]>
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) { | |
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) | |
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) { | |
86 | int next = (last + first) >> 1; | |
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 | ||
131 | static int estimate_similarity(struct diff_filespec *src, | |
132 | struct diff_filespec *dst, | |
133 | int minimum_score) | |
134 | { | |
135 | /* src points at a file that existed in the original tree (or | |
136 | * optionally a file in the destination tree) and dst points | |
137 | * at a newly created file. They may be quite similar, in which | |
138 | * case we want to say src is renamed to dst or src is copied into | |
139 | * dst, and then some edit has been applied to dst. | |
140 | * | |
141 | * Compare them and return how similar they are, representing | |
f0c6b2a2 JH |
142 | * the score as an integer between 0 and MAX_SCORE. |
143 | * | |
144 | * When there is an exact match, it is considered a better | |
145 | * match than anything else; the destination does not even | |
146 | * call into this function in that case. | |
427dcb4b | 147 | */ |
90bd932c | 148 | unsigned long max_size, delta_size, base_size, src_copied, literal_added; |
427dcb4b JH |
149 | int score; |
150 | ||
60896c7b JH |
151 | /* We deal only with regular files. Symlink renames are handled |
152 | * only when they are exact matches --- in other words, no edits | |
153 | * after renaming. | |
154 | */ | |
155 | if (!S_ISREG(src->mode) || !S_ISREG(dst->mode)) | |
156 | return 0; | |
157 | ||
81ac051d LT |
158 | /* |
159 | * Need to check that source and destination sizes are | |
160 | * filled in before comparing them. | |
161 | * | |
162 | * If we already have "cnt_data" filled in, we know it's | |
163 | * all good (avoid checking the size for zero, as that | |
164 | * is a possible size - we really should have a flag to | |
165 | * say whether the size is valid or not!) | |
166 | */ | |
8e5dd3d6 NTND |
167 | if (!src->cnt_data && |
168 | diff_populate_filespec(src, CHECK_SIZE_ONLY)) | |
81ac051d | 169 | return 0; |
8e5dd3d6 NTND |
170 | if (!dst->cnt_data && |
171 | diff_populate_filespec(dst, CHECK_SIZE_ONLY)) | |
81ac051d LT |
172 | return 0; |
173 | ||
90bd932c | 174 | max_size = ((src->size > dst->size) ? src->size : dst->size); |
58b103f5 | 175 | base_size = ((src->size < dst->size) ? src->size : dst->size); |
90bd932c | 176 | delta_size = max_size - base_size; |
427dcb4b | 177 | |
58b103f5 JH |
178 | /* We would not consider edits that change the file size so |
179 | * drastically. delta_size must be smaller than | |
cd1870ed | 180 | * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size). |
f0c6b2a2 | 181 | * |
58b103f5 JH |
182 | * Note that base_size == 0 case is handled here already |
183 | * and the final score computation below would not have a | |
184 | * divide-by-zero issue. | |
427dcb4b | 185 | */ |
3a4d6769 | 186 | if (max_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE) |
427dcb4b JH |
187 | return 0; |
188 | ||
885c716f BS |
189 | if (!src->cnt_data && diff_populate_filespec(src, 0)) |
190 | return 0; | |
191 | if (!dst->cnt_data && diff_populate_filespec(dst, 0)) | |
192 | return 0; | |
193 | ||
d8c3d03a | 194 | if (diffcore_count_changes(src, dst, |
c06c7966 | 195 | &src->cnt_data, &dst->cnt_data, |
65416758 | 196 | &src_copied, &literal_added)) |
85976974 | 197 | return 0; |
355e76a4 | 198 | |
1706306a JH |
199 | /* How similar are they? |
200 | * what percentage of material in dst are from source? | |
427dcb4b | 201 | */ |
90bd932c | 202 | if (!dst->size) |
1706306a | 203 | score = 0; /* should not happen */ |
cfc0aef1 | 204 | else |
dc49cd76 | 205 | score = (int)(src_copied * MAX_SCORE / max_size); |
427dcb4b JH |
206 | return score; |
207 | } | |
208 | ||
5098bafb | 209 | static void record_rename_pair(int dst_index, int src_index, int score) |
427dcb4b | 210 | { |
9fb88419 | 211 | struct diff_filespec *src, *dst; |
25d5ea41 | 212 | struct diff_filepair *dp; |
f7c1512a | 213 | |
25d5ea41 JH |
214 | if (rename_dst[dst_index].pair) |
215 | die("internal error: dst already matched."); | |
427dcb4b | 216 | |
e88d6bc6 | 217 | src = rename_src[src_index].p->one; |
64479711 | 218 | src->rename_used++; |
9fb88419 | 219 | src->count++; |
427dcb4b | 220 | |
25d5ea41 | 221 | dst = rename_dst[dst_index].two; |
9fb88419 | 222 | dst->count++; |
427dcb4b | 223 | |
9fb88419 | 224 | dp = diff_queue(NULL, src, dst); |
ef677686 | 225 | dp->renamed_pair = 1; |
fc580719 JH |
226 | if (!strcmp(src->path, dst->path)) |
227 | dp->score = rename_src[src_index].score; | |
228 | else | |
229 | dp->score = score; | |
25d5ea41 | 230 | rename_dst[dst_index].pair = dp; |
427dcb4b JH |
231 | } |
232 | ||
233 | /* | |
234 | * We sort the rename similarity matrix with the score, in descending | |
15d061b4 | 235 | * order (the most similar first). |
427dcb4b JH |
236 | */ |
237 | static int score_compare(const void *a_, const void *b_) | |
238 | { | |
239 | const struct diff_score *a = a_, *b = b_; | |
cfc0aef1 | 240 | |
6d24ad97 JH |
241 | /* sink the unused ones to the bottom */ |
242 | if (a->dst < 0) | |
243 | return (0 <= b->dst); | |
244 | else if (b->dst < 0) | |
245 | return -1; | |
246 | ||
cfc0aef1 RS |
247 | if (a->score == b->score) |
248 | return b->name_score - a->name_score; | |
249 | ||
427dcb4b JH |
250 | return b->score - a->score; |
251 | } | |
252 | ||
9027f53c | 253 | struct file_similarity { |
f79d9c58 | 254 | struct hashmap_entry entry; |
7c85f8ac | 255 | int index; |
9027f53c | 256 | struct diff_filespec *filespec; |
9027f53c LT |
257 | }; |
258 | ||
48f6407f KB |
259 | static unsigned int hash_filespec(struct diff_filespec *filespec) |
260 | { | |
41c9560e | 261 | if (!filespec->oid_valid) { |
48f6407f KB |
262 | if (diff_populate_filespec(filespec, 0)) |
263 | return 0; | |
f070facc PO |
264 | hash_object_file(filespec->data, filespec->size, "blob", |
265 | &filespec->oid); | |
48f6407f | 266 | } |
a0d12c44 | 267 | return sha1hash(filespec->oid.hash); |
48f6407f KB |
268 | } |
269 | ||
f79d9c58 | 270 | static int find_identical_files(struct hashmap *srcs, |
7c85f8ac | 271 | int dst_index, |
11f944dd | 272 | struct diff_options *options) |
9027f53c LT |
273 | { |
274 | int renames = 0; | |
275 | ||
7c85f8ac | 276 | struct diff_filespec *target = rename_dst[dst_index].two; |
ab73a9d1 | 277 | struct file_similarity *p, *best = NULL; |
48f6407f KB |
278 | int i = 100, best_score = -1; |
279 | ||
280 | /* | |
7c85f8ac | 281 | * Find the best source match for specified destination. |
48f6407f | 282 | */ |
ab73a9d1 KB |
283 | p = hashmap_get_from_hash(srcs, hash_filespec(target), NULL); |
284 | for (; p; p = hashmap_get_next(srcs, p)) { | |
48f6407f KB |
285 | int score; |
286 | struct diff_filespec *source = p->filespec; | |
287 | ||
288 | /* False hash collision? */ | |
a0d12c44 | 289 | if (oidcmp(&source->oid, &target->oid)) |
48f6407f KB |
290 | continue; |
291 | /* Non-regular files? If so, the modes must match! */ | |
292 | if (!S_ISREG(source->mode) || !S_ISREG(target->mode)) { | |
293 | if (source->mode != target->mode) | |
0940e5f2 | 294 | continue; |
9027f53c | 295 | } |
48f6407f KB |
296 | /* Give higher scores to sources that haven't been used already */ |
297 | score = !source->rename_used; | |
298 | if (source->rename_used && options->detect_rename != DIFF_DETECT_COPY) | |
299 | continue; | |
300 | score += basename_same(source, target); | |
301 | if (score > best_score) { | |
302 | best = p; | |
303 | best_score = score; | |
304 | if (score == 2) | |
305 | break; | |
9027f53c | 306 | } |
48f6407f KB |
307 | |
308 | /* Too many identical alternatives? Pick one */ | |
309 | if (!--i) | |
310 | break; | |
311 | } | |
312 | if (best) { | |
7c85f8ac | 313 | record_rename_pair(dst_index, best->index, MAX_SCORE); |
48f6407f KB |
314 | renames++; |
315 | } | |
9027f53c LT |
316 | return renames; |
317 | } | |
318 | ||
f79d9c58 | 319 | static void insert_file_table(struct hashmap *table, int index, struct diff_filespec *filespec) |
9027f53c | 320 | { |
9027f53c LT |
321 | struct file_similarity *entry = xmalloc(sizeof(*entry)); |
322 | ||
9027f53c LT |
323 | entry->index = index; |
324 | entry->filespec = filespec; | |
9027f53c | 325 | |
f79d9c58 KB |
326 | hashmap_entry_init(entry, hash_filespec(filespec)); |
327 | hashmap_add(table, entry); | |
9027f53c LT |
328 | } |
329 | ||
cb1491b6 LT |
330 | /* |
331 | * Find exact renames first. | |
332 | * | |
333 | * The first round matches up the up-to-date entries, | |
334 | * and then during the second round we try to match | |
335 | * cache-dirty entries as well. | |
cb1491b6 | 336 | */ |
11f944dd | 337 | static int find_exact_renames(struct diff_options *options) |
cb1491b6 | 338 | { |
7c85f8ac | 339 | int i, renames = 0; |
f79d9c58 | 340 | struct hashmap file_table; |
cb1491b6 | 341 | |
ca4e3ca0 SG |
342 | /* Add all sources to the hash table in reverse order, because |
343 | * later on they will be retrieved in LIFO order. | |
344 | */ | |
7663cdc8 | 345 | hashmap_init(&file_table, NULL, NULL, rename_src_nr); |
ca4e3ca0 | 346 | for (i = rename_src_nr-1; i >= 0; i--) |
7c85f8ac | 347 | insert_file_table(&file_table, i, rename_src[i].p->one); |
9027f53c | 348 | |
7c85f8ac | 349 | /* Walk the destinations and find best source match */ |
9027f53c | 350 | for (i = 0; i < rename_dst_nr; i++) |
7c85f8ac | 351 | renames += find_identical_files(&file_table, i, options); |
9027f53c | 352 | |
f79d9c58 KB |
353 | /* Free the hash data structure and entries */ |
354 | hashmap_free(&file_table, 1); | |
9027f53c | 355 | |
7c85f8ac | 356 | return renames; |
cb1491b6 LT |
357 | } |
358 | ||
6d24ad97 JH |
359 | #define NUM_CANDIDATE_PER_DST 4 |
360 | static void record_if_better(struct diff_score m[], struct diff_score *o) | |
361 | { | |
362 | int i, worst; | |
363 | ||
364 | /* find the worst one */ | |
365 | worst = 0; | |
366 | for (i = 1; i < NUM_CANDIDATE_PER_DST; i++) | |
367 | if (score_compare(&m[i], &m[worst]) > 0) | |
368 | worst = i; | |
369 | ||
370 | /* is it better than the worst one? */ | |
371 | if (score_compare(&m[worst], o) > 0) | |
372 | m[worst] = *o; | |
373 | } | |
374 | ||
f31027c9 JH |
375 | /* |
376 | * Returns: | |
377 | * 0 if we are under the limit; | |
378 | * 1 if we need to disable inexact rename detection; | |
379 | * 2 if we would be under the limit if we were given -C instead of -C -C. | |
380 | */ | |
9d8a5a50 JH |
381 | static int too_many_rename_candidates(int num_create, |
382 | struct diff_options *options) | |
383 | { | |
384 | int rename_limit = options->rename_limit; | |
385 | int num_src = rename_src_nr; | |
f31027c9 | 386 | int i; |
9d8a5a50 JH |
387 | |
388 | options->needed_rename_limit = 0; | |
389 | ||
390 | /* | |
391 | * This basically does a test for the rename matrix not | |
392 | * growing larger than a "rename_limit" square matrix, ie: | |
393 | * | |
394 | * num_create * num_src > rename_limit * rename_limit | |
9d8a5a50 | 395 | */ |
89973554 JT |
396 | if (rename_limit <= 0) |
397 | rename_limit = 32767; | |
9d8a5a50 | 398 | if ((num_create <= rename_limit || num_src <= rename_limit) && |
9f7e4bfa EN |
399 | ((uint64_t)num_create * (uint64_t)num_src |
400 | <= (uint64_t)rename_limit * (uint64_t)rename_limit)) | |
9d8a5a50 JH |
401 | return 0; |
402 | ||
403 | options->needed_rename_limit = | |
404 | num_src > num_create ? num_src : num_create; | |
f31027c9 JH |
405 | |
406 | /* Are we running under -C -C? */ | |
0d1e0e78 | 407 | if (!options->flags.find_copies_harder) |
f31027c9 JH |
408 | return 1; |
409 | ||
410 | /* Would we bust the limit if we were running under -C? */ | |
411 | for (num_src = i = 0; i < rename_src_nr; i++) { | |
412 | if (diff_unmodified_pair(rename_src[i].p)) | |
413 | continue; | |
414 | num_src++; | |
415 | } | |
416 | if ((num_create <= rename_limit || num_src <= rename_limit) && | |
9f7e4bfa EN |
417 | ((uint64_t)num_create * (uint64_t)num_src |
418 | <= (uint64_t)rename_limit * (uint64_t)rename_limit)) | |
f31027c9 | 419 | return 2; |
9d8a5a50 JH |
420 | return 1; |
421 | } | |
422 | ||
0940e5f2 LT |
423 | static int find_renames(struct diff_score *mx, int dst_cnt, int minimum_score, int copies) |
424 | { | |
425 | int count = 0, i; | |
426 | ||
427 | for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) { | |
428 | struct diff_rename_dst *dst; | |
429 | ||
430 | if ((mx[i].dst < 0) || | |
431 | (mx[i].score < minimum_score)) | |
432 | break; /* there is no more usable pair. */ | |
433 | dst = &rename_dst[mx[i].dst]; | |
434 | if (dst->pair) | |
435 | continue; /* already done, either exact or fuzzy. */ | |
e88d6bc6 | 436 | if (!copies && rename_src[mx[i].src].p->one->rename_used) |
0940e5f2 LT |
437 | continue; |
438 | record_rename_pair(mx[i].dst, mx[i].src, mx[i].score); | |
439 | count++; | |
440 | } | |
441 | return count; | |
442 | } | |
443 | ||
8082d8d3 | 444 | void diffcore_rename(struct diff_options *options) |
427dcb4b | 445 | { |
8082d8d3 JH |
446 | int detect_rename = options->detect_rename; |
447 | int minimum_score = options->rename_score; | |
38c6f780 | 448 | struct diff_queue_struct *q = &diff_queued_diff; |
5098bafb | 449 | struct diff_queue_struct outq; |
427dcb4b | 450 | struct diff_score *mx; |
f31027c9 | 451 | int i, j, rename_count, skip_unmodified = 0; |
dabdbee1 | 452 | int num_create, dst_cnt; |
3ac942d4 | 453 | struct progress *progress = NULL; |
427dcb4b | 454 | |
26dee0ad | 455 | if (!minimum_score) |
f345b0a0 | 456 | minimum_score = DEFAULT_RENAME_SCORE; |
427dcb4b | 457 | |
427dcb4b | 458 | for (i = 0; i < q->nr; i++) { |
52e95789 | 459 | struct diff_filepair *p = q->queue[i]; |
2f3f8b21 | 460 | if (!DIFF_FILE_VALID(p->one)) { |
81e50eab | 461 | if (!DIFF_FILE_VALID(p->two)) |
60896c7b | 462 | continue; /* unmerged */ |
2f3f8b21 JH |
463 | else if (options->single_follow && |
464 | strcmp(options->single_follow, p->two->path)) | |
465 | continue; /* not interested */ | |
0d1e0e78 | 466 | else if (!options->flags.rename_empty && |
02491b67 | 467 | is_empty_blob_oid(&p->two->oid)) |
90d43b07 | 468 | continue; |
4d6be03b JK |
469 | else if (add_rename_dst(p->two) < 0) { |
470 | warning("skipping rename detection, detected" | |
471 | " duplicate destination '%s'", | |
472 | p->two->path); | |
473 | goto cleanup; | |
474 | } | |
2f3f8b21 | 475 | } |
0d1e0e78 | 476 | else if (!options->flags.rename_empty && |
02491b67 | 477 | is_empty_blob_oid(&p->one->oid)) |
90d43b07 | 478 | continue; |
d7c9bf22 | 479 | else if (!DIFF_PAIR_UNMERGED(p) && !DIFF_FILE_VALID(p->two)) { |
64479711 LT |
480 | /* |
481 | * If the source is a broken "delete", and | |
2210100a JH |
482 | * they did not really want to get broken, |
483 | * that means the source actually stays. | |
64479711 LT |
484 | * So we increment the "rename_used" score |
485 | * by one, to indicate ourselves as a user | |
486 | */ | |
487 | if (p->broken_pair && !p->score) | |
488 | p->one->rename_used++; | |
e88d6bc6 | 489 | register_rename_src(p); |
64479711 LT |
490 | } |
491 | else if (detect_rename == DIFF_DETECT_COPY) { | |
492 | /* | |
493 | * Increment the "rename_used" score by | |
494 | * one, to indicate ourselves as a user. | |
2210100a | 495 | */ |
64479711 | 496 | p->one->rename_used++; |
e88d6bc6 | 497 | register_rename_src(p); |
2210100a | 498 | } |
427dcb4b | 499 | } |
0024a549 | 500 | if (rename_dst_nr == 0 || rename_src_nr == 0) |
427dcb4b JH |
501 | goto cleanup; /* nothing to do */ |
502 | ||
17559a64 LT |
503 | /* |
504 | * We really want to cull the candidates list early | |
505 | * with cheap tests in order to avoid doing deltas. | |
506 | */ | |
11f944dd | 507 | rename_count = find_exact_renames(options); |
17559a64 | 508 | |
42899ac8 LT |
509 | /* Did we only want exact renames? */ |
510 | if (minimum_score == MAX_SCORE) | |
511 | goto cleanup; | |
512 | ||
513 | /* | |
514 | * Calculate how many renames are left (but all the source | |
515 | * files still remain as options for rename/copies!) | |
516 | */ | |
517 | num_create = (rename_dst_nr - rename_count); | |
42899ac8 LT |
518 | |
519 | /* All done? */ | |
520 | if (!num_create) | |
521 | goto cleanup; | |
522 | ||
f31027c9 JH |
523 | switch (too_many_rename_candidates(num_create, options)) { |
524 | case 1: | |
0024a549 | 525 | goto cleanup; |
f31027c9 JH |
526 | case 2: |
527 | options->degraded_cc_to_c = 1; | |
528 | skip_unmodified = 1; | |
529 | break; | |
530 | default: | |
531 | break; | |
532 | } | |
0024a549 | 533 | |
3ac942d4 | 534 | if (options->show_rename_progress) { |
8aade107 | 535 | progress = start_delayed_progress( |
754dbc43 | 536 | _("Performing inexact rename detection"), |
d6861d02 | 537 | (uint64_t)rename_dst_nr * (uint64_t)rename_src_nr); |
3ac942d4 JK |
538 | } |
539 | ||
50492f7b | 540 | mx = xcalloc(st_mult(NUM_CANDIDATE_PER_DST, num_create), sizeof(*mx)); |
25d5ea41 | 541 | for (dst_cnt = i = 0; i < rename_dst_nr; i++) { |
25d5ea41 | 542 | struct diff_filespec *two = rename_dst[i].two; |
6d24ad97 JH |
543 | struct diff_score *m; |
544 | ||
25d5ea41 | 545 | if (rename_dst[i].pair) |
427dcb4b | 546 | continue; /* dealt with exact match already. */ |
6d24ad97 JH |
547 | |
548 | m = &mx[dst_cnt * NUM_CANDIDATE_PER_DST]; | |
549 | for (j = 0; j < NUM_CANDIDATE_PER_DST; j++) | |
550 | m[j].dst = -1; | |
551 | ||
25d5ea41 | 552 | for (j = 0; j < rename_src_nr; j++) { |
e88d6bc6 | 553 | struct diff_filespec *one = rename_src[j].p->one; |
6d24ad97 | 554 | struct diff_score this_src; |
f31027c9 JH |
555 | |
556 | if (skip_unmodified && | |
557 | diff_unmodified_pair(rename_src[j].p)) | |
558 | continue; | |
559 | ||
6d24ad97 JH |
560 | this_src.score = estimate_similarity(one, two, |
561 | minimum_score); | |
562 | this_src.name_score = basename_same(one, two); | |
563 | this_src.dst = i; | |
564 | this_src.src = j; | |
565 | record_if_better(m, &this_src); | |
809809bb JH |
566 | /* |
567 | * Once we run estimate_similarity, | |
568 | * We do not need the text anymore. | |
569 | */ | |
8ae92e63 | 570 | diff_free_filespec_blob(one); |
809809bb | 571 | diff_free_filespec_blob(two); |
427dcb4b JH |
572 | } |
573 | dst_cnt++; | |
d6861d02 | 574 | display_progress(progress, (uint64_t)(i+1)*(uint64_t)rename_src_nr); |
427dcb4b | 575 | } |
3ac942d4 | 576 | stop_progress(&progress); |
6d24ad97 | 577 | |
427dcb4b | 578 | /* cost matrix sorted by most to least similar pair */ |
9ed0d8d6 | 579 | QSORT(mx, dst_cnt * NUM_CANDIDATE_PER_DST, score_compare); |
6d24ad97 | 580 | |
0940e5f2 LT |
581 | rename_count += find_renames(mx, dst_cnt, minimum_score, 0); |
582 | if (detect_rename == DIFF_DETECT_COPY) | |
583 | rename_count += find_renames(mx, dst_cnt, minimum_score, 1); | |
427dcb4b | 584 | free(mx); |
427dcb4b | 585 | |
15d061b4 | 586 | cleanup: |
427dcb4b | 587 | /* At this point, we have found some renames and copies and they |
5098bafb | 588 | * are recorded in rename_dst. The original list is still in *q. |
427dcb4b | 589 | */ |
9ca5df90 | 590 | DIFF_QUEUE_CLEAR(&outq); |
427dcb4b | 591 | for (i = 0; i < q->nr; i++) { |
25d5ea41 | 592 | struct diff_filepair *p = q->queue[i]; |
25d5ea41 JH |
593 | struct diff_filepair *pair_to_free = NULL; |
594 | ||
d7c9bf22 MZ |
595 | if (DIFF_PAIR_UNMERGED(p)) { |
596 | diff_q(&outq, p); | |
597 | } | |
598 | else if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) { | |
2cd68882 JH |
599 | /* |
600 | * Creation | |
601 | * | |
602 | * We would output this create record if it has | |
603 | * not been turned into a rename/copy already. | |
604 | */ | |
f98c2f7e | 605 | struct diff_rename_dst *dst = locate_rename_dst(p->two); |
2cd68882 | 606 | if (dst && dst->pair) { |
25d5ea41 JH |
607 | diff_q(&outq, dst->pair); |
608 | pair_to_free = p; | |
609 | } | |
610 | else | |
2cd68882 JH |
611 | /* no matching rename/copy source, so |
612 | * record this as a creation. | |
25d5ea41 JH |
613 | */ |
614 | diff_q(&outq, p); | |
427dcb4b | 615 | } |
2cd68882 JH |
616 | else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) { |
617 | /* | |
618 | * Deletion | |
619 | * | |
f345b0a0 JH |
620 | * We would output this delete record if: |
621 | * | |
622 | * (1) this is a broken delete and the counterpart | |
623 | * broken create remains in the output; or | |
5098bafb JH |
624 | * (2) this is not a broken delete, and rename_dst |
625 | * does not have a rename/copy to move p->one->path | |
626 | * out of existence. | |
f345b0a0 JH |
627 | * |
628 | * Otherwise, the counterpart broken create | |
629 | * has been turned into a rename-edit; or | |
630 | * delete did not have a matching create to | |
631 | * begin with. | |
2cd68882 | 632 | */ |
f345b0a0 JH |
633 | if (DIFF_PAIR_BROKEN(p)) { |
634 | /* broken delete */ | |
f98c2f7e | 635 | struct diff_rename_dst *dst = locate_rename_dst(p->one); |
f345b0a0 JH |
636 | if (dst && dst->pair) |
637 | /* counterpart is now rename/copy */ | |
638 | pair_to_free = p; | |
639 | } | |
640 | else { | |
64479711 | 641 | if (p->one->rename_used) |
f345b0a0 JH |
642 | /* this path remains */ |
643 | pair_to_free = p; | |
644 | } | |
2cd68882 JH |
645 | |
646 | if (pair_to_free) | |
647 | ; | |
648 | else | |
649 | diff_q(&outq, p); | |
650 | } | |
25d5ea41 | 651 | else if (!diff_unmodified_pair(p)) |
15d061b4 | 652 | /* all the usual ones need to be kept */ |
25d5ea41 | 653 | diff_q(&outq, p); |
15d061b4 JH |
654 | else |
655 | /* no need to keep unmodified pairs */ | |
656 | pair_to_free = p; | |
657 | ||
226406f6 JH |
658 | if (pair_to_free) |
659 | diff_free_filepair(pair_to_free); | |
427dcb4b | 660 | } |
25d5ea41 | 661 | diff_debug_queue("done copying original", &outq); |
427dcb4b | 662 | |
25d5ea41 JH |
663 | free(q->queue); |
664 | *q = outq; | |
665 | diff_debug_queue("done collapsing", q); | |
427dcb4b | 666 | |
9fb88419 LT |
667 | for (i = 0; i < rename_dst_nr; i++) |
668 | free_filespec(rename_dst[i].two); | |
5098bafb | 669 | |
6a83d902 | 670 | FREE_AND_NULL(rename_dst); |
25d5ea41 | 671 | rename_dst_nr = rename_dst_alloc = 0; |
6a83d902 | 672 | FREE_AND_NULL(rename_src); |
25d5ea41 | 673 | rename_src_nr = rename_src_alloc = 0; |
427dcb4b JH |
674 | return; |
675 | } |