]>
Commit | Line | Data |
---|---|---|
db106d01 | 1 | /* |
359fc2d2 | 2 | * Copyright (C) the libgit2 contributors. All rights reserved. |
db106d01 RB |
3 | * |
4 | * This file is part of libgit2, distributed under the GNU GPL v2 with | |
5 | * a Linking Exception. For full terms see the included COPYING file. | |
6 | */ | |
7 | #include "common.h" | |
114f5a6c | 8 | |
db106d01 | 9 | #include "git2/config.h" |
960a04dd | 10 | #include "git2/blob.h" |
737b5051 | 11 | #include "git2/sys/hashsig.h" |
114f5a6c RB |
12 | |
13 | #include "diff.h" | |
114f5a6c RB |
14 | #include "path.h" |
15 | #include "fileops.h" | |
9f77b3f6 | 16 | #include "config.h" |
db106d01 RB |
17 | |
18 | static git_diff_delta *diff_delta__dup( | |
19 | const git_diff_delta *d, git_pool *pool) | |
20 | { | |
21 | git_diff_delta *delta = git__malloc(sizeof(git_diff_delta)); | |
22 | if (!delta) | |
23 | return NULL; | |
24 | ||
25 | memcpy(delta, d, sizeof(git_diff_delta)); | |
c68b09dc | 26 | GIT_DIFF_FLAG__CLEAR_INTERNAL(delta->flags); |
db106d01 | 27 | |
d958e37a RB |
28 | if (d->old_file.path != NULL) { |
29 | delta->old_file.path = git_pool_strdup(pool, d->old_file.path); | |
30 | if (delta->old_file.path == NULL) | |
31 | goto fail; | |
32 | } | |
db106d01 | 33 | |
d958e37a | 34 | if (d->new_file.path != d->old_file.path && d->new_file.path != NULL) { |
db106d01 RB |
35 | delta->new_file.path = git_pool_strdup(pool, d->new_file.path); |
36 | if (delta->new_file.path == NULL) | |
37 | goto fail; | |
38 | } else { | |
39 | delta->new_file.path = delta->old_file.path; | |
40 | } | |
41 | ||
42 | return delta; | |
43 | ||
44 | fail: | |
45 | git__free(delta); | |
46 | return NULL; | |
47 | } | |
48 | ||
49 | static git_diff_delta *diff_delta__merge_like_cgit( | |
3940310e RB |
50 | const git_diff_delta *a, |
51 | const git_diff_delta *b, | |
52 | git_pool *pool) | |
db106d01 RB |
53 | { |
54 | git_diff_delta *dup; | |
55 | ||
56 | /* Emulate C git for merging two diffs (a la 'git diff <sha>'). | |
57 | * | |
58 | * When C git does a diff between the work dir and a tree, it actually | |
59 | * diffs with the index but uses the workdir contents. This emulates | |
60 | * those choices so we can emulate the type of diff. | |
61 | * | |
62 | * We have three file descriptions here, let's call them: | |
63 | * f1 = a->old_file | |
64 | * f2 = a->new_file AND b->old_file | |
65 | * f3 = b->new_file | |
66 | */ | |
67 | ||
68 | /* if f2 == f3 or f2 is deleted, then just dup the 'a' diff */ | |
69 | if (b->status == GIT_DELTA_UNMODIFIED || a->status == GIT_DELTA_DELETED) | |
70 | return diff_delta__dup(a, pool); | |
71 | ||
72 | /* otherwise, base this diff on the 'b' diff */ | |
73 | if ((dup = diff_delta__dup(b, pool)) == NULL) | |
74 | return NULL; | |
75 | ||
76 | /* If 'a' status is uninteresting, then we're done */ | |
77 | if (a->status == GIT_DELTA_UNMODIFIED) | |
78 | return dup; | |
79 | ||
80 | assert(a->status != GIT_DELTA_UNMODIFIED); | |
81 | assert(b->status != GIT_DELTA_UNMODIFIED); | |
82 | ||
83 | /* A cgit exception is that the diff of a file that is only in the | |
84 | * index (i.e. not in HEAD nor workdir) is given as empty. | |
85 | */ | |
86 | if (dup->status == GIT_DELTA_DELETED) { | |
b9780823 | 87 | if (a->status == GIT_DELTA_ADDED) { |
db106d01 | 88 | dup->status = GIT_DELTA_UNMODIFIED; |
b9780823 POL |
89 | dup->nfiles = 2; |
90 | } | |
db106d01 RB |
91 | /* else don't overwrite DELETE status */ |
92 | } else { | |
93 | dup->status = a->status; | |
b9780823 | 94 | dup->nfiles = a->nfiles; |
db106d01 RB |
95 | } |
96 | ||
9950bb4e | 97 | git_oid_cpy(&dup->old_file.id, &a->old_file.id); |
db106d01 RB |
98 | dup->old_file.mode = a->old_file.mode; |
99 | dup->old_file.size = a->old_file.size; | |
100 | dup->old_file.flags = a->old_file.flags; | |
101 | ||
102 | return dup; | |
103 | } | |
104 | ||
e7c85120 RB |
105 | static git_diff_delta *diff_delta__merge_like_cgit_reversed( |
106 | const git_diff_delta *a, | |
107 | const git_diff_delta *b, | |
108 | git_pool *pool) | |
109 | { | |
110 | git_diff_delta *dup; | |
111 | ||
112 | /* reversed version of above logic */ | |
113 | ||
114 | if (a->status == GIT_DELTA_UNMODIFIED) | |
115 | return diff_delta__dup(b, pool); | |
116 | ||
117 | if ((dup = diff_delta__dup(a, pool)) == NULL) | |
118 | return NULL; | |
119 | ||
61bef72d | 120 | if (b->status == GIT_DELTA_UNMODIFIED || b->status == GIT_DELTA_UNTRACKED || b->status == GIT_DELTA_UNREADABLE) |
e7c85120 RB |
121 | return dup; |
122 | ||
123 | if (dup->status == GIT_DELTA_DELETED) { | |
b9780823 | 124 | if (b->status == GIT_DELTA_ADDED) { |
e7c85120 | 125 | dup->status = GIT_DELTA_UNMODIFIED; |
b9780823 POL |
126 | dup->nfiles = 2; |
127 | } | |
e7c85120 RB |
128 | } else { |
129 | dup->status = b->status; | |
b9780823 | 130 | dup->nfiles = b->nfiles; |
e7c85120 RB |
131 | } |
132 | ||
9950bb4e | 133 | git_oid_cpy(&dup->old_file.id, &b->old_file.id); |
e7c85120 RB |
134 | dup->old_file.mode = b->old_file.mode; |
135 | dup->old_file.size = b->old_file.size; | |
136 | dup->old_file.flags = b->old_file.flags; | |
137 | ||
138 | return dup; | |
139 | } | |
140 | ||
141 | int git_diff_merge(git_diff *onto, const git_diff *from) | |
db106d01 RB |
142 | { |
143 | int error = 0; | |
144 | git_pool onto_pool; | |
145 | git_vector onto_new; | |
146 | git_diff_delta *delta; | |
e7c85120 | 147 | bool ignore_case, reversed; |
db106d01 RB |
148 | unsigned int i, j; |
149 | ||
150 | assert(onto && from); | |
151 | ||
152 | if (!from->deltas.length) | |
153 | return 0; | |
154 | ||
e7c85120 RB |
155 | ignore_case = ((onto->opts.flags & GIT_DIFF_IGNORE_CASE) != 0); |
156 | reversed = ((onto->opts.flags & GIT_DIFF_REVERSE) != 0); | |
157 | ||
158 | if (ignore_case != ((from->opts.flags & GIT_DIFF_IGNORE_CASE) != 0) || | |
159 | reversed != ((from->opts.flags & GIT_DIFF_REVERSE) != 0)) { | |
3940310e RB |
160 | giterr_set(GITERR_INVALID, |
161 | "Attempt to merge diffs created with conflicting options"); | |
162 | return -1; | |
163 | } | |
164 | ||
db106d01 RB |
165 | if (git_vector_init( |
166 | &onto_new, onto->deltas.length, git_diff_delta__cmp) < 0 || | |
167 | git_pool_init(&onto_pool, 1, 0) < 0) | |
168 | return -1; | |
169 | ||
db106d01 RB |
170 | for (i = 0, j = 0; i < onto->deltas.length || j < from->deltas.length; ) { |
171 | git_diff_delta *o = GIT_VECTOR_GET(&onto->deltas, i); | |
172 | const git_diff_delta *f = GIT_VECTOR_GET(&from->deltas, j); | |
3940310e RB |
173 | int cmp = !f ? -1 : !o ? 1 : |
174 | STRCMP_CASESELECT(ignore_case, o->old_file.path, f->old_file.path); | |
db106d01 RB |
175 | |
176 | if (cmp < 0) { | |
177 | delta = diff_delta__dup(o, &onto_pool); | |
178 | i++; | |
179 | } else if (cmp > 0) { | |
180 | delta = diff_delta__dup(f, &onto_pool); | |
181 | j++; | |
182 | } else { | |
e7c85120 RB |
183 | delta = reversed ? |
184 | diff_delta__merge_like_cgit_reversed(o, f, &onto_pool) : | |
185 | diff_delta__merge_like_cgit(o, f, &onto_pool); | |
db106d01 RB |
186 | i++; |
187 | j++; | |
188 | } | |
189 | ||
190 | /* the ignore rules for the target may not match the source | |
191 | * or the result of a merged delta could be skippable... | |
192 | */ | |
193 | if (git_diff_delta__should_skip(&onto->opts, delta)) { | |
194 | git__free(delta); | |
195 | continue; | |
196 | } | |
197 | ||
198 | if ((error = !delta ? -1 : git_vector_insert(&onto_new, delta)) < 0) | |
199 | break; | |
200 | } | |
201 | ||
202 | if (!error) { | |
203 | git_vector_swap(&onto->deltas, &onto_new); | |
204 | git_pool_swap(&onto->pool, &onto_pool); | |
3940310e RB |
205 | |
206 | if ((onto->opts.flags & GIT_DIFF_REVERSE) != 0) | |
207 | onto->old_src = from->old_src; | |
208 | else | |
209 | onto->new_src = from->new_src; | |
db106d01 RB |
210 | |
211 | /* prefix strings also come from old pool, so recreate those.*/ | |
212 | onto->opts.old_prefix = | |
213 | git_pool_strdup_safe(&onto->pool, onto->opts.old_prefix); | |
214 | onto->opts.new_prefix = | |
215 | git_pool_strdup_safe(&onto->pool, onto->opts.new_prefix); | |
216 | } | |
217 | ||
9cfce273 | 218 | git_vector_free_deep(&onto_new); |
db106d01 RB |
219 | git_pool_clear(&onto_pool); |
220 | ||
221 | return error; | |
222 | } | |
223 | ||
0462fba5 | 224 | int git_diff_find_similar__hashsig_for_file( |
f8275890 RB |
225 | void **out, const git_diff_file *f, const char *path, void *p) |
226 | { | |
c8893d1f | 227 | git_hashsig_option_t opt = (git_hashsig_option_t)(intptr_t)p; |
aa408cbf | 228 | |
f8275890 | 229 | GIT_UNUSED(f); |
36fc5497 | 230 | return git_hashsig_create_fromfile((git_hashsig **)out, path, opt); |
f8275890 | 231 | } |
9bc8be3d | 232 | |
0462fba5 | 233 | int git_diff_find_similar__hashsig_for_buf( |
f8275890 RB |
234 | void **out, const git_diff_file *f, const char *buf, size_t len, void *p) |
235 | { | |
c8893d1f | 236 | git_hashsig_option_t opt = (git_hashsig_option_t)(intptr_t)p; |
0462fba5 | 237 | |
f8275890 | 238 | GIT_UNUSED(f); |
36fc5497 | 239 | return git_hashsig_create((git_hashsig **)out, buf, len, opt); |
f8275890 | 240 | } |
9bc8be3d | 241 | |
0462fba5 | 242 | void git_diff_find_similar__hashsig_free(void *sig, void *payload) |
9bc8be3d RB |
243 | { |
244 | GIT_UNUSED(payload); | |
245 | git_hashsig_free(sig); | |
246 | } | |
247 | ||
0462fba5 | 248 | int git_diff_find_similar__calc_similarity( |
9bc8be3d RB |
249 | int *score, void *siga, void *sigb, void *payload) |
250 | { | |
36fc5497 POL |
251 | int error; |
252 | ||
9bc8be3d | 253 | GIT_UNUSED(payload); |
36fc5497 POL |
254 | error = git_hashsig_compare(siga, sigb); |
255 | if (error < 0) | |
256 | return error; | |
257 | ||
258 | *score = error; | |
9bc8be3d RB |
259 | return 0; |
260 | } | |
261 | ||
db106d01 RB |
262 | #define DEFAULT_THRESHOLD 50 |
263 | #define DEFAULT_BREAK_REWRITE_THRESHOLD 60 | |
a21cbb12 | 264 | #define DEFAULT_RENAME_LIMIT 200 |
db106d01 RB |
265 | |
266 | static int normalize_find_opts( | |
3ff1d123 | 267 | git_diff *diff, |
db106d01 | 268 | git_diff_find_options *opts, |
10672e3e | 269 | const git_diff_find_options *given) |
db106d01 RB |
270 | { |
271 | git_config *cfg = NULL; | |
36fc5497 | 272 | git_hashsig_option_t hashsig_opts; |
db106d01 | 273 | |
5a52d6be BS |
274 | GITERR_CHECK_VERSION(given, GIT_DIFF_FIND_OPTIONS_VERSION, "git_diff_find_options"); |
275 | ||
db106d01 RB |
276 | if (diff->repo != NULL && |
277 | git_repository_config__weakptr(&cfg, diff->repo) < 0) | |
278 | return -1; | |
279 | ||
7e3ed419 | 280 | if (given) |
db106d01 | 281 | memcpy(opts, given, sizeof(*opts)); |
db106d01 | 282 | |
c56c6d69 BS |
283 | if (!given || |
284 | (given->flags & GIT_DIFF_FIND_ALL) == GIT_DIFF_FIND_BY_CONFIG) | |
285 | { | |
fe21d708 | 286 | char *rule = |
9f77b3f6 RB |
287 | git_config__get_string_force(cfg, "diff.renames", "true"); |
288 | int boolval; | |
289 | ||
290 | if (!git__parse_bool(&boolval, rule) && !boolval) | |
291 | /* don't set FIND_RENAMES if bool value is false */; | |
292 | else if (!strcasecmp(rule, "copies") || !strcasecmp(rule, "copy")) | |
293 | opts->flags |= GIT_DIFF_FIND_RENAMES | GIT_DIFF_FIND_COPIES; | |
294 | else | |
295 | opts->flags |= GIT_DIFF_FIND_RENAMES; | |
fe21d708 CMN |
296 | |
297 | git__free(rule); | |
db106d01 | 298 | } |
ca901e7b | 299 | |
db106d01 RB |
300 | /* some flags imply others */ |
301 | ||
9be5be47 RB |
302 | if (opts->flags & GIT_DIFF_FIND_EXACT_MATCH_ONLY) { |
303 | /* if we are only looking for exact matches, then don't turn | |
304 | * MODIFIED items into ADD/DELETE pairs because it's too picky | |
305 | */ | |
306 | opts->flags &= ~(GIT_DIFF_FIND_REWRITES | GIT_DIFF_BREAK_REWRITES); | |
307 | ||
308 | /* similarly, don't look for self-rewrites to split */ | |
309 | opts->flags &= ~GIT_DIFF_FIND_RENAMES_FROM_REWRITES; | |
310 | } | |
311 | ||
db106d01 RB |
312 | if (opts->flags & GIT_DIFF_FIND_RENAMES_FROM_REWRITES) |
313 | opts->flags |= GIT_DIFF_FIND_RENAMES; | |
314 | ||
315 | if (opts->flags & GIT_DIFF_FIND_COPIES_FROM_UNMODIFIED) | |
316 | opts->flags |= GIT_DIFF_FIND_COPIES; | |
317 | ||
d958e37a RB |
318 | if (opts->flags & GIT_DIFF_BREAK_REWRITES) |
319 | opts->flags |= GIT_DIFF_FIND_REWRITES; | |
320 | ||
db106d01 RB |
321 | #define USE_DEFAULT(X) ((X) == 0 || (X) > 100) |
322 | ||
323 | if (USE_DEFAULT(opts->rename_threshold)) | |
324 | opts->rename_threshold = DEFAULT_THRESHOLD; | |
325 | ||
326 | if (USE_DEFAULT(opts->rename_from_rewrite_threshold)) | |
327 | opts->rename_from_rewrite_threshold = DEFAULT_THRESHOLD; | |
328 | ||
329 | if (USE_DEFAULT(opts->copy_threshold)) | |
330 | opts->copy_threshold = DEFAULT_THRESHOLD; | |
331 | ||
332 | if (USE_DEFAULT(opts->break_rewrite_threshold)) | |
333 | opts->break_rewrite_threshold = DEFAULT_BREAK_REWRITE_THRESHOLD; | |
334 | ||
335 | #undef USE_DEFAULT | |
336 | ||
a21cbb12 | 337 | if (!opts->rename_limit) { |
9f77b3f6 RB |
338 | opts->rename_limit = git_config__get_int_force( |
339 | cfg, "diff.renamelimit", DEFAULT_RENAME_LIMIT); | |
db106d01 | 340 | |
9f77b3f6 RB |
341 | if (opts->rename_limit <= 0) |
342 | opts->rename_limit = DEFAULT_RENAME_LIMIT; | |
db106d01 RB |
343 | } |
344 | ||
f8275890 | 345 | /* assign the internal metric with whitespace flag as payload */ |
9bc8be3d | 346 | if (!opts->metric) { |
f8275890 RB |
347 | opts->metric = git__malloc(sizeof(git_diff_similarity_metric)); |
348 | GITERR_CHECK_ALLOC(opts->metric); | |
349 | ||
0462fba5 ET |
350 | opts->metric->file_signature = git_diff_find_similar__hashsig_for_file; |
351 | opts->metric->buffer_signature = git_diff_find_similar__hashsig_for_buf; | |
352 | opts->metric->free_signature = git_diff_find_similar__hashsig_free; | |
353 | opts->metric->similarity = git_diff_find_similar__calc_similarity; | |
f8275890 | 354 | |
9bc8be3d | 355 | if (opts->flags & GIT_DIFF_FIND_IGNORE_WHITESPACE) |
36fc5497 | 356 | hashsig_opts = GIT_HASHSIG_IGNORE_WHITESPACE; |
9bc8be3d | 357 | else if (opts->flags & GIT_DIFF_FIND_DONT_IGNORE_WHITESPACE) |
36fc5497 | 358 | hashsig_opts = GIT_HASHSIG_NORMAL; |
9bc8be3d | 359 | else |
36fc5497 POL |
360 | hashsig_opts = GIT_HASHSIG_SMART_WHITESPACE; |
361 | hashsig_opts |= GIT_HASHSIG_ALLOW_SMALL_FILES; | |
362 | opts->metric->payload = (void *)hashsig_opts; | |
9bc8be3d RB |
363 | } |
364 | ||
db106d01 RB |
365 | return 0; |
366 | } | |
367 | ||
2123a17f RB |
368 | static int insert_delete_side_of_split( |
369 | git_diff *diff, git_vector *onto, const git_diff_delta *delta) | |
370 | { | |
371 | /* make new record for DELETED side of split */ | |
372 | git_diff_delta *deleted = diff_delta__dup(delta, &diff->pool); | |
373 | GITERR_CHECK_ALLOC(deleted); | |
374 | ||
375 | deleted->status = GIT_DELTA_DELETED; | |
376 | deleted->nfiles = 1; | |
377 | memset(&deleted->new_file, 0, sizeof(deleted->new_file)); | |
378 | deleted->new_file.path = deleted->old_file.path; | |
9950bb4e | 379 | deleted->new_file.flags |= GIT_DIFF_FLAG_VALID_ID; |
2123a17f RB |
380 | |
381 | return git_vector_insert(onto, deleted); | |
382 | } | |
383 | ||
d958e37a | 384 | static int apply_splits_and_deletes( |
3ff1d123 | 385 | git_diff *diff, size_t expected_size, bool actually_split) |
db106d01 RB |
386 | { |
387 | git_vector onto = GIT_VECTOR_INIT; | |
388 | size_t i; | |
2123a17f | 389 | git_diff_delta *delta; |
db106d01 RB |
390 | |
391 | if (git_vector_init(&onto, expected_size, git_diff_delta__cmp) < 0) | |
392 | return -1; | |
393 | ||
394 | /* build new delta list without TO_DELETE and splitting TO_SPLIT */ | |
395 | git_vector_foreach(&diff->deltas, i, delta) { | |
71a3d27e | 396 | if ((delta->flags & GIT_DIFF_FLAG__TO_DELETE) != 0) |
db106d01 | 397 | continue; |
db106d01 | 398 | |
a21cbb12 | 399 | if ((delta->flags & GIT_DIFF_FLAG__TO_SPLIT) != 0 && actually_split) { |
d958e37a RB |
400 | delta->similarity = 0; |
401 | ||
2123a17f | 402 | if (insert_delete_side_of_split(diff, &onto, delta) < 0) |
11d9f6b3 | 403 | goto on_error; |
db106d01 | 404 | |
9be5be47 RB |
405 | if (diff->new_src == GIT_ITERATOR_TYPE_WORKDIR) |
406 | delta->status = GIT_DELTA_UNTRACKED; | |
407 | else | |
408 | delta->status = GIT_DELTA_ADDED; | |
74a627f0 | 409 | delta->nfiles = 1; |
db106d01 RB |
410 | memset(&delta->old_file, 0, sizeof(delta->old_file)); |
411 | delta->old_file.path = delta->new_file.path; | |
9950bb4e | 412 | delta->old_file.flags |= GIT_DIFF_FLAG_VALID_ID; |
db106d01 RB |
413 | } |
414 | ||
c68b09dc RB |
415 | /* clean up delta before inserting into new list */ |
416 | GIT_DIFF_FLAG__CLEAR_INTERNAL(delta->flags); | |
417 | ||
418 | if (delta->status != GIT_DELTA_COPIED && | |
419 | delta->status != GIT_DELTA_RENAMED && | |
420 | (delta->status != GIT_DELTA_MODIFIED || actually_split)) | |
421 | delta->similarity = 0; | |
422 | ||
423 | /* insert into new list */ | |
11d9f6b3 PK |
424 | if (git_vector_insert(&onto, delta) < 0) |
425 | goto on_error; | |
db106d01 RB |
426 | } |
427 | ||
11d9f6b3 | 428 | /* cannot return an error past this point */ |
c68b09dc RB |
429 | |
430 | /* free deltas from old list that didn't make it to the new one */ | |
a21cbb12 | 431 | git_vector_foreach(&diff->deltas, i, delta) { |
71a3d27e | 432 | if ((delta->flags & GIT_DIFF_FLAG__TO_DELETE) != 0) |
11d9f6b3 | 433 | git__free(delta); |
a21cbb12 RB |
434 | } |
435 | ||
db106d01 | 436 | /* swap new delta list into place */ |
db106d01 RB |
437 | git_vector_swap(&diff->deltas, &onto); |
438 | git_vector_free(&onto); | |
a21cbb12 | 439 | git_vector_sort(&diff->deltas); |
db106d01 RB |
440 | |
441 | return 0; | |
11d9f6b3 PK |
442 | |
443 | on_error: | |
9cfce273 | 444 | git_vector_free_deep(&onto); |
11d9f6b3 PK |
445 | |
446 | return -1; | |
db106d01 RB |
447 | } |
448 | ||
3ff1d123 | 449 | GIT_INLINE(git_diff_file *) similarity_get_file(git_diff *diff, size_t idx) |
960a04dd RB |
450 | { |
451 | git_diff_delta *delta = git_vector_get(&diff->deltas, idx / 2); | |
452 | return (idx & 1) ? &delta->new_file : &delta->old_file; | |
453 | } | |
99ba8f23 | 454 | |
a5140f4d RB |
455 | typedef struct { |
456 | size_t idx; | |
457 | git_iterator_type_t src; | |
458 | git_repository *repo; | |
459 | git_diff_file *file; | |
460 | git_buf data; | |
effdbeb3 | 461 | git_odb_object *odb_obj; |
a5140f4d | 462 | git_blob *blob; |
a5140f4d RB |
463 | } similarity_info; |
464 | ||
effdbeb3 | 465 | static int similarity_init( |
3ff1d123 | 466 | similarity_info *info, git_diff *diff, size_t file_idx) |
a5140f4d RB |
467 | { |
468 | info->idx = file_idx; | |
469 | info->src = (file_idx & 1) ? diff->new_src : diff->old_src; | |
470 | info->repo = diff->repo; | |
471 | info->file = similarity_get_file(diff, file_idx); | |
effdbeb3 | 472 | info->odb_obj = NULL; |
a5140f4d | 473 | info->blob = NULL; |
a5140f4d | 474 | git_buf_init(&info->data, 0); |
09fae31d | 475 | |
410a8e6f | 476 | if (info->file->size > 0 || info->src == GIT_ITERATOR_TYPE_WORKDIR) |
effdbeb3 | 477 | return 0; |
5e5848eb | 478 | |
effdbeb3 RB |
479 | return git_diff_file__resolve_zero_size( |
480 | info->file, &info->odb_obj, info->repo); | |
a5140f4d | 481 | } |
960a04dd | 482 | |
d730d3f4 | 483 | static int similarity_sig( |
a5140f4d RB |
484 | similarity_info *info, |
485 | const git_diff_find_options *opts, | |
486 | void **cache) | |
487 | { | |
488 | int error = 0; | |
effdbeb3 | 489 | git_diff_file *file = info->file; |
8cfd54f0 | 490 | |
effdbeb3 RB |
491 | if (info->src == GIT_ITERATOR_TYPE_WORKDIR) { |
492 | if ((error = git_buf_joinpath( | |
493 | &info->data, git_repository_workdir(info->repo), file->path)) < 0) | |
494 | return error; | |
960a04dd | 495 | |
effdbeb3 RB |
496 | /* if path is not a regular file, just skip this item */ |
497 | if (!git_path_isfile(info->data.ptr)) | |
498 | return 0; | |
a5140f4d | 499 | |
a5140f4d RB |
500 | /* TODO: apply wd-to-odb filters to file data if necessary */ |
501 | ||
502 | error = opts->metric->file_signature( | |
503 | &cache[info->idx], info->file, | |
504 | info->data.ptr, opts->metric->payload); | |
505 | } else { | |
effdbeb3 RB |
506 | /* if we didn't initially know the size, we might have an odb_obj |
507 | * around from earlier, so convert that, otherwise load the blob now | |
508 | */ | |
509 | if (info->odb_obj != NULL) | |
510 | error = git_object__from_odb_object( | |
511 | (git_object **)&info->blob, info->repo, | |
512 | info->odb_obj, GIT_OBJ_BLOB); | |
513 | else | |
9950bb4e | 514 | error = git_blob_lookup(&info->blob, info->repo, &file->id); |
effdbeb3 RB |
515 | |
516 | if (error < 0) { | |
517 | /* if lookup fails, just skip this item in similarity calc */ | |
518 | giterr_clear(); | |
519 | } else { | |
a16e4172 RB |
520 | size_t sz; |
521 | ||
522 | /* index size may not be actual blob size if filtered */ | |
523 | if (file->size != git_blob_rawsize(info->blob)) | |
524 | file->size = git_blob_rawsize(info->blob); | |
525 | ||
526 | sz = (size_t)(git__is_sizet(file->size) ? file->size : -1); | |
effdbeb3 RB |
527 | |
528 | error = opts->metric->buffer_signature( | |
529 | &cache[info->idx], info->file, | |
530 | git_blob_rawcontent(info->blob), sz, opts->metric->payload); | |
531 | } | |
960a04dd RB |
532 | } |
533 | ||
534 | return error; | |
535 | } | |
536 | ||
effdbeb3 RB |
537 | static void similarity_unload(similarity_info *info) |
538 | { | |
539 | if (info->odb_obj) | |
540 | git_odb_object_free(info->odb_obj); | |
541 | ||
542 | if (info->blob) | |
543 | git_blob_free(info->blob); | |
544 | else | |
545 | git_buf_free(&info->data); | |
546 | } | |
547 | ||
a21cbb12 | 548 | #define FLAG_SET(opts,flag_name) (((opts)->flags & flag_name) != 0) |
9be5be47 RB |
549 | |
550 | /* - score < 0 means files cannot be compared | |
551 | * - score >= 100 means files are exact match | |
552 | * - score == 0 means files are completely different | |
553 | */ | |
960a04dd | 554 | static int similarity_measure( |
9be5be47 | 555 | int *score, |
3ff1d123 | 556 | git_diff *diff, |
a21cbb12 | 557 | const git_diff_find_options *opts, |
960a04dd RB |
558 | void **cache, |
559 | size_t a_idx, | |
560 | size_t b_idx) | |
561 | { | |
960a04dd RB |
562 | git_diff_file *a_file = similarity_get_file(diff, a_idx); |
563 | git_diff_file *b_file = similarity_get_file(diff, b_idx); | |
a21cbb12 | 564 | bool exact_match = FLAG_SET(opts, GIT_DIFF_FIND_EXACT_MATCH_ONLY); |
a5140f4d RB |
565 | int error = 0; |
566 | similarity_info a_info, b_info; | |
9be5be47 RB |
567 | |
568 | *score = -1; | |
960a04dd | 569 | |
9be5be47 | 570 | /* don't try to compare files of different types */ |
960a04dd RB |
571 | if (GIT_MODE_TYPE(a_file->mode) != GIT_MODE_TYPE(b_file->mode)) |
572 | return 0; | |
573 | ||
a1683f28 | 574 | /* if exact match is requested, force calculation of missing OIDs now */ |
9be5be47 | 575 | if (exact_match) { |
9950bb4e | 576 | if (git_oid_iszero(&a_file->id) && |
9be5be47 | 577 | diff->old_src == GIT_ITERATOR_TYPE_WORKDIR && |
240f4af3 RB |
578 | !git_diff__oid_for_file(&a_file->id, |
579 | diff, a_file->path, a_file->mode, a_file->size)) | |
9950bb4e | 580 | a_file->flags |= GIT_DIFF_FLAG_VALID_ID; |
9be5be47 | 581 | |
9950bb4e | 582 | if (git_oid_iszero(&b_file->id) && |
9be5be47 | 583 | diff->new_src == GIT_ITERATOR_TYPE_WORKDIR && |
240f4af3 RB |
584 | !git_diff__oid_for_file(&b_file->id, |
585 | diff, b_file->path, b_file->mode, b_file->size)) | |
9950bb4e | 586 | b_file->flags |= GIT_DIFF_FLAG_VALID_ID; |
9be5be47 RB |
587 | } |
588 | ||
589 | /* check OID match as a quick test */ | |
9950bb4e | 590 | if (git_oid__cmp(&a_file->id, &b_file->id) == 0) { |
9be5be47 RB |
591 | *score = 100; |
592 | return 0; | |
593 | } | |
594 | ||
595 | /* don't calculate signatures if we are doing exact match */ | |
596 | if (exact_match) { | |
597 | *score = 0; | |
598 | return 0; | |
599 | } | |
db106d01 | 600 | |
effdbeb3 RB |
601 | memset(&a_info, 0, sizeof(a_info)); |
602 | memset(&b_info, 0, sizeof(b_info)); | |
a5140f4d | 603 | |
effdbeb3 RB |
604 | /* set up similarity data (will try to update missing file sizes) */ |
605 | if (!cache[a_idx] && (error = similarity_init(&a_info, diff, a_idx)) < 0) | |
606 | return error; | |
607 | if (!cache[b_idx] && (error = similarity_init(&b_info, diff, b_idx)) < 0) | |
608 | goto cleanup; | |
a5140f4d | 609 | |
f5c4d022 | 610 | /* check if file sizes are nowhere near each other */ |
18e9efc4 RB |
611 | if (a_file->size > 127 && |
612 | b_file->size > 127 && | |
d730d3f4 RB |
613 | (a_file->size > (b_file->size << 3) || |
614 | b_file->size > (a_file->size << 3))) | |
effdbeb3 | 615 | goto cleanup; |
18e9efc4 | 616 | |
960a04dd | 617 | /* update signature cache if needed */ |
d730d3f4 RB |
618 | if (!cache[a_idx]) { |
619 | if ((error = similarity_sig(&a_info, opts, cache)) < 0) | |
620 | goto cleanup; | |
621 | } | |
622 | if (!cache[b_idx]) { | |
623 | if ((error = similarity_sig(&b_info, opts, cache)) < 0) | |
624 | goto cleanup; | |
625 | } | |
1fed6b07 | 626 | |
a5140f4d RB |
627 | /* calculate similarity provided that the metric choose to process |
628 | * both the a and b files (some may not if file is too big, etc). | |
629 | */ | |
630 | if (cache[a_idx] && cache[b_idx]) | |
631 | error = opts->metric->similarity( | |
632 | score, cache[a_idx], cache[b_idx], opts->metric->payload); | |
db106d01 | 633 | |
effdbeb3 | 634 | cleanup: |
a5140f4d RB |
635 | similarity_unload(&a_info); |
636 | similarity_unload(&b_info); | |
637 | ||
638 | return error; | |
db106d01 RB |
639 | } |
640 | ||
a21cbb12 | 641 | static int calc_self_similarity( |
3ff1d123 | 642 | git_diff *diff, |
a21cbb12 RB |
643 | const git_diff_find_options *opts, |
644 | size_t delta_idx, | |
645 | void **cache) | |
9be5be47 | 646 | { |
a21cbb12 RB |
647 | int error, similarity = -1; |
648 | git_diff_delta *delta = GIT_VECTOR_GET(&diff->deltas, delta_idx); | |
649 | ||
650 | if ((delta->flags & GIT_DIFF_FLAG__HAS_SELF_SIMILARITY) != 0) | |
651 | return 0; | |
652 | ||
653 | error = similarity_measure( | |
654 | &similarity, diff, opts, cache, 2 * delta_idx, 2 * delta_idx + 1); | |
655 | if (error < 0) | |
656 | return error; | |
657 | ||
658 | if (similarity >= 0) { | |
74a627f0 | 659 | delta->similarity = (uint16_t)similarity; |
a21cbb12 RB |
660 | delta->flags |= GIT_DIFF_FLAG__HAS_SELF_SIMILARITY; |
661 | } | |
662 | ||
663 | return 0; | |
664 | } | |
665 | ||
666 | static bool is_rename_target( | |
3ff1d123 | 667 | git_diff *diff, |
a21cbb12 RB |
668 | const git_diff_find_options *opts, |
669 | size_t delta_idx, | |
670 | void **cache) | |
671 | { | |
672 | git_diff_delta *delta = GIT_VECTOR_GET(&diff->deltas, delta_idx); | |
673 | ||
674 | /* skip things that aren't plain blobs */ | |
675 | if (!GIT_MODE_ISBLOB(delta->new_file.mode)) | |
676 | return false; | |
677 | ||
678 | /* only consider ADDED, RENAMED, COPIED, and split MODIFIED as | |
50456801 | 679 | * targets; maybe include UNTRACKED if requested. |
a21cbb12 RB |
680 | */ |
681 | switch (delta->status) { | |
682 | case GIT_DELTA_UNMODIFIED: | |
683 | case GIT_DELTA_DELETED: | |
50456801 POL |
684 | case GIT_DELTA_IGNORED: |
685 | case GIT_DELTA_CONFLICTED: | |
a21cbb12 RB |
686 | return false; |
687 | ||
688 | case GIT_DELTA_MODIFIED: | |
689 | if (!FLAG_SET(opts, GIT_DIFF_FIND_REWRITES) && | |
690 | !FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES)) | |
691 | return false; | |
692 | ||
693 | if (calc_self_similarity(diff, opts, delta_idx, cache) < 0) | |
694 | return false; | |
695 | ||
696 | if (FLAG_SET(opts, GIT_DIFF_BREAK_REWRITES) && | |
697 | delta->similarity < opts->break_rewrite_threshold) { | |
698 | delta->flags |= GIT_DIFF_FLAG__TO_SPLIT; | |
699 | break; | |
700 | } | |
701 | if (FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES) && | |
702 | delta->similarity < opts->rename_from_rewrite_threshold) | |
703 | break; | |
704 | ||
705 | return false; | |
706 | ||
707 | case GIT_DELTA_UNTRACKED: | |
a21cbb12 RB |
708 | if (!FLAG_SET(opts, GIT_DIFF_FIND_FOR_UNTRACKED)) |
709 | return false; | |
710 | break; | |
711 | ||
712 | default: /* all other status values should be checked */ | |
713 | break; | |
714 | } | |
715 | ||
716 | delta->flags |= GIT_DIFF_FLAG__IS_RENAME_TARGET; | |
717 | return true; | |
718 | } | |
719 | ||
720 | static bool is_rename_source( | |
3ff1d123 | 721 | git_diff *diff, |
a21cbb12 RB |
722 | const git_diff_find_options *opts, |
723 | size_t delta_idx, | |
724 | void **cache) | |
725 | { | |
726 | git_diff_delta *delta = GIT_VECTOR_GET(&diff->deltas, delta_idx); | |
727 | ||
728 | /* skip things that aren't blobs */ | |
729 | if (!GIT_MODE_ISBLOB(delta->old_file.mode)) | |
730 | return false; | |
731 | ||
732 | switch (delta->status) { | |
733 | case GIT_DELTA_ADDED: | |
734 | case GIT_DELTA_UNTRACKED: | |
61bef72d | 735 | case GIT_DELTA_UNREADABLE: |
a21cbb12 | 736 | case GIT_DELTA_IGNORED: |
50456801 | 737 | case GIT_DELTA_CONFLICTED: |
a21cbb12 RB |
738 | return false; |
739 | ||
740 | case GIT_DELTA_DELETED: | |
741 | case GIT_DELTA_TYPECHANGE: | |
742 | break; | |
743 | ||
744 | case GIT_DELTA_UNMODIFIED: | |
745 | if (!FLAG_SET(opts, GIT_DIFF_FIND_COPIES_FROM_UNMODIFIED)) | |
746 | return false; | |
f62c174d | 747 | if (FLAG_SET(opts, GIT_DIFF_FIND_REMOVE_UNMODIFIED)) |
97ad85b8 | 748 | delta->flags |= GIT_DIFF_FLAG__TO_DELETE; |
a21cbb12 RB |
749 | break; |
750 | ||
751 | default: /* MODIFIED, RENAMED, COPIED */ | |
752 | /* if we're finding copies, this could be a source */ | |
753 | if (FLAG_SET(opts, GIT_DIFF_FIND_COPIES)) | |
754 | break; | |
755 | ||
756 | /* otherwise, this is only a source if we can split it */ | |
757 | if (!FLAG_SET(opts, GIT_DIFF_FIND_REWRITES) && | |
758 | !FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES)) | |
759 | return false; | |
760 | ||
761 | if (calc_self_similarity(diff, opts, delta_idx, cache) < 0) | |
762 | return false; | |
763 | ||
764 | if (FLAG_SET(opts, GIT_DIFF_BREAK_REWRITES) && | |
765 | delta->similarity < opts->break_rewrite_threshold) { | |
766 | delta->flags |= GIT_DIFF_FLAG__TO_SPLIT; | |
767 | break; | |
768 | } | |
769 | ||
770 | if (FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES) && | |
771 | delta->similarity < opts->rename_from_rewrite_threshold) | |
772 | break; | |
773 | ||
774 | return false; | |
775 | } | |
776 | ||
777 | delta->flags |= GIT_DIFF_FLAG__IS_RENAME_SOURCE; | |
778 | return true; | |
779 | } | |
780 | ||
781 | GIT_INLINE(bool) delta_is_split(git_diff_delta *delta) | |
782 | { | |
783 | return (delta->status == GIT_DELTA_TYPECHANGE || | |
784 | (delta->flags & GIT_DIFF_FLAG__TO_SPLIT) != 0); | |
785 | } | |
786 | ||
787 | GIT_INLINE(bool) delta_is_new_only(git_diff_delta *delta) | |
788 | { | |
789 | return (delta->status == GIT_DELTA_ADDED || | |
790 | delta->status == GIT_DELTA_UNTRACKED || | |
61bef72d | 791 | delta->status == GIT_DELTA_UNREADABLE || |
a21cbb12 | 792 | delta->status == GIT_DELTA_IGNORED); |
9be5be47 | 793 | } |
db106d01 | 794 | |
e4acc3ba | 795 | GIT_INLINE(void) delta_make_rename( |
74a627f0 | 796 | git_diff_delta *to, const git_diff_delta *from, uint16_t similarity) |
e4acc3ba RB |
797 | { |
798 | to->status = GIT_DELTA_RENAMED; | |
799 | to->similarity = similarity; | |
74a627f0 | 800 | to->nfiles = 2; |
e4acc3ba RB |
801 | memcpy(&to->old_file, &from->old_file, sizeof(to->old_file)); |
802 | to->flags &= ~GIT_DIFF_FLAG__TO_SPLIT; | |
803 | } | |
804 | ||
d958e37a | 805 | typedef struct { |
74a627f0 RB |
806 | size_t idx; |
807 | uint16_t similarity; | |
d958e37a RB |
808 | } diff_find_match; |
809 | ||
db106d01 | 810 | int git_diff_find_similar( |
3ff1d123 | 811 | git_diff *diff, |
10672e3e | 812 | const git_diff_find_options *given_opts) |
db106d01 | 813 | { |
d730d3f4 | 814 | size_t s, t; |
74a627f0 RB |
815 | int error = 0, result; |
816 | uint16_t similarity; | |
d730d3f4 | 817 | git_diff_delta *src, *tgt; |
7e3ed419 | 818 | git_diff_find_options opts = GIT_DIFF_FIND_OPTIONS_INIT; |
d730d3f4 RB |
819 | size_t num_deltas, num_srcs = 0, num_tgts = 0; |
820 | size_t tried_srcs = 0, tried_tgts = 0; | |
e4acc3ba | 821 | size_t num_rewrites = 0, num_updates = 0, num_bumped = 0; |
f1453c59 | 822 | size_t sigcache_size; |
7e3ed419 | 823 | void **sigcache = NULL; /* cache of similarity metric file signatures */ |
d730d3f4 RB |
824 | diff_find_match *tgt2src = NULL; |
825 | diff_find_match *src2tgt = NULL; | |
826 | diff_find_match *tgt2src_copy = NULL; | |
827 | diff_find_match *best_match; | |
e4acc3ba | 828 | git_diff_file swap; |
db106d01 | 829 | |
960a04dd | 830 | if ((error = normalize_find_opts(diff, &opts, given_opts)) < 0) |
628e92cd BS |
831 | return error; |
832 | ||
d730d3f4 RB |
833 | num_deltas = diff->deltas.length; |
834 | ||
a21cbb12 | 835 | /* TODO: maybe abort if deltas.length > rename_limit ??? */ |
d730d3f4 | 836 | if (!git__is_uint32(num_deltas)) |
7e3ed419 RB |
837 | goto cleanup; |
838 | ||
839 | /* No flags set; nothing to do */ | |
840 | if ((opts.flags & GIT_DIFF_FIND_ALL) == 0) | |
841 | goto cleanup; | |
960a04dd | 842 | |
f1453c59 ET |
843 | GITERR_CHECK_ALLOC_MULTIPLY(&sigcache_size, num_deltas, 2); |
844 | sigcache = git__calloc(sigcache_size, sizeof(void *)); | |
e4acc3ba RB |
845 | GITERR_CHECK_ALLOC(sigcache); |
846 | ||
847 | /* Label rename sources and targets | |
848 | * | |
849 | * This will also set self-similarity scores for MODIFIED files and | |
850 | * mark them for splitting if break-rewrites is enabled | |
851 | */ | |
d730d3f4 RB |
852 | git_vector_foreach(&diff->deltas, t, tgt) { |
853 | if (is_rename_source(diff, &opts, t, sigcache)) | |
e4acc3ba RB |
854 | ++num_srcs; |
855 | ||
d730d3f4 | 856 | if (is_rename_target(diff, &opts, t, sigcache)) |
e4acc3ba | 857 | ++num_tgts; |
17c7fbf6 ET |
858 | |
859 | if ((tgt->flags & GIT_DIFF_FLAG__TO_SPLIT) != 0) | |
860 | num_rewrites++; | |
e4acc3ba | 861 | } |
960a04dd | 862 | |
e4acc3ba RB |
863 | /* if there are no candidate srcs or tgts, we're done */ |
864 | if (!num_srcs || !num_tgts) | |
865 | goto cleanup; | |
960a04dd | 866 | |
d730d3f4 RB |
867 | src2tgt = git__calloc(num_deltas, sizeof(diff_find_match)); |
868 | GITERR_CHECK_ALLOC(src2tgt); | |
869 | tgt2src = git__calloc(num_deltas, sizeof(diff_find_match)); | |
870 | GITERR_CHECK_ALLOC(tgt2src); | |
871 | ||
872 | if (FLAG_SET(&opts, GIT_DIFF_FIND_COPIES)) { | |
873 | tgt2src_copy = git__calloc(num_deltas, sizeof(diff_find_match)); | |
874 | GITERR_CHECK_ALLOC(tgt2src_copy); | |
875 | } | |
db106d01 | 876 | |
e4acc3ba RB |
877 | /* |
878 | * Find best-fit matches for rename / copy candidates | |
879 | */ | |
d958e37a | 880 | |
e4acc3ba RB |
881 | find_best_matches: |
882 | tried_tgts = num_bumped = 0; | |
db106d01 | 883 | |
d730d3f4 | 884 | git_vector_foreach(&diff->deltas, t, tgt) { |
a21cbb12 | 885 | /* skip things that are not rename targets */ |
d730d3f4 | 886 | if ((tgt->flags & GIT_DIFF_FLAG__IS_RENAME_TARGET) == 0) |
d958e37a RB |
887 | continue; |
888 | ||
e4acc3ba | 889 | tried_srcs = 0; |
db106d01 | 890 | |
d730d3f4 | 891 | git_vector_foreach(&diff->deltas, s, src) { |
a21cbb12 | 892 | /* skip things that are not rename sources */ |
d730d3f4 | 893 | if ((src->flags & GIT_DIFF_FLAG__IS_RENAME_SOURCE) == 0) |
960a04dd RB |
894 | continue; |
895 | ||
d958e37a | 896 | /* calculate similarity for this pair and find best match */ |
d730d3f4 | 897 | if (s == t) |
74a627f0 | 898 | result = -1; /* don't measure self-similarity here */ |
e4acc3ba | 899 | else if ((error = similarity_measure( |
74a627f0 | 900 | &result, diff, &opts, sigcache, 2 * s, 2 * t + 1)) < 0) |
960a04dd | 901 | goto cleanup; |
a21cbb12 | 902 | |
74a627f0 | 903 | if (result < 0) |
d730d3f4 | 904 | continue; |
74a627f0 | 905 | similarity = (uint16_t)result; |
d730d3f4 RB |
906 | |
907 | /* is this a better rename? */ | |
74a627f0 RB |
908 | if (tgt2src[t].similarity < similarity && |
909 | src2tgt[s].similarity < similarity) | |
e4acc3ba | 910 | { |
d730d3f4 RB |
911 | /* eject old mapping */ |
912 | if (src2tgt[s].similarity > 0) { | |
913 | tgt2src[src2tgt[s].idx].similarity = 0; | |
914 | num_bumped++; | |
915 | } | |
916 | if (tgt2src[t].similarity > 0) { | |
917 | src2tgt[tgt2src[t].idx].similarity = 0; | |
918 | num_bumped++; | |
e4acc3ba RB |
919 | } |
920 | ||
d730d3f4 | 921 | /* write new mapping */ |
74a627f0 RB |
922 | tgt2src[t].idx = s; |
923 | tgt2src[t].similarity = similarity; | |
924 | src2tgt[s].idx = t; | |
925 | src2tgt[s].similarity = similarity; | |
d730d3f4 | 926 | } |
a21cbb12 | 927 | |
d730d3f4 RB |
928 | /* keep best absolute match for copies */ |
929 | if (tgt2src_copy != NULL && | |
74a627f0 | 930 | tgt2src_copy[t].similarity < similarity) |
d730d3f4 | 931 | { |
74a627f0 RB |
932 | tgt2src_copy[t].idx = s; |
933 | tgt2src_copy[t].similarity = similarity; | |
db106d01 | 934 | } |
e4acc3ba RB |
935 | |
936 | if (++tried_srcs >= num_srcs) | |
937 | break; | |
938 | ||
d730d3f4 | 939 | /* cap on maximum targets we'll examine (per "tgt" file) */ |
e4acc3ba RB |
940 | if (tried_srcs > opts.rename_limit) |
941 | break; | |
db106d01 | 942 | } |
e4acc3ba RB |
943 | |
944 | if (++tried_tgts >= num_tgts) | |
945 | break; | |
db106d01 RB |
946 | } |
947 | ||
e4acc3ba RB |
948 | if (num_bumped > 0) /* try again if we bumped some items */ |
949 | goto find_best_matches; | |
950 | ||
951 | /* | |
952 | * Rewrite the diffs with renames / copies | |
953 | */ | |
954 | ||
d730d3f4 | 955 | git_vector_foreach(&diff->deltas, t, tgt) { |
e4acc3ba | 956 | /* skip things that are not rename targets */ |
d730d3f4 | 957 | if ((tgt->flags & GIT_DIFF_FLAG__IS_RENAME_TARGET) == 0) |
a21cbb12 | 958 | continue; |
690bf41c | 959 | |
e4acc3ba | 960 | /* check if this delta was the target of a similarity */ |
d730d3f4 RB |
961 | if (tgt2src[t].similarity) |
962 | best_match = &tgt2src[t]; | |
963 | else if (tgt2src_copy && tgt2src_copy[t].similarity) | |
964 | best_match = &tgt2src_copy[t]; | |
965 | else | |
e4acc3ba | 966 | continue; |
d958e37a | 967 | |
d730d3f4 RB |
968 | s = best_match->idx; |
969 | src = GIT_VECTOR_GET(&diff->deltas, s); | |
d958e37a | 970 | |
a21cbb12 RB |
971 | /* possible scenarios: |
972 | * 1. from DELETE to ADD/UNTRACK/IGNORE = RENAME | |
973 | * 2. from DELETE to SPLIT/TYPECHANGE = RENAME + DELETE | |
974 | * 3. from SPLIT/TYPECHANGE to ADD/UNTRACK/IGNORE = ADD + RENAME | |
975 | * 4. from SPLIT/TYPECHANGE to SPLIT/TYPECHANGE = RENAME + SPLIT | |
976 | * 5. from OTHER to ADD/UNTRACK/IGNORE = OTHER + COPY | |
db106d01 RB |
977 | */ |
978 | ||
d730d3f4 | 979 | if (src->status == GIT_DELTA_DELETED) { |
db106d01 | 980 | |
d730d3f4 | 981 | if (delta_is_new_only(tgt)) { |
db106d01 | 982 | |
e4acc3ba | 983 | if (best_match->similarity < opts.rename_threshold) |
a21cbb12 | 984 | continue; |
960a04dd | 985 | |
d730d3f4 | 986 | delta_make_rename(tgt, src, best_match->similarity); |
d958e37a | 987 | |
d730d3f4 | 988 | src->flags |= GIT_DIFF_FLAG__TO_DELETE; |
a21cbb12 RB |
989 | num_rewrites++; |
990 | } else { | |
d730d3f4 | 991 | assert(delta_is_split(tgt)); |
960a04dd | 992 | |
e4acc3ba | 993 | if (best_match->similarity < opts.rename_from_rewrite_threshold) |
a21cbb12 | 994 | continue; |
db106d01 | 995 | |
d730d3f4 | 996 | memcpy(&swap, &tgt->old_file, sizeof(swap)); |
db106d01 | 997 | |
d730d3f4 | 998 | delta_make_rename(tgt, src, best_match->similarity); |
e4acc3ba RB |
999 | num_rewrites--; |
1000 | ||
74a627f0 | 1001 | assert(src->status == GIT_DELTA_DELETED); |
d730d3f4 RB |
1002 | memcpy(&src->old_file, &swap, sizeof(src->old_file)); |
1003 | memset(&src->new_file, 0, sizeof(src->new_file)); | |
1004 | src->new_file.path = src->old_file.path; | |
9950bb4e | 1005 | src->new_file.flags |= GIT_DIFF_FLAG_VALID_ID; |
db106d01 | 1006 | |
d958e37a | 1007 | num_updates++; |
7edb74d3 RB |
1008 | |
1009 | if (src2tgt[t].similarity > 0 && src2tgt[t].idx > t) { | |
1010 | /* what used to be at src t is now at src s */ | |
74a627f0 | 1011 | tgt2src[src2tgt[t].idx].idx = s; |
7edb74d3 | 1012 | } |
db106d01 RB |
1013 | } |
1014 | } | |
1015 | ||
d730d3f4 | 1016 | else if (delta_is_split(src)) { |
a21cbb12 | 1017 | |
d730d3f4 | 1018 | if (delta_is_new_only(tgt)) { |
db106d01 | 1019 | |
e4acc3ba | 1020 | if (best_match->similarity < opts.rename_threshold) |
a21cbb12 | 1021 | continue; |
d958e37a | 1022 | |
d730d3f4 | 1023 | delta_make_rename(tgt, src, best_match->similarity); |
a21cbb12 | 1024 | |
d730d3f4 | 1025 | src->status = (diff->new_src == GIT_ITERATOR_TYPE_WORKDIR) ? |
a21cbb12 | 1026 | GIT_DELTA_UNTRACKED : GIT_DELTA_ADDED; |
74a627f0 | 1027 | src->nfiles = 1; |
d730d3f4 RB |
1028 | memset(&src->old_file, 0, sizeof(src->old_file)); |
1029 | src->old_file.path = src->new_file.path; | |
9950bb4e | 1030 | src->old_file.flags |= GIT_DIFF_FLAG_VALID_ID; |
e4acc3ba | 1031 | |
d730d3f4 | 1032 | src->flags &= ~GIT_DIFF_FLAG__TO_SPLIT; |
e4acc3ba | 1033 | num_rewrites--; |
a21cbb12 RB |
1034 | |
1035 | num_updates++; | |
1036 | } else { | |
d730d3f4 | 1037 | assert(delta_is_split(src)); |
a21cbb12 | 1038 | |
e4acc3ba | 1039 | if (best_match->similarity < opts.rename_from_rewrite_threshold) |
a21cbb12 RB |
1040 | continue; |
1041 | ||
d730d3f4 | 1042 | memcpy(&swap, &tgt->old_file, sizeof(swap)); |
a21cbb12 | 1043 | |
d730d3f4 | 1044 | delta_make_rename(tgt, src, best_match->similarity); |
e4acc3ba RB |
1045 | num_rewrites--; |
1046 | num_updates++; | |
a21cbb12 | 1047 | |
d730d3f4 | 1048 | memcpy(&src->old_file, &swap, sizeof(src->old_file)); |
a21cbb12 | 1049 | |
e4acc3ba RB |
1050 | /* if we've just swapped the new element into the correct |
1051 | * place, clear the SPLIT flag | |
67db583d | 1052 | */ |
d730d3f4 RB |
1053 | if (tgt2src[s].idx == t && |
1054 | tgt2src[s].similarity > | |
67db583d | 1055 | opts.rename_from_rewrite_threshold) { |
d730d3f4 RB |
1056 | src->status = GIT_DELTA_RENAMED; |
1057 | src->similarity = tgt2src[s].similarity; | |
1058 | tgt2src[s].similarity = 0; | |
1059 | src->flags &= ~GIT_DIFF_FLAG__TO_SPLIT; | |
67db583d RB |
1060 | num_rewrites--; |
1061 | } | |
e4acc3ba | 1062 | /* otherwise, if we just overwrote a source, update mapping */ |
7edb74d3 | 1063 | else if (src2tgt[t].similarity > 0 && src2tgt[t].idx > t) { |
d730d3f4 | 1064 | /* what used to be at src t is now at src s */ |
74a627f0 | 1065 | tgt2src[src2tgt[t].idx].idx = s; |
e4acc3ba | 1066 | } |
67db583d | 1067 | |
a21cbb12 RB |
1068 | num_updates++; |
1069 | } | |
1070 | } | |
1071 | ||
2123a17f | 1072 | else if (FLAG_SET(&opts, GIT_DIFF_FIND_COPIES)) { |
d730d3f4 | 1073 | if (tgt2src_copy[t].similarity < opts.copy_threshold) |
a21cbb12 RB |
1074 | continue; |
1075 | ||
d730d3f4 RB |
1076 | /* always use best possible source for copy */ |
1077 | best_match = &tgt2src_copy[t]; | |
1078 | src = GIT_VECTOR_GET(&diff->deltas, best_match->idx); | |
1079 | ||
2123a17f RB |
1080 | if (delta_is_split(tgt)) { |
1081 | error = insert_delete_side_of_split(diff, &diff->deltas, tgt); | |
1082 | if (error < 0) | |
1083 | goto cleanup; | |
1084 | num_rewrites--; | |
1085 | } | |
1086 | ||
1087 | if (!delta_is_split(tgt) && !delta_is_new_only(tgt)) | |
1088 | continue; | |
1089 | ||
d730d3f4 RB |
1090 | tgt->status = GIT_DELTA_COPIED; |
1091 | tgt->similarity = best_match->similarity; | |
74a627f0 | 1092 | tgt->nfiles = 2; |
d730d3f4 | 1093 | memcpy(&tgt->old_file, &src->old_file, sizeof(tgt->old_file)); |
2123a17f | 1094 | tgt->flags &= ~GIT_DIFF_FLAG__TO_SPLIT; |
a21cbb12 RB |
1095 | |
1096 | num_updates++; | |
1097 | } | |
db106d01 RB |
1098 | } |
1099 | ||
e4acc3ba RB |
1100 | /* |
1101 | * Actually split and delete entries as needed | |
1102 | */ | |
1103 | ||
a21cbb12 | 1104 | if (num_rewrites > 0 || num_updates > 0) |
960a04dd | 1105 | error = apply_splits_and_deletes( |
d958e37a | 1106 | diff, diff->deltas.length - num_rewrites, |
17c7fbf6 ET |
1107 | FLAG_SET(&opts, GIT_DIFF_BREAK_REWRITES) && |
1108 | !FLAG_SET(&opts, GIT_DIFF_BREAK_REWRITES_FOR_RENAMES_ONLY)); | |
d958e37a | 1109 | |
960a04dd | 1110 | cleanup: |
d730d3f4 RB |
1111 | git__free(tgt2src); |
1112 | git__free(src2tgt); | |
1113 | git__free(tgt2src_copy); | |
db106d01 | 1114 | |
7e3ed419 RB |
1115 | if (sigcache) { |
1116 | for (t = 0; t < num_deltas * 2; ++t) { | |
1117 | if (sigcache[t] != NULL) | |
1118 | opts.metric->free_signature(sigcache[t], opts.metric->payload); | |
1119 | } | |
1120 | git__free(sigcache); | |
db106d01 RB |
1121 | } |
1122 | ||
f8275890 RB |
1123 | if (!given_opts || !given_opts->metric) |
1124 | git__free(opts.metric); | |
1125 | ||
960a04dd | 1126 | return error; |
db106d01 RB |
1127 | } |
1128 | ||
1129 | #undef FLAG_SET |