |
355e76a4 | 199 | |
1706306a JH |
200 | /* How similar are they? |
201 | * what percentage of material in dst are from source? | |
427dcb4b | 202 | */ |
90bd932c | 203 | if (!dst->size) |
1706306a | 204 | score = 0; /* should not happen */ |
cfc0aef1 | 205 | else |
dc49cd76 | 206 | score = (int)(src_copied * MAX_SCORE / max_size); |
427dcb4b JH |
207 | return score; |
208 | } | |
209 | ||
5098bafb | 210 | static void record_rename_pair(int dst_index, int src_index, int score) |
427dcb4b | 211 | { |
9fb88419 | 212 | struct diff_filespec *src, *dst; |
25d5ea41 | 213 | struct diff_filepair *dp; |
f7c1512a | 214 | |
25d5ea41 JH |
215 | if (rename_dst[dst_index].pair) |
216 | die("internal error: dst already matched."); | |
427dcb4b | 217 | |
e88d6bc6 | 218 | src = rename_src[src_index].p->one; |
64479711 | 219 | src->rename_used++; |
9fb88419 | 220 | src->count++; |
427dcb4b | 221 | |
25d5ea41 | 222 | dst = rename_dst[dst_index].two; |
9fb88419 | 223 | dst->count++; |
427dcb4b | 224 | |
9fb88419 | 225 | dp = diff_queue(NULL, src, dst); |
ef677686 | 226 | dp->renamed_pair = 1; |
fc580719 JH |
227 | if (!strcmp(src->path, dst->path)) |
228 | dp->score = rename_src[src_index].score; | |
229 | else | |
230 | dp->score = score; | |
25d5ea41 | 231 | rename_dst[dst_index].pair = dp; |
427dcb4b JH |
232 | } |
233 | ||
234 | /* | |
235 | * We sort the rename similarity matrix with the score, in descending | |
15d061b4 | 236 | * order (the most similar first). |
427dcb4b JH |
237 | */ |
238 | static int score_compare(const void *a_, const void *b_) | |
239 | { | |
240 | const struct diff_score *a = a_, *b = b_; | |
cfc0aef1 | 241 | |
6d24ad97 JH |
242 | /* sink the unused ones to the bottom */ |
243 | if (a->dst < 0) | |
244 | return (0 <= b->dst); | |
245 | else if (b->dst < 0) | |
246 | return -1; | |
247 | ||
cfc0aef1 RS |
248 | if (a->score == b->score) |
249 | return b->name_score - a->name_score; | |
250 | ||
427dcb4b JH |
251 | return b->score - a->score; |
252 | } | |
253 | ||
9027f53c | 254 | struct file_similarity { |
f79d9c58 | 255 | struct hashmap_entry entry; |
7c85f8ac | 256 | int index; |
9027f53c | 257 | struct diff_filespec *filespec; |
9027f53c LT |
258 | }; |
259 | ||
b78ea5fc NTND |
260 | static unsigned int hash_filespec(struct repository *r, |
261 | struct diff_filespec *filespec) | |
48f6407f | 262 | { |
41c9560e | 263 | if (!filespec->oid_valid) { |
b78ea5fc | 264 | if (diff_populate_filespec(r, filespec, 0)) |
48f6407f | 265 | return 0; |
f070facc PO |
266 | hash_object_file(filespec->data, filespec->size, "blob", |
267 | &filespec->oid); | |
48f6407f | 268 | } |
d40abc8e | 269 | return oidhash(&filespec->oid); |
48f6407f KB |
270 | } |
271 | ||
f79d9c58 | 272 | static int find_identical_files(struct hashmap *srcs, |
7c85f8ac | 273 | int dst_index, |
11f944dd | 274 | struct diff_options *options) |
9027f53c LT |
275 | { |
276 | int renames = 0; | |
6bcbdfb2 | 277 | struct hashmap_entry *ent; |
7c85f8ac | 278 | struct diff_filespec *target = rename_dst[dst_index].two; |
ab73a9d1 | 279 | struct file_similarity *p, *best = NULL; |
48f6407f KB |
280 | int i = 100, best_score = -1; |
281 | ||
282 | /* | |
7c85f8ac | 283 | * Find the best source match for specified destination. |
48f6407f | 284 | */ |
6bcbdfb2 | 285 | ent = hashmap_get_from_hash(srcs, |
b78ea5fc NTND |
286 | hash_filespec(options->repo, target), |
287 | NULL); | |
6bcbdfb2 | 288 | for (; ent; ent = hashmap_get_next(srcs, ent)) { |
48f6407f | 289 | int score; |
6bcbdfb2 EW |
290 | struct diff_filespec *source; |
291 | ||
292 | p = container_of(ent, struct file_similarity, entry); | |
293 | source = p->filespec; | |
48f6407f KB |
294 | |
295 | /* False hash collision? */ | |
9001dc2a | 296 | if (!oideq(&source->oid, &target->oid)) |
48f6407f KB |
297 | continue; |
298 | /* Non-regular files? If so, the modes must match! */ | |
299 | if (!S_ISREG(source->mode) || !S_ISREG(target->mode)) { | |
300 | if (source->mode != target->mode) | |
0940e5f2 | 301 | continue; |
9027f53c | 302 | } |
48f6407f KB |
303 | /* Give higher scores to sources that haven't been used already */ |
304 | score = !source->rename_used; | |
305 | if (source->rename_used && options->detect_rename != DIFF_DETECT_COPY) | |
306 | continue; | |
307 | score += basename_same(source, target); | |
308 | if (score > best_score) { | |
309 | best = p; | |
310 | best_score = score; | |
311 | if (score == 2) | |
312 | break; | |
9027f53c | 313 | } |
48f6407f KB |
314 | |
315 | /* Too many identical alternatives? Pick one */ | |
316 | if (!--i) | |
317 | break; | |
318 | } | |
319 | if (best) { | |
7c85f8ac | 320 | record_rename_pair(dst_index, best->index, MAX_SCORE); |
48f6407f KB |
321 | renames++; |
322 | } | |
9027f53c LT |
323 | return renames; |
324 | } | |
325 | ||
b78ea5fc NTND |
326 | static void insert_file_table(struct repository *r, |
327 | struct hashmap *table, int index, | |
328 | struct diff_filespec *filespec) | |
9027f53c | 329 | { |
9027f53c LT |
330 | struct file_similarity *entry = xmalloc(sizeof(*entry)); |
331 | ||
9027f53c LT |
332 | entry->index = index; |
333 | entry->filespec = filespec; | |
9027f53c | 334 | |
d22245a2 | 335 | hashmap_entry_init(&entry->entry, hash_filespec(r, filespec)); |
b94e5c1d | 336 | hashmap_add(table, &entry->entry); |
9027f53c LT |
337 | } |
338 | ||
cb1491b6 LT |
339 | /* |
340 | * Find exact renames first. | |
341 | * | |
342 | * The first round matches up the up-to-date entries, | |
343 | * and then during the second round we try to match | |
344 | * cache-dirty entries as well. | |
cb1491b6 | 345 | */ |
11f944dd | 346 | static int find_exact_renames(struct diff_options *options) |
cb1491b6 | 347 | { |
7c85f8ac | 348 | int i, renames = 0; |
f79d9c58 | 349 | struct hashmap file_table; |
cb1491b6 | 350 | |
ca4e3ca0 SG |
351 | /* Add all sources to the hash table in reverse order, because |
352 | * later on they will be retrieved in LIFO order. | |
353 | */ | |
7663cdc8 | 354 | hashmap_init(&file_table, NULL, NULL, rename_src_nr); |
ca4e3ca0 | 355 | for (i = rename_src_nr-1; i >= 0; i--) |
b78ea5fc NTND |
356 | insert_file_table(options->repo, |
357 | &file_table, i, | |
358 | rename_src[i].p->one); | |
9027f53c | 359 | |
7c85f8ac | 360 | /* Walk the destinations and find best source match */ |
9027f53c | 361 | for (i = 0; i < rename_dst_nr; i++) |
7c85f8ac | 362 | renames += find_identical_files(&file_table, i, options); |
9027f53c | 363 | |
f79d9c58 KB |
364 | /* Free the hash data structure and entries */ |
365 | hashmap_free(&file_table, 1); | |
9027f53c | 366 | |
7c85f8ac | 367 | return renames; |
cb1491b6 LT |
368 | } |
369 | ||
6d24ad97 JH |
370 | #define NUM_CANDIDATE_PER_DST 4 |
371 | static void record_if_better(struct diff_score m[], struct diff_score *o) | |
372 | { | |
373 | int i, worst; | |
374 | ||
375 | /* find the worst one */ | |
376 | worst = 0; | |
377 | for (i = 1; i < NUM_CANDIDATE_PER_DST; i++) | |
378 | if (score_compare(&m[i], &m[worst]) > 0) | |
379 | worst = i; | |
380 | ||
381 | /* is it better than the worst one? */ | |
382 | if (score_compare(&m[worst], o) > 0) | |
383 | m[worst] = *o; | |
384 | } | |
385 | ||
f31027c9 JH |
386 | /* |
387 | * Returns: | |
388 | * 0 if we are under the limit; | |
389 | * 1 if we need to disable inexact rename detection; | |
390 | * 2 if we would be under the limit if we were given -C instead of -C -C. | |
391 | */ | |
9d8a5a50 JH |
392 | static int too_many_rename_candidates(int num_create, |
393 | struct diff_options *options) | |
394 | { | |
395 | int rename_limit = options->rename_limit; | |
396 | int num_src = rename_src_nr; | |
f31027c9 | 397 | int i; |
9d8a5a50 JH |
398 | |
399 | options->needed_rename_limit = 0; | |
400 | ||
401 | /* | |
402 | * This basically does a test for the rename matrix not | |
403 | * growing larger than a "rename_limit" square matrix, ie: | |
404 | * | |
405 | * num_create * num_src > rename_limit * rename_limit | |
9d8a5a50 | 406 | */ |
89973554 JT |
407 | if (rename_limit <= 0) |
408 | rename_limit = 32767; | |
9d8a5a50 | 409 | if ((num_create <= rename_limit || num_src <= rename_limit) && |
9f7e4bfa EN |
410 | ((uint64_t)num_create * (uint64_t)num_src |
411 | <= (uint64_t)rename_limit * (uint64_t)rename_limit)) | |
9d8a5a50 JH |
412 | return 0; |
413 | ||
414 | options->needed_rename_limit = | |
415 | num_src > num_create ? num_src : num_create; | |
f31027c9 JH |
416 | |
417 | /* Are we running under -C -C? */ | |
0d1e0e78 | 418 | if (!options->flags.find_copies_harder) |
f31027c9 JH |
419 | return 1; |
420 | ||
421 | /* Would we bust the limit if we were running under -C? */ | |
422 | for (num_src = i = 0; i < rename_src_nr; i++) { | |
423 | if (diff_unmodified_pair(rename_src[i].p)) | |
424 | continue; | |
425 | num_src++; | |
426 | } | |
427 | if ((num_create <= rename_limit || num_src <= rename_limit) && | |
9f7e4bfa EN |
428 | ((uint64_t)num_create * (uint64_t)num_src |
429 | <= (uint64_t)rename_limit * (uint64_t)rename_limit)) | |
f31027c9 | 430 | return 2; |
9d8a5a50 JH |
431 | return 1; |
432 | } | |
433 | ||
0940e5f2 LT |
434 | static int find_renames(struct diff_score *mx, int dst_cnt, int minimum_score, int copies) |
435 | { | |
436 | int count = 0, i; | |
437 | ||
438 | for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) { | |
439 | struct diff_rename_dst *dst; | |
440 | ||
441 | if ((mx[i].dst < 0) || | |
442 | (mx[i].score < minimum_score)) | |
443 | break; /* there is no more usable pair. */ | |
444 | dst = &rename_dst[mx[i].dst]; | |
445 | if (dst->pair) | |
446 | continue; /* already done, either exact or fuzzy. */ | |
e88d6bc6 | 447 | if (!copies && rename_src[mx[i].src].p->one->rename_used) |
0940e5f2 LT |
448 | continue; |
449 | record_rename_pair(mx[i].dst, mx[i].src, mx[i].score); | |
450 | count++; | |
451 | } | |
452 | return count; | |
453 | } | |
454 | ||
8082d8d3 | 455 | void diffcore_rename(struct diff_options *options) |
427dcb4b | 456 | { |
8082d8d3 JH |
457 | int detect_rename = options->detect_rename; |
458 | int minimum_score = options->rename_score; | |
38c6f780 | 459 | struct diff_queue_struct *q = &diff_queued_diff; |
5098bafb | 460 | struct diff_queue_struct outq; |
427dcb4b | 461 | struct diff_score *mx; |
f31027c9 | 462 | int i, j, rename_count, skip_unmodified = 0; |
dabdbee1 | 463 | int num_create, dst_cnt; |
3ac942d4 | 464 | struct progress *progress = NULL; |
427dcb4b | 465 | |
26dee0ad | 466 | if (!minimum_score) |
f345b0a0 | 467 | minimum_score = DEFAULT_RENAME_SCORE; |
427dcb4b | 468 | |
427dcb4b | 469 | for (i = 0; i < q->nr; i++) { |
52e95789 | 470 | struct diff_filepair *p = q->queue[i]; |
2f3f8b21 | 471 | if (!DIFF_FILE_VALID(p->one)) { |
81e50eab | 472 | if (!DIFF_FILE_VALID(p->two)) |
60896c7b | 473 | continue; /* unmerged */ |
2f3f8b21 JH |
474 | else if (options->single_follow && |
475 | strcmp(options->single_follow, p->two->path)) | |
476 | continue; /* not interested */ | |
0d1e0e78 | 477 | else if (!options->flags.rename_empty && |
02491b67 | 478 | is_empty_blob_oid(&p->two->oid)) |
90d43b07 | 479 | continue; |
4d6be03b JK |
480 | else if (add_rename_dst(p->two) < 0) { |
481 | warning("skipping rename detection, detected" | |
482 | " duplicate destination '%s'", | |
483 | p->two->path); | |
484 | goto cleanup; | |
485 | } | |
2f3f8b21 | 486 | } |
0d1e0e78 | 487 | else if (!options->flags.rename_empty && |
02491b67 | 488 | is_empty_blob_oid(&p->one->oid)) |
90d43b07 | 489 | continue; |
d7c9bf22 | 490 | else if (!DIFF_PAIR_UNMERGED(p) && !DIFF_FILE_VALID(p->two)) { |
64479711 LT |
491 | /* |
492 | * If the source is a broken "delete", and | |
2210100a JH |
493 | * they did not really want to get broken, |
494 | * that means the source actually stays. | |
64479711 LT |
495 | * So we increment the "rename_used" score |
496 | * by one, to indicate ourselves as a user | |
497 | */ | |
498 | if (p->broken_pair && !p->score) | |
499 | p->one->rename_used++; | |
e88d6bc6 | 500 | register_rename_src(p); |
64479711 LT |
501 | } |
502 | else if (detect_rename == DIFF_DETECT_COPY) { | |
503 | /* | |
504 | * Increment the "rename_used" score by | |
505 | * one, to indicate ourselves as a user. | |
2210100a | 506 | */ |
64479711 | 507 | p->one->rename_used++; |
e88d6bc6 | 508 | register_rename_src(p); |
2210100a | 509 | } |
427dcb4b | 510 | } |
0024a549 | 511 | if (rename_dst_nr == 0 || rename_src_nr == 0) |
427dcb4b JH |
512 | goto cleanup; /* nothing to do */ |
513 | ||
17559a64 LT |
514 | /* |
515 | * We really want to cull the candidates list early | |
516 | * with cheap tests in order to avoid doing deltas. | |
517 | */ | |
11f944dd | 518 | rename_count = find_exact_renames(options); |
17559a64 | 519 | |
42899ac8 LT |
520 | /* Did we only want exact renames? */ |
521 | if (minimum_score == MAX_SCORE) | |
522 | goto cleanup; | |
523 | ||
524 | /* | |
525 | * Calculate how many renames are left (but all the source | |
526 | * files still remain as options for rename/copies!) | |
527 | */ | |
528 | num_create = (rename_dst_nr - rename_count); | |
42899ac8 LT |
529 | |
530 | /* All done? */ | |
531 | if (!num_create) | |
532 | goto cleanup; | |
533 | ||
f31027c9 JH |
534 | switch (too_many_rename_candidates(num_create, options)) { |
535 | case 1: | |
0024a549 | 536 | goto cleanup; |
f31027c9 JH |
537 | case 2: |
538 | options->degraded_cc_to_c = 1; | |
539 | skip_unmodified = 1; | |
540 | break; | |
541 | default: | |
542 | break; | |
543 | } | |
0024a549 | 544 | |
3ac942d4 | 545 | if (options->show_rename_progress) { |
8aade107 | 546 | progress = start_delayed_progress( |
754dbc43 | 547 | _("Performing inexact rename detection"), |
d6861d02 | 548 | (uint64_t)rename_dst_nr * (uint64_t)rename_src_nr); |
3ac942d4 JK |
549 | } |
550 | ||
50492f7b | 551 | mx = xcalloc(st_mult(NUM_CANDIDATE_PER_DST, num_create), sizeof(*mx)); |
25d5ea41 | 552 | for (dst_cnt = i = 0; i < rename_dst_nr; i++) { |
25d5ea41 | 553 | struct diff_filespec *two = rename_dst[i].two; |
6d24ad97 JH |
554 | struct diff_score *m; |
555 | ||
25d5ea41 | 556 | if (rename_dst[i].pair) |
427dcb4b | 557 | continue; /* dealt with exact match already. */ |
6d24ad97 JH |
558 | |
559 | m = &mx[dst_cnt * NUM_CANDIDATE_PER_DST]; | |
560 | for (j = 0; j < NUM_CANDIDATE_PER_DST; j++) | |
561 | m[j].dst = -1; | |
562 | ||
25d5ea41 | 563 | for (j = 0; j < rename_src_nr; j++) { |
e88d6bc6 | 564 | struct diff_filespec *one = rename_src[j].p->one; |
6d24ad97 | 565 | struct diff_score this_src; |
f31027c9 JH |
566 | |
567 | if (skip_unmodified && | |
568 | diff_unmodified_pair(rename_src[j].p)) | |
569 | continue; | |
570 | ||
b78ea5fc NTND |
571 | this_src.score = estimate_similarity(options->repo, |
572 | one, two, | |
6d24ad97 JH |
573 | minimum_score); |
574 | this_src.name_score = basename_same(one, two); | |
575 | this_src.dst = i; | |
576 | this_src.src = j; | |
577 | record_if_better(m, &this_src); | |
809809bb JH |
578 | /* |
579 | * Once we run estimate_similarity, | |
580 | * We do not need the text anymore. | |
581 | */ | |
8ae92e63 | 582 | diff_free_filespec_blob(one); |
809809bb | 583 | diff_free_filespec_blob(two); |
427dcb4b JH |
584 | } |
585 | dst_cnt++; | |
d6861d02 | 586 | display_progress(progress, (uint64_t)(i+1)*(uint64_t)rename_src_nr); |
427dcb4b | 587 | } |
3ac942d4 | 588 | stop_progress(&progress); |
6d24ad97 | 589 | |
427dcb4b | 590 | /* cost matrix sorted by most to least similar pair */ |
9ed0d8d6 | 591 | QSORT(mx, dst_cnt * NUM_CANDIDATE_PER_DST, score_compare); |
6d24ad97 | 592 | |
0940e5f2 LT |
593 | rename_count += find_renames(mx, dst_cnt, minimum_score, 0); |
594 | if (detect_rename == DIFF_DETECT_COPY) | |
595 | rename_count += find_renames(mx, dst_cnt, minimum_score, 1); | |
427dcb4b | 596 | free(mx); |
427dcb4b | 597 | |
15d061b4 | 598 | cleanup: |
427dcb4b | 599 | /* At this point, we have found some renames and copies and they |
5098bafb | 600 | * are recorded in rename_dst. The original list is still in *q. |
427dcb4b | 601 | */ |
9ca5df90 | 602 | DIFF_QUEUE_CLEAR(&outq); |
427dcb4b | 603 | for (i = 0; i < q->nr; i++) { |
25d5ea41 | 604 | struct diff_filepair *p = q->queue[i]; |
25d5ea41 JH |
605 | struct diff_filepair *pair_to_free = NULL; |
606 | ||
d7c9bf22 MZ |
607 | if (DIFF_PAIR_UNMERGED(p)) { |
608 | diff_q(&outq, p); | |
609 | } | |
610 | else if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) { | |
2cd68882 JH |
611 | /* |
612 | * Creation | |
613 | * | |
614 | * We would output this create record if it has | |
615 | * not been turned into a rename/copy already. | |
616 | */ | |
f98c2f7e | 617 | struct diff_rename_dst *dst = locate_rename_dst(p->two); |
2cd68882 | 618 | if (dst && dst->pair) { |
25d5ea41 JH |
619 | diff_q(&outq, dst->pair); |
620 | pair_to_free = p; | |
621 | } | |
622 | else | |
2cd68882 JH |
623 | /* no matching rename/copy source, so |
624 | * record this as a creation. | |
25d5ea41 JH |
625 | */ |
626 | diff_q(&outq, p); | |
427dcb4b | 627 | } |
2cd68882 JH |
628 | else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) { |
629 | /* | |
630 | * Deletion | |
631 | * | |
f345b0a0 JH |
632 | * We would output this delete record if: |
633 | * | |
634 | * (1) this is a broken delete and the counterpart | |
635 | * broken create remains in the output; or | |
5098bafb JH |
636 | * (2) this is not a broken delete, and rename_dst |
637 | * does not have a rename/copy to move p->one->path | |
638 | * out of existence. | |
f345b0a0 JH |
639 | * |
640 | * Otherwise, the counterpart broken create | |
641 | * has been turned into a rename-edit; or | |
642 | * delete did not have a matching create to | |
643 | * begin with. | |
2cd68882 | 644 | */ |
f345b0a0 JH |
645 | if (DIFF_PAIR_BROKEN(p)) { |
646 | /* broken delete */ | |
f98c2f7e | 647 | struct diff_rename_dst *dst = locate_rename_dst(p->one); |
f345b0a0 JH |
648 | if (dst && dst->pair) |
649 | /* counterpart is now rename/copy */ | |
650 | pair_to_free = p; | |
651 | } | |
652 | else { | |
64479711 | 653 | if (p->one->rename_used) |
f345b0a0 JH |
654 | /* this path remains */ |
655 | pair_to_free = p; | |
656 | } | |
2cd68882 JH |
657 | |
658 | if (pair_to_free) | |
659 | ; | |
660 | else | |
661 | diff_q(&outq, p); | |
662 | } | |
25d5ea41 | 663 | else if (!diff_unmodified_pair(p)) |
15d061b4 | 664 | /* all the usual ones need to be kept */ |
25d5ea41 | 665 | diff_q(&outq, p); |
15d061b4 JH |
666 | else |
667 | /* no need to keep unmodified pairs */ | |
668 | pair_to_free = p; | |
669 | ||
226406f6 JH |
670 | if (pair_to_free) |
671 | diff_free_filepair(pair_to_free); | |
427dcb4b | 672 | } |
25d5ea41 | 673 | diff_debug_queue("done copying original", &outq); |
427dcb4b | 674 | |
25d5ea41 JH |
675 | free(q->queue); |
676 | *q = outq; | |
677 | diff_debug_queue("done collapsing", q); | |
427dcb4b | 678 | |
9fb88419 LT |
679 | for (i = 0; i < rename_dst_nr; i++) |
680 | free_filespec(rename_dst[i].two); | |
5098bafb | 681 | |
6a83d902 | 682 | FREE_AND_NULL(rename_dst); |
25d5ea41 | 683 | rename_dst_nr = rename_dst_alloc = 0; |
6a83d902 | 684 | FREE_AND_NULL(rename_src); |
25d5ea41 | 685 | rename_src_nr = rename_src_alloc = 0; |
427dcb4b JH |
686 | return; |
687 | } |