]>
Commit | Line | Data |
---|---|---|
427dcb4b | 1 | /* |
95acf11a | 2 | * |
427dcb4b JH |
3 | * Copyright (C) 2005 Junio C Hamano |
4 | */ | |
5 | #include "cache.h" | |
6 | #include "diff.h" | |
7 | #include "diffcore.h" | |
cbd53a21 | 8 | #include "object-store.h" |
f79d9c58 | 9 | #include "hashmap.h" |
3ac942d4 | 10 | #include "progress.h" |
95acf11a | 11 | #include "promisor-remote.h" |
9db2ac56 | 12 | #include "strmap.h" |
427dcb4b | 13 | |
25d5ea41 JH |
14 | /* Table of rename/copy destinations */ |
15 | ||
16 | static struct diff_rename_dst { | |
9db2ac56 EN |
17 | struct diff_filepair *p; |
18 | struct diff_filespec *filespec_to_free; | |
19 | int is_rename; /* false -> just a create; true -> rename or copy */ | |
25d5ea41 JH |
20 | } *rename_dst; |
21 | static int rename_dst_nr, rename_dst_alloc; | |
9db2ac56 EN |
22 | /* Mapping from break source pathname to break destination index */ |
23 | static struct strintmap *break_idx = NULL; | |
427dcb4b | 24 | |
9db2ac56 | 25 | static struct diff_rename_dst *locate_rename_dst(struct diff_filepair *p) |
f98c2f7e | 26 | { |
9db2ac56 EN |
27 | /* Lookup by p->ONE->path */ |
28 | int idx = break_idx ? strintmap_get(break_idx, p->one->path) : -1; | |
29 | return (idx == -1) ? NULL : &rename_dst[idx]; | |
f98c2f7e JK |
30 | } |
31 | ||
32 | /* | |
33 | * Returns 0 on success, -1 if we found a duplicate. | |
34 | */ | |
9db2ac56 | 35 | static int add_rename_dst(struct diff_filepair *p) |
f98c2f7e | 36 | { |
337ce247 | 37 | ALLOC_GROW(rename_dst, rename_dst_nr + 1, rename_dst_alloc); |
9db2ac56 EN |
38 | rename_dst[rename_dst_nr].p = p; |
39 | rename_dst[rename_dst_nr].filespec_to_free = NULL; | |
40 | rename_dst[rename_dst_nr].is_rename = 0; | |
25d5ea41 | 41 | rename_dst_nr++; |
f98c2f7e | 42 | return 0; |
427dcb4b JH |
43 | } |
44 | ||
15d061b4 | 45 | /* Table of rename/copy src files */ |
25d5ea41 | 46 | static struct diff_rename_src { |
e88d6bc6 | 47 | struct diff_filepair *p; |
fc580719 | 48 | unsigned short score; /* to remember the break score */ |
25d5ea41 JH |
49 | } *rename_src; |
50 | static int rename_src_nr, rename_src_alloc; | |
427dcb4b | 51 | |
b970b4ef | 52 | static void register_rename_src(struct diff_filepair *p) |
25d5ea41 | 53 | { |
9db2ac56 EN |
54 | if (p->broken_pair) { |
55 | if (!break_idx) { | |
56 | break_idx = xmalloc(sizeof(*break_idx)); | |
57 | strintmap_init(break_idx, -1); | |
58 | } | |
59 | strintmap_set(break_idx, p->one->path, rename_dst_nr); | |
60 | } | |
61 | ||
337ce247 | 62 | ALLOC_GROW(rename_src, rename_src_nr + 1, rename_src_alloc); |
b970b4ef EN |
63 | rename_src[rename_src_nr].p = p; |
64 | rename_src[rename_src_nr].score = p->score; | |
25d5ea41 | 65 | rename_src_nr++; |
427dcb4b JH |
66 | } |
67 | ||
0ce39643 JS |
68 | static int basename_same(struct diff_filespec *src, struct diff_filespec *dst) |
69 | { | |
70 | int src_len = strlen(src->path), dst_len = strlen(dst->path); | |
71 | while (src_len && dst_len) { | |
72 | char c1 = src->path[--src_len]; | |
73 | char c2 = dst->path[--dst_len]; | |
74 | if (c1 != c2) | |
75 | return 0; | |
76 | if (c1 == '/') | |
77 | return 1; | |
78 | } | |
79 | return (!src_len || src->path[src_len - 1] == '/') && | |
80 | (!dst_len || dst->path[dst_len - 1] == '/'); | |
81 | } | |
82 | ||
427dcb4b | 83 | struct diff_score { |
25d5ea41 JH |
84 | int src; /* index in rename_src */ |
85 | int dst; /* index in rename_dst */ | |
6d24ad97 JH |
86 | unsigned short score; |
87 | short name_score; | |
427dcb4b JH |
88 | }; |
89 | ||
95acf11a JT |
90 | struct prefetch_options { |
91 | struct repository *repo; | |
92 | int skip_unmodified; | |
93 | }; | |
94 | static void prefetch(void *prefetch_options) | |
95 | { | |
96 | struct prefetch_options *options = prefetch_options; | |
97 | int i; | |
98 | struct oid_array to_fetch = OID_ARRAY_INIT; | |
99 | ||
100 | for (i = 0; i < rename_dst_nr; i++) { | |
9db2ac56 | 101 | if (rename_dst[i].p->renamed_pair) |
95acf11a JT |
102 | /* |
103 | * The loop in diffcore_rename() will not need these | |
104 | * blobs, so skip prefetching. | |
105 | */ | |
106 | continue; /* already found exact match */ | |
107 | diff_add_if_missing(options->repo, &to_fetch, | |
9db2ac56 | 108 | rename_dst[i].p->two); |
95acf11a JT |
109 | } |
110 | for (i = 0; i < rename_src_nr; i++) { | |
111 | if (options->skip_unmodified && | |
112 | diff_unmodified_pair(rename_src[i].p)) | |
113 | /* | |
114 | * The loop in diffcore_rename() will not need these | |
115 | * blobs, so skip prefetching. | |
116 | */ | |
117 | continue; | |
118 | diff_add_if_missing(options->repo, &to_fetch, | |
119 | rename_src[i].p->one); | |
120 | } | |
121 | promisor_remote_get_direct(options->repo, to_fetch.oid, to_fetch.nr); | |
122 | oid_array_clear(&to_fetch); | |
123 | } | |
124 | ||
b78ea5fc NTND |
125 | static int estimate_similarity(struct repository *r, |
126 | struct diff_filespec *src, | |
427dcb4b | 127 | struct diff_filespec *dst, |
95acf11a JT |
128 | int minimum_score, |
129 | int skip_unmodified) | |
427dcb4b JH |
130 | { |
131 | /* src points at a file that existed in the original tree (or | |
132 | * optionally a file in the destination tree) and dst points | |
133 | * at a newly created file. They may be quite similar, in which | |
134 | * case we want to say src is renamed to dst or src is copied into | |
135 | * dst, and then some edit has been applied to dst. | |
136 | * | |
137 | * Compare them and return how similar they are, representing | |
f0c6b2a2 JH |
138 | * the score as an integer between 0 and MAX_SCORE. |
139 | * | |
140 | * When there is an exact match, it is considered a better | |
141 | * match than anything else; the destination does not even | |
142 | * call into this function in that case. | |
427dcb4b | 143 | */ |
90bd932c | 144 | unsigned long max_size, delta_size, base_size, src_copied, literal_added; |
427dcb4b | 145 | int score; |
1c37e86a JT |
146 | struct diff_populate_filespec_options dpf_options = { |
147 | .check_size_only = 1 | |
148 | }; | |
95acf11a JT |
149 | struct prefetch_options prefetch_options = {r, skip_unmodified}; |
150 | ||
151 | if (r == the_repository && has_promisor_remote()) { | |
152 | dpf_options.missing_object_cb = prefetch; | |
153 | dpf_options.missing_object_data = &prefetch_options; | |
154 | } | |
427dcb4b | 155 | |
60896c7b JH |
156 | /* We deal only with regular files. Symlink renames are handled |
157 | * only when they are exact matches --- in other words, no edits | |
158 | * after renaming. | |
159 | */ | |
160 | if (!S_ISREG(src->mode) || !S_ISREG(dst->mode)) | |
161 | return 0; | |
162 | ||
81ac051d LT |
163 | /* |
164 | * Need to check that source and destination sizes are | |
165 | * filled in before comparing them. | |
166 | * | |
167 | * If we already have "cnt_data" filled in, we know it's | |
168 | * all good (avoid checking the size for zero, as that | |
169 | * is a possible size - we really should have a flag to | |
170 | * say whether the size is valid or not!) | |
171 | */ | |
8e5dd3d6 | 172 | if (!src->cnt_data && |
1c37e86a | 173 | diff_populate_filespec(r, src, &dpf_options)) |
81ac051d | 174 | return 0; |
8e5dd3d6 | 175 | if (!dst->cnt_data && |
1c37e86a | 176 | diff_populate_filespec(r, dst, &dpf_options)) |
81ac051d LT |
177 | return 0; |
178 | ||
90bd932c | 179 | max_size = ((src->size > dst->size) ? src->size : dst->size); |
58b103f5 | 180 | base_size = ((src->size < dst->size) ? src->size : dst->size); |
90bd932c | 181 | delta_size = max_size - base_size; |
427dcb4b | 182 | |
58b103f5 JH |
183 | /* We would not consider edits that change the file size so |
184 | * drastically. delta_size must be smaller than | |
cd1870ed | 185 | * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size). |
f0c6b2a2 | 186 | * |
58b103f5 JH |
187 | * Note that base_size == 0 case is handled here already |
188 | * and the final score computation below would not have a | |
189 | * divide-by-zero issue. | |
427dcb4b | 190 | */ |
3a4d6769 | 191 | if (max_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE) |
427dcb4b JH |
192 | return 0; |
193 | ||
95acf11a JT |
194 | dpf_options.check_size_only = 0; |
195 | ||
196 | if (!src->cnt_data && diff_populate_filespec(r, src, &dpf_options)) | |
885c716f | 197 | return 0; |
95acf11a | 198 | if (!dst->cnt_data && diff_populate_filespec(r, dst, &dpf_options)) |
885c716f BS |
199 | return 0; |
200 | ||
b78ea5fc | 201 | if (diffcore_count_changes(r, src, dst, |
c06c7966 | 202 | &src->cnt_data, &dst->cnt_data, |
65416758 | 203 | &src_copied, &literal_added)) |
85976974 | 204 | return 0; |
355e76a4 | 205 | |
1706306a JH |
206 | /* How similar are they? |
207 | * what percentage of material in dst are from source? | |
427dcb4b | 208 | */ |
90bd932c | 209 | if (!dst->size) |
1706306a | 210 | score = 0; /* should not happen */ |
cfc0aef1 | 211 | else |
dc49cd76 | 212 | score = (int)(src_copied * MAX_SCORE / max_size); |
427dcb4b JH |
213 | return score; |
214 | } | |
215 | ||
5098bafb | 216 | static void record_rename_pair(int dst_index, int src_index, int score) |
427dcb4b | 217 | { |
9db2ac56 EN |
218 | struct diff_filepair *src = rename_src[src_index].p; |
219 | struct diff_filepair *dst = rename_dst[dst_index].p; | |
f7c1512a | 220 | |
9db2ac56 | 221 | if (dst->renamed_pair) |
25d5ea41 | 222 | die("internal error: dst already matched."); |
427dcb4b | 223 | |
9db2ac56 EN |
224 | src->one->rename_used++; |
225 | src->one->count++; | |
427dcb4b | 226 | |
9db2ac56 EN |
227 | rename_dst[dst_index].filespec_to_free = dst->one; |
228 | rename_dst[dst_index].is_rename = 1; | |
427dcb4b | 229 | |
9db2ac56 EN |
230 | dst->one = src->one; |
231 | dst->renamed_pair = 1; | |
232 | if (!strcmp(dst->one->path, dst->two->path)) | |
233 | dst->score = rename_src[src_index].score; | |
fc580719 | 234 | else |
9db2ac56 | 235 | dst->score = score; |
427dcb4b JH |
236 | } |
237 | ||
238 | /* | |
239 | * We sort the rename similarity matrix with the score, in descending | |
15d061b4 | 240 | * order (the most similar first). |
427dcb4b JH |
241 | */ |
242 | static int score_compare(const void *a_, const void *b_) | |
243 | { | |
244 | const struct diff_score *a = a_, *b = b_; | |
cfc0aef1 | 245 | |
6d24ad97 JH |
246 | /* sink the unused ones to the bottom */ |
247 | if (a->dst < 0) | |
248 | return (0 <= b->dst); | |
249 | else if (b->dst < 0) | |
250 | return -1; | |
251 | ||
cfc0aef1 RS |
252 | if (a->score == b->score) |
253 | return b->name_score - a->name_score; | |
254 | ||
427dcb4b JH |
255 | return b->score - a->score; |
256 | } | |
257 | ||
9027f53c | 258 | struct file_similarity { |
f79d9c58 | 259 | struct hashmap_entry entry; |
7c85f8ac | 260 | int index; |
9027f53c | 261 | struct diff_filespec *filespec; |
9027f53c LT |
262 | }; |
263 | ||
b78ea5fc NTND |
264 | static unsigned int hash_filespec(struct repository *r, |
265 | struct diff_filespec *filespec) | |
48f6407f | 266 | { |
41c9560e | 267 | if (!filespec->oid_valid) { |
1c37e86a | 268 | if (diff_populate_filespec(r, filespec, NULL)) |
48f6407f | 269 | return 0; |
2dcde20e MT |
270 | hash_object_file(r->hash_algo, filespec->data, filespec->size, |
271 | "blob", &filespec->oid); | |
48f6407f | 272 | } |
d40abc8e | 273 | return oidhash(&filespec->oid); |
48f6407f KB |
274 | } |
275 | ||
f79d9c58 | 276 | static int find_identical_files(struct hashmap *srcs, |
7c85f8ac | 277 | int dst_index, |
11f944dd | 278 | struct diff_options *options) |
9027f53c LT |
279 | { |
280 | int renames = 0; | |
9db2ac56 | 281 | struct diff_filespec *target = rename_dst[dst_index].p->two; |
ab73a9d1 | 282 | struct file_similarity *p, *best = NULL; |
48f6407f | 283 | int i = 100, best_score = -1; |
f0e63c41 | 284 | unsigned int hash = hash_filespec(options->repo, target); |
48f6407f KB |
285 | |
286 | /* | |
7c85f8ac | 287 | * Find the best source match for specified destination. |
48f6407f | 288 | */ |
f0e63c41 EW |
289 | p = hashmap_get_entry_from_hash(srcs, hash, NULL, |
290 | struct file_similarity, entry); | |
23dee69f | 291 | hashmap_for_each_entry_from(srcs, p, entry) { |
48f6407f KB |
292 | int score; |
293 | struct diff_filespec *source = p->filespec; | |
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 | 364 | /* Free the hash data structure and entries */ |
6da1a258 | 365 | hashmap_clear_and_free(&file_table, struct file_similarity, entry); |
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 | */ | |
00b8cccd | 392 | static int too_many_rename_candidates(int num_destinations, int num_sources, |
9d8a5a50 JH |
393 | struct diff_options *options) |
394 | { | |
395 | int rename_limit = options->rename_limit; | |
00b8cccd | 396 | int i, limited_sources; |
9d8a5a50 JH |
397 | |
398 | options->needed_rename_limit = 0; | |
399 | ||
400 | /* | |
401 | * This basically does a test for the rename matrix not | |
402 | * growing larger than a "rename_limit" square matrix, ie: | |
403 | * | |
00b8cccd | 404 | * num_destinations * num_sources > rename_limit * rename_limit |
ad8a1be5 EN |
405 | * |
406 | * We use st_mult() to check overflow conditions; in the | |
407 | * exceptional circumstance that size_t isn't large enough to hold | |
408 | * the multiplication, the system won't be able to allocate enough | |
409 | * memory for the matrix anyway. | |
9d8a5a50 | 410 | */ |
89973554 JT |
411 | if (rename_limit <= 0) |
412 | rename_limit = 32767; | |
ad8a1be5 EN |
413 | if (st_mult(num_destinations, num_sources) |
414 | <= st_mult(rename_limit, rename_limit)) | |
9d8a5a50 JH |
415 | return 0; |
416 | ||
417 | options->needed_rename_limit = | |
00b8cccd | 418 | num_sources > num_destinations ? num_sources : num_destinations; |
f31027c9 JH |
419 | |
420 | /* Are we running under -C -C? */ | |
0d1e0e78 | 421 | if (!options->flags.find_copies_harder) |
f31027c9 JH |
422 | return 1; |
423 | ||
424 | /* Would we bust the limit if we were running under -C? */ | |
00b8cccd | 425 | for (limited_sources = i = 0; i < num_sources; i++) { |
f31027c9 JH |
426 | if (diff_unmodified_pair(rename_src[i].p)) |
427 | continue; | |
00b8cccd | 428 | limited_sources++; |
f31027c9 | 429 | } |
ad8a1be5 EN |
430 | if (st_mult(num_destinations, limited_sources) |
431 | <= st_mult(rename_limit, rename_limit)) | |
f31027c9 | 432 | return 2; |
9d8a5a50 JH |
433 | return 1; |
434 | } | |
435 | ||
0940e5f2 LT |
436 | static int find_renames(struct diff_score *mx, int dst_cnt, int minimum_score, int copies) |
437 | { | |
438 | int count = 0, i; | |
439 | ||
440 | for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) { | |
441 | struct diff_rename_dst *dst; | |
442 | ||
443 | if ((mx[i].dst < 0) || | |
444 | (mx[i].score < minimum_score)) | |
445 | break; /* there is no more usable pair. */ | |
446 | dst = &rename_dst[mx[i].dst]; | |
9db2ac56 | 447 | if (dst->is_rename) |
0940e5f2 | 448 | continue; /* already done, either exact or fuzzy. */ |
e88d6bc6 | 449 | if (!copies && rename_src[mx[i].src].p->one->rename_used) |
0940e5f2 LT |
450 | continue; |
451 | record_rename_pair(mx[i].dst, mx[i].src, mx[i].score); | |
452 | count++; | |
453 | } | |
454 | return count; | |
455 | } | |
456 | ||
8082d8d3 | 457 | void diffcore_rename(struct diff_options *options) |
427dcb4b | 458 | { |
8082d8d3 JH |
459 | int detect_rename = options->detect_rename; |
460 | int minimum_score = options->rename_score; | |
38c6f780 | 461 | struct diff_queue_struct *q = &diff_queued_diff; |
5098bafb | 462 | struct diff_queue_struct outq; |
427dcb4b | 463 | struct diff_score *mx; |
f31027c9 | 464 | int i, j, rename_count, skip_unmodified = 0; |
26a66a6b | 465 | int num_destinations, dst_cnt; |
3ac942d4 | 466 | struct progress *progress = NULL; |
427dcb4b | 467 | |
26dee0ad | 468 | if (!minimum_score) |
f345b0a0 | 469 | minimum_score = DEFAULT_RENAME_SCORE; |
427dcb4b | 470 | |
427dcb4b | 471 | for (i = 0; i < q->nr; i++) { |
52e95789 | 472 | struct diff_filepair *p = q->queue[i]; |
2f3f8b21 | 473 | if (!DIFF_FILE_VALID(p->one)) { |
81e50eab | 474 | if (!DIFF_FILE_VALID(p->two)) |
60896c7b | 475 | continue; /* unmerged */ |
2f3f8b21 JH |
476 | else if (options->single_follow && |
477 | strcmp(options->single_follow, p->two->path)) | |
478 | continue; /* not interested */ | |
0d1e0e78 | 479 | else if (!options->flags.rename_empty && |
02491b67 | 480 | is_empty_blob_oid(&p->two->oid)) |
90d43b07 | 481 | continue; |
9db2ac56 | 482 | else if (add_rename_dst(p) < 0) { |
4d6be03b JK |
483 | warning("skipping rename detection, detected" |
484 | " duplicate destination '%s'", | |
485 | p->two->path); | |
486 | goto cleanup; | |
487 | } | |
2f3f8b21 | 488 | } |
0d1e0e78 | 489 | else if (!options->flags.rename_empty && |
02491b67 | 490 | is_empty_blob_oid(&p->one->oid)) |
90d43b07 | 491 | continue; |
d7c9bf22 | 492 | else if (!DIFF_PAIR_UNMERGED(p) && !DIFF_FILE_VALID(p->two)) { |
64479711 LT |
493 | /* |
494 | * If the source is a broken "delete", and | |
2210100a JH |
495 | * they did not really want to get broken, |
496 | * that means the source actually stays. | |
64479711 LT |
497 | * So we increment the "rename_used" score |
498 | * by one, to indicate ourselves as a user | |
499 | */ | |
500 | if (p->broken_pair && !p->score) | |
501 | p->one->rename_used++; | |
e88d6bc6 | 502 | register_rename_src(p); |
64479711 LT |
503 | } |
504 | else if (detect_rename == DIFF_DETECT_COPY) { | |
505 | /* | |
506 | * Increment the "rename_used" score by | |
507 | * one, to indicate ourselves as a user. | |
2210100a | 508 | */ |
64479711 | 509 | p->one->rename_used++; |
e88d6bc6 | 510 | register_rename_src(p); |
2210100a | 511 | } |
427dcb4b | 512 | } |
0024a549 | 513 | if (rename_dst_nr == 0 || rename_src_nr == 0) |
427dcb4b JH |
514 | goto cleanup; /* nothing to do */ |
515 | ||
17559a64 LT |
516 | /* |
517 | * We really want to cull the candidates list early | |
518 | * with cheap tests in order to avoid doing deltas. | |
519 | */ | |
11f944dd | 520 | rename_count = find_exact_renames(options); |
17559a64 | 521 | |
42899ac8 LT |
522 | /* Did we only want exact renames? */ |
523 | if (minimum_score == MAX_SCORE) | |
524 | goto cleanup; | |
525 | ||
526 | /* | |
527 | * Calculate how many renames are left (but all the source | |
528 | * files still remain as options for rename/copies!) | |
529 | */ | |
26a66a6b | 530 | num_destinations = (rename_dst_nr - rename_count); |
42899ac8 LT |
531 | |
532 | /* All done? */ | |
26a66a6b | 533 | if (!num_destinations) |
42899ac8 LT |
534 | goto cleanup; |
535 | ||
00b8cccd EN |
536 | switch (too_many_rename_candidates(num_destinations, rename_src_nr, |
537 | options)) { | |
f31027c9 | 538 | case 1: |
0024a549 | 539 | goto cleanup; |
f31027c9 JH |
540 | case 2: |
541 | options->degraded_cc_to_c = 1; | |
542 | skip_unmodified = 1; | |
543 | break; | |
544 | default: | |
545 | break; | |
546 | } | |
0024a549 | 547 | |
3ac942d4 | 548 | if (options->show_rename_progress) { |
8aade107 | 549 | progress = start_delayed_progress( |
754dbc43 | 550 | _("Performing inexact rename detection"), |
81c4bf02 | 551 | (uint64_t)num_destinations * (uint64_t)rename_src_nr); |
3ac942d4 JK |
552 | } |
553 | ||
26a66a6b EN |
554 | mx = xcalloc(st_mult(NUM_CANDIDATE_PER_DST, num_destinations), |
555 | sizeof(*mx)); | |
25d5ea41 | 556 | for (dst_cnt = i = 0; i < rename_dst_nr; i++) { |
9db2ac56 | 557 | struct diff_filespec *two = rename_dst[i].p->two; |
6d24ad97 JH |
558 | struct diff_score *m; |
559 | ||
9db2ac56 | 560 | if (rename_dst[i].is_rename) |
427dcb4b | 561 | continue; /* dealt with exact match already. */ |
6d24ad97 JH |
562 | |
563 | m = &mx[dst_cnt * NUM_CANDIDATE_PER_DST]; | |
564 | for (j = 0; j < NUM_CANDIDATE_PER_DST; j++) | |
565 | m[j].dst = -1; | |
566 | ||
25d5ea41 | 567 | for (j = 0; j < rename_src_nr; j++) { |
e88d6bc6 | 568 | struct diff_filespec *one = rename_src[j].p->one; |
6d24ad97 | 569 | struct diff_score this_src; |
f31027c9 JH |
570 | |
571 | if (skip_unmodified && | |
572 | diff_unmodified_pair(rename_src[j].p)) | |
573 | continue; | |
574 | ||
b78ea5fc NTND |
575 | this_src.score = estimate_similarity(options->repo, |
576 | one, two, | |
95acf11a JT |
577 | minimum_score, |
578 | skip_unmodified); | |
6d24ad97 JH |
579 | this_src.name_score = basename_same(one, two); |
580 | this_src.dst = i; | |
581 | this_src.src = j; | |
582 | record_if_better(m, &this_src); | |
809809bb JH |
583 | /* |
584 | * Once we run estimate_similarity, | |
585 | * We do not need the text anymore. | |
586 | */ | |
8ae92e63 | 587 | diff_free_filespec_blob(one); |
809809bb | 588 | diff_free_filespec_blob(two); |
427dcb4b JH |
589 | } |
590 | dst_cnt++; | |
81c4bf02 EN |
591 | display_progress(progress, |
592 | (uint64_t)dst_cnt * (uint64_t)rename_src_nr); | |
427dcb4b | 593 | } |
3ac942d4 | 594 | stop_progress(&progress); |
6d24ad97 | 595 | |
427dcb4b | 596 | /* cost matrix sorted by most to least similar pair */ |
2049b8dc | 597 | STABLE_QSORT(mx, dst_cnt * NUM_CANDIDATE_PER_DST, score_compare); |
6d24ad97 | 598 | |
0940e5f2 LT |
599 | rename_count += find_renames(mx, dst_cnt, minimum_score, 0); |
600 | if (detect_rename == DIFF_DETECT_COPY) | |
601 | rename_count += find_renames(mx, dst_cnt, minimum_score, 1); | |
427dcb4b | 602 | free(mx); |
427dcb4b | 603 | |
15d061b4 | 604 | cleanup: |
427dcb4b | 605 | /* At this point, we have found some renames and copies and they |
5098bafb | 606 | * are recorded in rename_dst. The original list is still in *q. |
427dcb4b | 607 | */ |
9ca5df90 | 608 | DIFF_QUEUE_CLEAR(&outq); |
427dcb4b | 609 | for (i = 0; i < q->nr; i++) { |
25d5ea41 | 610 | struct diff_filepair *p = q->queue[i]; |
25d5ea41 JH |
611 | struct diff_filepair *pair_to_free = NULL; |
612 | ||
d7c9bf22 MZ |
613 | if (DIFF_PAIR_UNMERGED(p)) { |
614 | diff_q(&outq, p); | |
615 | } | |
616 | else if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) { | |
9db2ac56 EN |
617 | /* Creation */ |
618 | diff_q(&outq, p); | |
427dcb4b | 619 | } |
2cd68882 JH |
620 | else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) { |
621 | /* | |
622 | * Deletion | |
623 | * | |
f345b0a0 JH |
624 | * We would output this delete record if: |
625 | * | |
626 | * (1) this is a broken delete and the counterpart | |
627 | * broken create remains in the output; or | |
5098bafb JH |
628 | * (2) this is not a broken delete, and rename_dst |
629 | * does not have a rename/copy to move p->one->path | |
630 | * out of existence. | |
f345b0a0 JH |
631 | * |
632 | * Otherwise, the counterpart broken create | |
633 | * has been turned into a rename-edit; or | |
634 | * delete did not have a matching create to | |
635 | * begin with. | |
2cd68882 | 636 | */ |
f345b0a0 JH |
637 | if (DIFF_PAIR_BROKEN(p)) { |
638 | /* broken delete */ | |
9db2ac56 EN |
639 | struct diff_rename_dst *dst = locate_rename_dst(p); |
640 | if (!dst) | |
641 | BUG("tracking failed somehow; failed to find associated dst for broken pair"); | |
642 | if (dst->is_rename) | |
f345b0a0 JH |
643 | /* counterpart is now rename/copy */ |
644 | pair_to_free = p; | |
645 | } | |
646 | else { | |
64479711 | 647 | if (p->one->rename_used) |
f345b0a0 JH |
648 | /* this path remains */ |
649 | pair_to_free = p; | |
650 | } | |
2cd68882 | 651 | |
9db2ac56 | 652 | if (!pair_to_free) |
2cd68882 JH |
653 | diff_q(&outq, p); |
654 | } | |
25d5ea41 | 655 | else if (!diff_unmodified_pair(p)) |
15d061b4 | 656 | /* all the usual ones need to be kept */ |
25d5ea41 | 657 | diff_q(&outq, p); |
15d061b4 | 658 | else |
9db2ac56 | 659 | /* no need to keep unmodified pairs; FIXME: remove earlier? */ |
15d061b4 JH |
660 | pair_to_free = p; |
661 | ||
226406f6 JH |
662 | if (pair_to_free) |
663 | diff_free_filepair(pair_to_free); | |
427dcb4b | 664 | } |
25d5ea41 | 665 | diff_debug_queue("done copying original", &outq); |
427dcb4b | 666 | |
25d5ea41 JH |
667 | free(q->queue); |
668 | *q = outq; | |
669 | diff_debug_queue("done collapsing", q); | |
427dcb4b | 670 | |
9fb88419 | 671 | for (i = 0; i < rename_dst_nr; i++) |
9db2ac56 EN |
672 | if (rename_dst[i].filespec_to_free) |
673 | free_filespec(rename_dst[i].filespec_to_free); | |
5098bafb | 674 | |
6a83d902 | 675 | FREE_AND_NULL(rename_dst); |
25d5ea41 | 676 | rename_dst_nr = rename_dst_alloc = 0; |
6a83d902 | 677 | FREE_AND_NULL(rename_src); |
25d5ea41 | 678 | rename_src_nr = rename_src_alloc = 0; |
9db2ac56 EN |
679 | if (break_idx) { |
680 | strintmap_clear(break_idx); | |
681 | FREE_AND_NULL(break_idx); | |
682 | } | |
427dcb4b JH |
683 | return; |
684 | } |