]>
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" | |
f79d9c58 | 7 | #include "hashmap.h" |
3ac942d4 | 8 | #include "progress.h" |
427dcb4b | 9 | |
25d5ea41 JH |
10 | /* Table of rename/copy destinations */ |
11 | ||
12 | static struct diff_rename_dst { | |
13 | struct diff_filespec *two; | |
14 | struct diff_filepair *pair; | |
15 | } *rename_dst; | |
16 | static int rename_dst_nr, rename_dst_alloc; | |
427dcb4b | 17 | |
f98c2f7e | 18 | static int find_rename_dst(struct diff_filespec *two) |
427dcb4b | 19 | { |
25d5ea41 JH |
20 | int first, last; |
21 | ||
22 | first = 0; | |
23 | last = rename_dst_nr; | |
24 | while (last > first) { | |
25 | int next = (last + first) >> 1; | |
26 | struct diff_rename_dst *dst = &(rename_dst[next]); | |
27 | int cmp = strcmp(two->path, dst->two->path); | |
28 | if (!cmp) | |
f98c2f7e | 29 | return next; |
25d5ea41 JH |
30 | if (cmp < 0) { |
31 | last = next; | |
32 | continue; | |
33 | } | |
34 | first = next+1; | |
35 | } | |
f98c2f7e JK |
36 | return -first - 1; |
37 | } | |
38 | ||
39 | static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two) | |
40 | { | |
41 | int ofs = find_rename_dst(two); | |
42 | return ofs < 0 ? NULL : &rename_dst[ofs]; | |
43 | } | |
44 | ||
45 | /* | |
46 | * Returns 0 on success, -1 if we found a duplicate. | |
47 | */ | |
48 | static int add_rename_dst(struct diff_filespec *two) | |
49 | { | |
50 | int first = find_rename_dst(two); | |
51 | ||
52 | if (first >= 0) | |
53 | return -1; | |
54 | first = -first - 1; | |
55 | ||
25d5ea41 | 56 | /* insert to make it at "first" */ |
337ce247 | 57 | ALLOC_GROW(rename_dst, rename_dst_nr + 1, rename_dst_alloc); |
25d5ea41 JH |
58 | rename_dst_nr++; |
59 | if (first < rename_dst_nr) | |
60 | memmove(rename_dst + first + 1, rename_dst + first, | |
61 | (rename_dst_nr - first - 1) * sizeof(*rename_dst)); | |
5098bafb | 62 | rename_dst[first].two = alloc_filespec(two->path); |
41c9560e | 63 | fill_filespec(rename_dst[first].two, two->oid.hash, two->oid_valid, |
a0d12c44 | 64 | two->mode); |
25d5ea41 | 65 | rename_dst[first].pair = NULL; |
f98c2f7e | 66 | return 0; |
427dcb4b JH |
67 | } |
68 | ||
15d061b4 | 69 | /* Table of rename/copy src files */ |
25d5ea41 | 70 | static struct diff_rename_src { |
e88d6bc6 | 71 | struct diff_filepair *p; |
fc580719 | 72 | unsigned short score; /* to remember the break score */ |
25d5ea41 JH |
73 | } *rename_src; |
74 | static int rename_src_nr, rename_src_alloc; | |
427dcb4b | 75 | |
e88d6bc6 | 76 | static struct diff_rename_src *register_rename_src(struct diff_filepair *p) |
25d5ea41 JH |
77 | { |
78 | int first, last; | |
e88d6bc6 JH |
79 | struct diff_filespec *one = p->one; |
80 | unsigned short score = p->score; | |
25d5ea41 JH |
81 | |
82 | first = 0; | |
83 | last = rename_src_nr; | |
84 | while (last > first) { | |
85 | int next = (last + first) >> 1; | |
86 | struct diff_rename_src *src = &(rename_src[next]); | |
e88d6bc6 | 87 | int cmp = strcmp(one->path, src->p->one->path); |
25d5ea41 JH |
88 | if (!cmp) |
89 | return src; | |
90 | if (cmp < 0) { | |
91 | last = next; | |
92 | continue; | |
93 | } | |
94 | first = next+1; | |
95 | } | |
15d061b4 | 96 | |
25d5ea41 | 97 | /* insert to make it at "first" */ |
337ce247 | 98 | ALLOC_GROW(rename_src, rename_src_nr + 1, rename_src_alloc); |
25d5ea41 JH |
99 | rename_src_nr++; |
100 | if (first < rename_src_nr) | |
101 | memmove(rename_src + first + 1, rename_src + first, | |
102 | (rename_src_nr - first - 1) * sizeof(*rename_src)); | |
e88d6bc6 | 103 | rename_src[first].p = p; |
fc580719 | 104 | rename_src[first].score = score; |
25d5ea41 | 105 | return &(rename_src[first]); |
427dcb4b JH |
106 | } |
107 | ||
0ce39643 JS |
108 | static int basename_same(struct diff_filespec *src, struct diff_filespec *dst) |
109 | { | |
110 | int src_len = strlen(src->path), dst_len = strlen(dst->path); | |
111 | while (src_len && dst_len) { | |
112 | char c1 = src->path[--src_len]; | |
113 | char c2 = dst->path[--dst_len]; | |
114 | if (c1 != c2) | |
115 | return 0; | |
116 | if (c1 == '/') | |
117 | return 1; | |
118 | } | |
119 | return (!src_len || src->path[src_len - 1] == '/') && | |
120 | (!dst_len || dst->path[dst_len - 1] == '/'); | |
121 | } | |
122 | ||
427dcb4b | 123 | struct diff_score { |
25d5ea41 JH |
124 | int src; /* index in rename_src */ |
125 | int dst; /* index in rename_dst */ | |
6d24ad97 JH |
126 | unsigned short score; |
127 | short name_score; | |
427dcb4b JH |
128 | }; |
129 | ||
130 | static int estimate_similarity(struct diff_filespec *src, | |
131 | struct diff_filespec *dst, | |
132 | int minimum_score) | |
133 | { | |
134 | /* src points at a file that existed in the original tree (or | |
135 | * optionally a file in the destination tree) and dst points | |
136 | * at a newly created file. They may be quite similar, in which | |
137 | * case we want to say src is renamed to dst or src is copied into | |
138 | * dst, and then some edit has been applied to dst. | |
139 | * | |
140 | * Compare them and return how similar they are, representing | |
f0c6b2a2 JH |
141 | * the score as an integer between 0 and MAX_SCORE. |
142 | * | |
143 | * When there is an exact match, it is considered a better | |
144 | * match than anything else; the destination does not even | |
145 | * call into this function in that case. | |
427dcb4b | 146 | */ |
90bd932c | 147 | unsigned long max_size, delta_size, base_size, src_copied, literal_added; |
75c660ac | 148 | unsigned long delta_limit; |
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 | ||
dc49cd76 SP |
194 | delta_limit = (unsigned long) |
195 | (base_size * (MAX_SCORE-minimum_score) / MAX_SCORE); | |
d8c3d03a | 196 | if (diffcore_count_changes(src, dst, |
c06c7966 | 197 | &src->cnt_data, &dst->cnt_data, |
65416758 JH |
198 | delta_limit, |
199 | &src_copied, &literal_added)) | |
85976974 | 200 | return 0; |
355e76a4 | 201 | |
1706306a JH |
202 | /* How similar are they? |
203 | * what percentage of material in dst are from source? | |
427dcb4b | 204 | */ |
90bd932c | 205 | if (!dst->size) |
1706306a | 206 | score = 0; /* should not happen */ |
cfc0aef1 | 207 | else |
dc49cd76 | 208 | score = (int)(src_copied * MAX_SCORE / max_size); |
427dcb4b JH |
209 | return score; |
210 | } | |
211 | ||
5098bafb | 212 | static void record_rename_pair(int dst_index, int src_index, int score) |
427dcb4b | 213 | { |
9fb88419 | 214 | struct diff_filespec *src, *dst; |
25d5ea41 | 215 | struct diff_filepair *dp; |
f7c1512a | 216 | |
25d5ea41 JH |
217 | if (rename_dst[dst_index].pair) |
218 | die("internal error: dst already matched."); | |
427dcb4b | 219 | |
e88d6bc6 | 220 | src = rename_src[src_index].p->one; |
64479711 | 221 | src->rename_used++; |
9fb88419 | 222 | src->count++; |
427dcb4b | 223 | |
25d5ea41 | 224 | dst = rename_dst[dst_index].two; |
9fb88419 | 225 | dst->count++; |
427dcb4b | 226 | |
9fb88419 | 227 | dp = diff_queue(NULL, src, dst); |
ef677686 | 228 | dp->renamed_pair = 1; |
fc580719 JH |
229 | if (!strcmp(src->path, dst->path)) |
230 | dp->score = rename_src[src_index].score; | |
231 | else | |
232 | dp->score = score; | |
25d5ea41 | 233 | rename_dst[dst_index].pair = dp; |
427dcb4b JH |
234 | } |
235 | ||
236 | /* | |
237 | * We sort the rename similarity matrix with the score, in descending | |
15d061b4 | 238 | * order (the most similar first). |
427dcb4b JH |
239 | */ |
240 | static int score_compare(const void *a_, const void *b_) | |
241 | { | |
242 | const struct diff_score *a = a_, *b = b_; | |
cfc0aef1 | 243 | |
6d24ad97 JH |
244 | /* sink the unused ones to the bottom */ |
245 | if (a->dst < 0) | |
246 | return (0 <= b->dst); | |
247 | else if (b->dst < 0) | |
248 | return -1; | |
249 | ||
cfc0aef1 RS |
250 | if (a->score == b->score) |
251 | return b->name_score - a->name_score; | |
252 | ||
427dcb4b JH |
253 | return b->score - a->score; |
254 | } | |
255 | ||
9027f53c | 256 | struct file_similarity { |
f79d9c58 | 257 | struct hashmap_entry entry; |
7c85f8ac | 258 | int index; |
9027f53c | 259 | struct diff_filespec *filespec; |
9027f53c LT |
260 | }; |
261 | ||
48f6407f KB |
262 | static unsigned int hash_filespec(struct diff_filespec *filespec) |
263 | { | |
41c9560e | 264 | if (!filespec->oid_valid) { |
48f6407f KB |
265 | if (diff_populate_filespec(filespec, 0)) |
266 | return 0; | |
a0d12c44 | 267 | hash_sha1_file(filespec->data, filespec->size, "blob", |
268 | filespec->oid.hash); | |
48f6407f | 269 | } |
a0d12c44 | 270 | return sha1hash(filespec->oid.hash); |
48f6407f KB |
271 | } |
272 | ||
f79d9c58 | 273 | static int find_identical_files(struct hashmap *srcs, |
7c85f8ac | 274 | int dst_index, |
11f944dd | 275 | struct diff_options *options) |
9027f53c LT |
276 | { |
277 | int renames = 0; | |
278 | ||
7c85f8ac | 279 | struct diff_filespec *target = rename_dst[dst_index].two; |
ab73a9d1 | 280 | struct file_similarity *p, *best = NULL; |
48f6407f KB |
281 | int i = 100, best_score = -1; |
282 | ||
283 | /* | |
7c85f8ac | 284 | * Find the best source match for specified destination. |
48f6407f | 285 | */ |
ab73a9d1 KB |
286 | p = hashmap_get_from_hash(srcs, hash_filespec(target), NULL); |
287 | for (; p; p = hashmap_get_next(srcs, p)) { | |
48f6407f KB |
288 | int score; |
289 | struct diff_filespec *source = p->filespec; | |
290 | ||
291 | /* False hash collision? */ | |
a0d12c44 | 292 | if (oidcmp(&source->oid, &target->oid)) |
48f6407f KB |
293 | continue; |
294 | /* Non-regular files? If so, the modes must match! */ | |
295 | if (!S_ISREG(source->mode) || !S_ISREG(target->mode)) { | |
296 | if (source->mode != target->mode) | |
0940e5f2 | 297 | continue; |
9027f53c | 298 | } |
48f6407f KB |
299 | /* Give higher scores to sources that haven't been used already */ |
300 | score = !source->rename_used; | |
301 | if (source->rename_used && options->detect_rename != DIFF_DETECT_COPY) | |
302 | continue; | |
303 | score += basename_same(source, target); | |
304 | if (score > best_score) { | |
305 | best = p; | |
306 | best_score = score; | |
307 | if (score == 2) | |
308 | break; | |
9027f53c | 309 | } |
48f6407f KB |
310 | |
311 | /* Too many identical alternatives? Pick one */ | |
312 | if (!--i) | |
313 | break; | |
314 | } | |
315 | if (best) { | |
7c85f8ac | 316 | record_rename_pair(dst_index, best->index, MAX_SCORE); |
48f6407f KB |
317 | renames++; |
318 | } | |
9027f53c LT |
319 | return renames; |
320 | } | |
321 | ||
f79d9c58 | 322 | static void insert_file_table(struct hashmap *table, int index, struct diff_filespec *filespec) |
9027f53c | 323 | { |
9027f53c LT |
324 | struct file_similarity *entry = xmalloc(sizeof(*entry)); |
325 | ||
9027f53c LT |
326 | entry->index = index; |
327 | entry->filespec = filespec; | |
9027f53c | 328 | |
f79d9c58 KB |
329 | hashmap_entry_init(entry, hash_filespec(filespec)); |
330 | hashmap_add(table, entry); | |
9027f53c LT |
331 | } |
332 | ||
cb1491b6 LT |
333 | /* |
334 | * Find exact renames first. | |
335 | * | |
336 | * The first round matches up the up-to-date entries, | |
337 | * and then during the second round we try to match | |
338 | * cache-dirty entries as well. | |
cb1491b6 | 339 | */ |
11f944dd | 340 | static int find_exact_renames(struct diff_options *options) |
cb1491b6 | 341 | { |
7c85f8ac | 342 | int i, renames = 0; |
f79d9c58 | 343 | struct hashmap file_table; |
cb1491b6 | 344 | |
ca4e3ca0 SG |
345 | /* Add all sources to the hash table in reverse order, because |
346 | * later on they will be retrieved in LIFO order. | |
347 | */ | |
f79d9c58 | 348 | hashmap_init(&file_table, NULL, rename_src_nr); |
ca4e3ca0 | 349 | for (i = rename_src_nr-1; i >= 0; i--) |
7c85f8ac | 350 | insert_file_table(&file_table, i, rename_src[i].p->one); |
9027f53c | 351 | |
7c85f8ac | 352 | /* Walk the destinations and find best source match */ |
9027f53c | 353 | for (i = 0; i < rename_dst_nr; i++) |
7c85f8ac | 354 | renames += find_identical_files(&file_table, i, options); |
9027f53c | 355 | |
f79d9c58 KB |
356 | /* Free the hash data structure and entries */ |
357 | hashmap_free(&file_table, 1); | |
9027f53c | 358 | |
7c85f8ac | 359 | return renames; |
cb1491b6 LT |
360 | } |
361 | ||
6d24ad97 JH |
362 | #define NUM_CANDIDATE_PER_DST 4 |
363 | static void record_if_better(struct diff_score m[], struct diff_score *o) | |
364 | { | |
365 | int i, worst; | |
366 | ||
367 | /* find the worst one */ | |
368 | worst = 0; | |
369 | for (i = 1; i < NUM_CANDIDATE_PER_DST; i++) | |
370 | if (score_compare(&m[i], &m[worst]) > 0) | |
371 | worst = i; | |
372 | ||
373 | /* is it better than the worst one? */ | |
374 | if (score_compare(&m[worst], o) > 0) | |
375 | m[worst] = *o; | |
376 | } | |
377 | ||
f31027c9 JH |
378 | /* |
379 | * Returns: | |
380 | * 0 if we are under the limit; | |
381 | * 1 if we need to disable inexact rename detection; | |
382 | * 2 if we would be under the limit if we were given -C instead of -C -C. | |
383 | */ | |
9d8a5a50 JH |
384 | static int too_many_rename_candidates(int num_create, |
385 | struct diff_options *options) | |
386 | { | |
387 | int rename_limit = options->rename_limit; | |
388 | int num_src = rename_src_nr; | |
f31027c9 | 389 | int i; |
9d8a5a50 JH |
390 | |
391 | options->needed_rename_limit = 0; | |
392 | ||
393 | /* | |
394 | * This basically does a test for the rename matrix not | |
395 | * growing larger than a "rename_limit" square matrix, ie: | |
396 | * | |
397 | * num_create * num_src > rename_limit * rename_limit | |
398 | * | |
399 | * but handles the potential overflow case specially (and we | |
400 | * assume at least 32-bit integers) | |
401 | */ | |
402 | if (rename_limit <= 0 || rename_limit > 32767) | |
403 | rename_limit = 32767; | |
404 | if ((num_create <= rename_limit || num_src <= rename_limit) && | |
405 | (num_create * num_src <= rename_limit * rename_limit)) | |
406 | return 0; | |
407 | ||
408 | options->needed_rename_limit = | |
409 | num_src > num_create ? num_src : num_create; | |
f31027c9 JH |
410 | |
411 | /* Are we running under -C -C? */ | |
412 | if (!DIFF_OPT_TST(options, FIND_COPIES_HARDER)) | |
413 | return 1; | |
414 | ||
415 | /* Would we bust the limit if we were running under -C? */ | |
416 | for (num_src = i = 0; i < rename_src_nr; i++) { | |
417 | if (diff_unmodified_pair(rename_src[i].p)) | |
418 | continue; | |
419 | num_src++; | |
420 | } | |
421 | if ((num_create <= rename_limit || num_src <= rename_limit) && | |
422 | (num_create * num_src <= rename_limit * rename_limit)) | |
423 | return 2; | |
9d8a5a50 JH |
424 | return 1; |
425 | } | |
426 | ||
0940e5f2 LT |
427 | static int find_renames(struct diff_score *mx, int dst_cnt, int minimum_score, int copies) |
428 | { | |
429 | int count = 0, i; | |
430 | ||
431 | for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) { | |
432 | struct diff_rename_dst *dst; | |
433 | ||
434 | if ((mx[i].dst < 0) || | |
435 | (mx[i].score < minimum_score)) | |
436 | break; /* there is no more usable pair. */ | |
437 | dst = &rename_dst[mx[i].dst]; | |
438 | if (dst->pair) | |
439 | continue; /* already done, either exact or fuzzy. */ | |
e88d6bc6 | 440 | if (!copies && rename_src[mx[i].src].p->one->rename_used) |
0940e5f2 LT |
441 | continue; |
442 | record_rename_pair(mx[i].dst, mx[i].src, mx[i].score); | |
443 | count++; | |
444 | } | |
445 | return count; | |
446 | } | |
447 | ||
8082d8d3 | 448 | void diffcore_rename(struct diff_options *options) |
427dcb4b | 449 | { |
8082d8d3 JH |
450 | int detect_rename = options->detect_rename; |
451 | int minimum_score = options->rename_score; | |
38c6f780 | 452 | struct diff_queue_struct *q = &diff_queued_diff; |
5098bafb | 453 | struct diff_queue_struct outq; |
427dcb4b | 454 | struct diff_score *mx; |
f31027c9 | 455 | int i, j, rename_count, skip_unmodified = 0; |
dabdbee1 | 456 | int num_create, dst_cnt; |
3ac942d4 | 457 | struct progress *progress = NULL; |
427dcb4b | 458 | |
26dee0ad | 459 | if (!minimum_score) |
f345b0a0 | 460 | minimum_score = DEFAULT_RENAME_SCORE; |
427dcb4b | 461 | |
427dcb4b | 462 | for (i = 0; i < q->nr; i++) { |
52e95789 | 463 | struct diff_filepair *p = q->queue[i]; |
2f3f8b21 | 464 | if (!DIFF_FILE_VALID(p->one)) { |
81e50eab | 465 | if (!DIFF_FILE_VALID(p->two)) |
60896c7b | 466 | continue; /* unmerged */ |
2f3f8b21 JH |
467 | else if (options->single_follow && |
468 | strcmp(options->single_follow, p->two->path)) | |
469 | continue; /* not interested */ | |
90d43b07 | 470 | else if (!DIFF_OPT_TST(options, RENAME_EMPTY) && |
a0d12c44 | 471 | is_empty_blob_sha1(p->two->oid.hash)) |
90d43b07 | 472 | continue; |
4d6be03b JK |
473 | else if (add_rename_dst(p->two) < 0) { |
474 | warning("skipping rename detection, detected" | |
475 | " duplicate destination '%s'", | |
476 | p->two->path); | |
477 | goto cleanup; | |
478 | } | |
2f3f8b21 | 479 | } |
90d43b07 | 480 | else if (!DIFF_OPT_TST(options, RENAME_EMPTY) && |
a0d12c44 | 481 | is_empty_blob_sha1(p->one->oid.hash)) |
90d43b07 | 482 | continue; |
d7c9bf22 | 483 | else if (!DIFF_PAIR_UNMERGED(p) && !DIFF_FILE_VALID(p->two)) { |
64479711 LT |
484 | /* |
485 | * If the source is a broken "delete", and | |
2210100a JH |
486 | * they did not really want to get broken, |
487 | * that means the source actually stays. | |
64479711 LT |
488 | * So we increment the "rename_used" score |
489 | * by one, to indicate ourselves as a user | |
490 | */ | |
491 | if (p->broken_pair && !p->score) | |
492 | p->one->rename_used++; | |
e88d6bc6 | 493 | register_rename_src(p); |
64479711 LT |
494 | } |
495 | else if (detect_rename == DIFF_DETECT_COPY) { | |
496 | /* | |
497 | * Increment the "rename_used" score by | |
498 | * one, to indicate ourselves as a user. | |
2210100a | 499 | */ |
64479711 | 500 | p->one->rename_used++; |
e88d6bc6 | 501 | register_rename_src(p); |
2210100a | 502 | } |
427dcb4b | 503 | } |
0024a549 | 504 | if (rename_dst_nr == 0 || rename_src_nr == 0) |
427dcb4b JH |
505 | goto cleanup; /* nothing to do */ |
506 | ||
17559a64 LT |
507 | /* |
508 | * We really want to cull the candidates list early | |
509 | * with cheap tests in order to avoid doing deltas. | |
510 | */ | |
11f944dd | 511 | rename_count = find_exact_renames(options); |
17559a64 | 512 | |
42899ac8 LT |
513 | /* Did we only want exact renames? */ |
514 | if (minimum_score == MAX_SCORE) | |
515 | goto cleanup; | |
516 | ||
517 | /* | |
518 | * Calculate how many renames are left (but all the source | |
519 | * files still remain as options for rename/copies!) | |
520 | */ | |
521 | num_create = (rename_dst_nr - rename_count); | |
42899ac8 LT |
522 | |
523 | /* All done? */ | |
524 | if (!num_create) | |
525 | goto cleanup; | |
526 | ||
f31027c9 JH |
527 | switch (too_many_rename_candidates(num_create, options)) { |
528 | case 1: | |
0024a549 | 529 | goto cleanup; |
f31027c9 JH |
530 | case 2: |
531 | options->degraded_cc_to_c = 1; | |
532 | skip_unmodified = 1; | |
533 | break; | |
534 | default: | |
535 | break; | |
536 | } | |
0024a549 | 537 | |
3ac942d4 JK |
538 | if (options->show_rename_progress) { |
539 | progress = start_progress_delay( | |
754dbc43 | 540 | _("Performing inexact rename detection"), |
3ac942d4 JK |
541 | rename_dst_nr * rename_src_nr, 50, 1); |
542 | } | |
543 | ||
50a6c8ef | 544 | mx = xcalloc(st_mult(num_create, NUM_CANDIDATE_PER_DST), sizeof(*mx)); |
25d5ea41 | 545 | for (dst_cnt = i = 0; i < rename_dst_nr; i++) { |
25d5ea41 | 546 | struct diff_filespec *two = rename_dst[i].two; |
6d24ad97 JH |
547 | struct diff_score *m; |
548 | ||
25d5ea41 | 549 | if (rename_dst[i].pair) |
427dcb4b | 550 | continue; /* dealt with exact match already. */ |
6d24ad97 JH |
551 | |
552 | m = &mx[dst_cnt * NUM_CANDIDATE_PER_DST]; | |
553 | for (j = 0; j < NUM_CANDIDATE_PER_DST; j++) | |
554 | m[j].dst = -1; | |
555 | ||
25d5ea41 | 556 | for (j = 0; j < rename_src_nr; j++) { |
e88d6bc6 | 557 | struct diff_filespec *one = rename_src[j].p->one; |
6d24ad97 | 558 | struct diff_score this_src; |
f31027c9 JH |
559 | |
560 | if (skip_unmodified && | |
561 | diff_unmodified_pair(rename_src[j].p)) | |
562 | continue; | |
563 | ||
6d24ad97 JH |
564 | this_src.score = estimate_similarity(one, two, |
565 | minimum_score); | |
566 | this_src.name_score = basename_same(one, two); | |
567 | this_src.dst = i; | |
568 | this_src.src = j; | |
569 | record_if_better(m, &this_src); | |
809809bb JH |
570 | /* |
571 | * Once we run estimate_similarity, | |
572 | * We do not need the text anymore. | |
573 | */ | |
8ae92e63 | 574 | diff_free_filespec_blob(one); |
809809bb | 575 | diff_free_filespec_blob(two); |
427dcb4b JH |
576 | } |
577 | dst_cnt++; | |
3ac942d4 | 578 | display_progress(progress, (i+1)*rename_src_nr); |
427dcb4b | 579 | } |
3ac942d4 | 580 | stop_progress(&progress); |
6d24ad97 | 581 | |
427dcb4b | 582 | /* cost matrix sorted by most to least similar pair */ |
6d24ad97 JH |
583 | qsort(mx, dst_cnt * NUM_CANDIDATE_PER_DST, sizeof(*mx), score_compare); |
584 | ||
0940e5f2 LT |
585 | rename_count += find_renames(mx, dst_cnt, minimum_score, 0); |
586 | if (detect_rename == DIFF_DETECT_COPY) | |
587 | rename_count += find_renames(mx, dst_cnt, minimum_score, 1); | |
427dcb4b | 588 | free(mx); |
427dcb4b | 589 | |
15d061b4 | 590 | cleanup: |
427dcb4b | 591 | /* At this point, we have found some renames and copies and they |
5098bafb | 592 | * are recorded in rename_dst. The original list is still in *q. |
427dcb4b | 593 | */ |
9ca5df90 | 594 | DIFF_QUEUE_CLEAR(&outq); |
427dcb4b | 595 | for (i = 0; i < q->nr; i++) { |
25d5ea41 | 596 | struct diff_filepair *p = q->queue[i]; |
25d5ea41 JH |
597 | struct diff_filepair *pair_to_free = NULL; |
598 | ||
d7c9bf22 MZ |
599 | if (DIFF_PAIR_UNMERGED(p)) { |
600 | diff_q(&outq, p); | |
601 | } | |
602 | else if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) { | |
2cd68882 JH |
603 | /* |
604 | * Creation | |
605 | * | |
606 | * We would output this create record if it has | |
607 | * not been turned into a rename/copy already. | |
608 | */ | |
f98c2f7e | 609 | struct diff_rename_dst *dst = locate_rename_dst(p->two); |
2cd68882 | 610 | if (dst && dst->pair) { |
25d5ea41 JH |
611 | diff_q(&outq, dst->pair); |
612 | pair_to_free = p; | |
613 | } | |
614 | else | |
2cd68882 JH |
615 | /* no matching rename/copy source, so |
616 | * record this as a creation. | |
25d5ea41 JH |
617 | */ |
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 */ | |
f98c2f7e | 639 | struct diff_rename_dst *dst = locate_rename_dst(p->one); |
f345b0a0 JH |
640 | if (dst && dst->pair) |
641 | /* counterpart is now rename/copy */ | |
642 | pair_to_free = p; | |
643 | } | |
644 | else { | |
64479711 | 645 | if (p->one->rename_used) |
f345b0a0 JH |
646 | /* this path remains */ |
647 | pair_to_free = p; | |
648 | } | |
2cd68882 JH |
649 | |
650 | if (pair_to_free) | |
651 | ; | |
652 | else | |
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 JH |
658 | else |
659 | /* no need to keep unmodified pairs */ | |
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 LT |
671 | for (i = 0; i < rename_dst_nr; i++) |
672 | free_filespec(rename_dst[i].two); | |
5098bafb | 673 | |
25d5ea41 JH |
674 | free(rename_dst); |
675 | rename_dst = NULL; | |
676 | rename_dst_nr = rename_dst_alloc = 0; | |
677 | free(rename_src); | |
678 | rename_src = NULL; | |
679 | rename_src_nr = rename_src_alloc = 0; | |
427dcb4b JH |
680 | return; |
681 | } |