]>
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" |
114f5a6c RB |
11 | |
12 | #include "diff.h" | |
5e5848eb | 13 | #include "hashsig.h" |
114f5a6c RB |
14 | #include "path.h" |
15 | #include "fileops.h" | |
db106d01 RB |
16 | |
17 | static git_diff_delta *diff_delta__dup( | |
18 | const git_diff_delta *d, git_pool *pool) | |
19 | { | |
20 | git_diff_delta *delta = git__malloc(sizeof(git_diff_delta)); | |
21 | if (!delta) | |
22 | return NULL; | |
23 | ||
24 | memcpy(delta, d, sizeof(git_diff_delta)); | |
c68b09dc | 25 | GIT_DIFF_FLAG__CLEAR_INTERNAL(delta->flags); |
db106d01 | 26 | |
d958e37a RB |
27 | if (d->old_file.path != NULL) { |
28 | delta->old_file.path = git_pool_strdup(pool, d->old_file.path); | |
29 | if (delta->old_file.path == NULL) | |
30 | goto fail; | |
31 | } | |
db106d01 | 32 | |
d958e37a | 33 | if (d->new_file.path != d->old_file.path && d->new_file.path != NULL) { |
db106d01 RB |
34 | delta->new_file.path = git_pool_strdup(pool, d->new_file.path); |
35 | if (delta->new_file.path == NULL) | |
36 | goto fail; | |
37 | } else { | |
38 | delta->new_file.path = delta->old_file.path; | |
39 | } | |
40 | ||
41 | return delta; | |
42 | ||
43 | fail: | |
44 | git__free(delta); | |
45 | return NULL; | |
46 | } | |
47 | ||
48 | static git_diff_delta *diff_delta__merge_like_cgit( | |
49 | const git_diff_delta *a, const git_diff_delta *b, git_pool *pool) | |
50 | { | |
51 | git_diff_delta *dup; | |
52 | ||
53 | /* Emulate C git for merging two diffs (a la 'git diff <sha>'). | |
54 | * | |
55 | * When C git does a diff between the work dir and a tree, it actually | |
56 | * diffs with the index but uses the workdir contents. This emulates | |
57 | * those choices so we can emulate the type of diff. | |
58 | * | |
59 | * We have three file descriptions here, let's call them: | |
60 | * f1 = a->old_file | |
61 | * f2 = a->new_file AND b->old_file | |
62 | * f3 = b->new_file | |
63 | */ | |
64 | ||
65 | /* if f2 == f3 or f2 is deleted, then just dup the 'a' diff */ | |
66 | if (b->status == GIT_DELTA_UNMODIFIED || a->status == GIT_DELTA_DELETED) | |
67 | return diff_delta__dup(a, pool); | |
68 | ||
69 | /* otherwise, base this diff on the 'b' diff */ | |
70 | if ((dup = diff_delta__dup(b, pool)) == NULL) | |
71 | return NULL; | |
72 | ||
73 | /* If 'a' status is uninteresting, then we're done */ | |
74 | if (a->status == GIT_DELTA_UNMODIFIED) | |
75 | return dup; | |
76 | ||
77 | assert(a->status != GIT_DELTA_UNMODIFIED); | |
78 | assert(b->status != GIT_DELTA_UNMODIFIED); | |
79 | ||
80 | /* A cgit exception is that the diff of a file that is only in the | |
81 | * index (i.e. not in HEAD nor workdir) is given as empty. | |
82 | */ | |
83 | if (dup->status == GIT_DELTA_DELETED) { | |
84 | if (a->status == GIT_DELTA_ADDED) | |
85 | dup->status = GIT_DELTA_UNMODIFIED; | |
86 | /* else don't overwrite DELETE status */ | |
87 | } else { | |
88 | dup->status = a->status; | |
89 | } | |
90 | ||
91 | git_oid_cpy(&dup->old_file.oid, &a->old_file.oid); | |
92 | dup->old_file.mode = a->old_file.mode; | |
93 | dup->old_file.size = a->old_file.size; | |
94 | dup->old_file.flags = a->old_file.flags; | |
95 | ||
96 | return dup; | |
97 | } | |
98 | ||
99 | int git_diff_merge( | |
100 | git_diff_list *onto, | |
101 | const git_diff_list *from) | |
102 | { | |
103 | int error = 0; | |
104 | git_pool onto_pool; | |
105 | git_vector onto_new; | |
106 | git_diff_delta *delta; | |
107 | bool ignore_case = false; | |
108 | unsigned int i, j; | |
109 | ||
110 | assert(onto && from); | |
111 | ||
112 | if (!from->deltas.length) | |
113 | return 0; | |
114 | ||
115 | if (git_vector_init( | |
116 | &onto_new, onto->deltas.length, git_diff_delta__cmp) < 0 || | |
117 | git_pool_init(&onto_pool, 1, 0) < 0) | |
118 | return -1; | |
119 | ||
120 | if ((onto->opts.flags & GIT_DIFF_DELTAS_ARE_ICASE) != 0 || | |
121 | (from->opts.flags & GIT_DIFF_DELTAS_ARE_ICASE) != 0) | |
122 | { | |
123 | ignore_case = true; | |
124 | ||
125 | /* This function currently only supports merging diff lists that | |
126 | * are sorted identically. */ | |
127 | assert((onto->opts.flags & GIT_DIFF_DELTAS_ARE_ICASE) != 0 && | |
128 | (from->opts.flags & GIT_DIFF_DELTAS_ARE_ICASE) != 0); | |
129 | } | |
130 | ||
131 | for (i = 0, j = 0; i < onto->deltas.length || j < from->deltas.length; ) { | |
132 | git_diff_delta *o = GIT_VECTOR_GET(&onto->deltas, i); | |
133 | const git_diff_delta *f = GIT_VECTOR_GET(&from->deltas, j); | |
134 | int cmp = !f ? -1 : !o ? 1 : STRCMP_CASESELECT(ignore_case, o->old_file.path, f->old_file.path); | |
135 | ||
136 | if (cmp < 0) { | |
137 | delta = diff_delta__dup(o, &onto_pool); | |
138 | i++; | |
139 | } else if (cmp > 0) { | |
140 | delta = diff_delta__dup(f, &onto_pool); | |
141 | j++; | |
142 | } else { | |
143 | delta = diff_delta__merge_like_cgit(o, f, &onto_pool); | |
144 | i++; | |
145 | j++; | |
146 | } | |
147 | ||
148 | /* the ignore rules for the target may not match the source | |
149 | * or the result of a merged delta could be skippable... | |
150 | */ | |
151 | if (git_diff_delta__should_skip(&onto->opts, delta)) { | |
152 | git__free(delta); | |
153 | continue; | |
154 | } | |
155 | ||
156 | if ((error = !delta ? -1 : git_vector_insert(&onto_new, delta)) < 0) | |
157 | break; | |
158 | } | |
159 | ||
160 | if (!error) { | |
161 | git_vector_swap(&onto->deltas, &onto_new); | |
162 | git_pool_swap(&onto->pool, &onto_pool); | |
163 | onto->new_src = from->new_src; | |
164 | ||
165 | /* prefix strings also come from old pool, so recreate those.*/ | |
166 | onto->opts.old_prefix = | |
167 | git_pool_strdup_safe(&onto->pool, onto->opts.old_prefix); | |
168 | onto->opts.new_prefix = | |
169 | git_pool_strdup_safe(&onto->pool, onto->opts.new_prefix); | |
170 | } | |
171 | ||
172 | git_vector_foreach(&onto_new, i, delta) | |
173 | git__free(delta); | |
174 | git_vector_free(&onto_new); | |
175 | git_pool_clear(&onto_pool); | |
176 | ||
177 | return error; | |
178 | } | |
179 | ||
0462fba5 | 180 | int git_diff_find_similar__hashsig_for_file( |
f8275890 RB |
181 | void **out, const git_diff_file *f, const char *path, void *p) |
182 | { | |
183 | git_hashsig_option_t opt = (git_hashsig_option_t)p; | |
aa408cbf ET |
184 | int error = 0; |
185 | ||
f8275890 | 186 | GIT_UNUSED(f); |
aa408cbf | 187 | error = git_hashsig_create_fromfile((git_hashsig **)out, path, opt); |
1fed6b07 | 188 | |
aa408cbf ET |
189 | if (error == GIT_EBUFS) { |
190 | error = 0; | |
191 | giterr_clear(); | |
192 | } | |
193 | ||
194 | return error; | |
f8275890 | 195 | } |
9bc8be3d | 196 | |
0462fba5 | 197 | int git_diff_find_similar__hashsig_for_buf( |
f8275890 RB |
198 | void **out, const git_diff_file *f, const char *buf, size_t len, void *p) |
199 | { | |
200 | git_hashsig_option_t opt = (git_hashsig_option_t)p; | |
aa408cbf | 201 | int error = 0; |
0462fba5 | 202 | |
f8275890 | 203 | GIT_UNUSED(f); |
aa408cbf | 204 | error = git_hashsig_create((git_hashsig **)out, buf, len, opt); |
1fed6b07 | 205 | |
aa408cbf ET |
206 | if (error == GIT_EBUFS) { |
207 | error = 0; | |
208 | giterr_clear(); | |
209 | } | |
210 | ||
211 | return error; | |
f8275890 | 212 | } |
9bc8be3d | 213 | |
0462fba5 | 214 | void git_diff_find_similar__hashsig_free(void *sig, void *payload) |
9bc8be3d RB |
215 | { |
216 | GIT_UNUSED(payload); | |
217 | git_hashsig_free(sig); | |
218 | } | |
219 | ||
0462fba5 | 220 | int git_diff_find_similar__calc_similarity( |
9bc8be3d RB |
221 | int *score, void *siga, void *sigb, void *payload) |
222 | { | |
223 | GIT_UNUSED(payload); | |
224 | *score = git_hashsig_compare(siga, sigb); | |
225 | return 0; | |
226 | } | |
227 | ||
db106d01 RB |
228 | #define DEFAULT_THRESHOLD 50 |
229 | #define DEFAULT_BREAK_REWRITE_THRESHOLD 60 | |
a21cbb12 | 230 | #define DEFAULT_RENAME_LIMIT 200 |
db106d01 RB |
231 | |
232 | static int normalize_find_opts( | |
233 | git_diff_list *diff, | |
234 | git_diff_find_options *opts, | |
235 | git_diff_find_options *given) | |
236 | { | |
237 | git_config *cfg = NULL; | |
db106d01 RB |
238 | |
239 | if (diff->repo != NULL && | |
240 | git_repository_config__weakptr(&cfg, diff->repo) < 0) | |
241 | return -1; | |
242 | ||
243 | if (given != NULL) | |
244 | memcpy(opts, given, sizeof(*opts)); | |
245 | else { | |
0a008913 RB |
246 | const char *val = NULL; |
247 | ||
248 | GIT_INIT_STRUCTURE(opts, GIT_DIFF_FIND_OPTIONS_VERSION); | |
db106d01 RB |
249 | |
250 | opts->flags = GIT_DIFF_FIND_RENAMES; | |
251 | ||
252 | if (git_config_get_string(&val, cfg, "diff.renames") < 0) | |
253 | giterr_clear(); | |
254 | else if (val && | |
255 | (!strcasecmp(val, "copies") || !strcasecmp(val, "copy"))) | |
256 | opts->flags = GIT_DIFF_FIND_RENAMES | GIT_DIFF_FIND_COPIES; | |
257 | } | |
258 | ||
c7231c45 | 259 | GITERR_CHECK_VERSION(opts, GIT_DIFF_FIND_OPTIONS_VERSION, "git_diff_find_options"); |
ca901e7b | 260 | |
db106d01 RB |
261 | /* some flags imply others */ |
262 | ||
9be5be47 RB |
263 | if (opts->flags & GIT_DIFF_FIND_EXACT_MATCH_ONLY) { |
264 | /* if we are only looking for exact matches, then don't turn | |
265 | * MODIFIED items into ADD/DELETE pairs because it's too picky | |
266 | */ | |
267 | opts->flags &= ~(GIT_DIFF_FIND_REWRITES | GIT_DIFF_BREAK_REWRITES); | |
268 | ||
269 | /* similarly, don't look for self-rewrites to split */ | |
270 | opts->flags &= ~GIT_DIFF_FIND_RENAMES_FROM_REWRITES; | |
271 | } | |
272 | ||
db106d01 RB |
273 | if (opts->flags & GIT_DIFF_FIND_RENAMES_FROM_REWRITES) |
274 | opts->flags |= GIT_DIFF_FIND_RENAMES; | |
275 | ||
276 | if (opts->flags & GIT_DIFF_FIND_COPIES_FROM_UNMODIFIED) | |
277 | opts->flags |= GIT_DIFF_FIND_COPIES; | |
278 | ||
d958e37a RB |
279 | if (opts->flags & GIT_DIFF_BREAK_REWRITES) |
280 | opts->flags |= GIT_DIFF_FIND_REWRITES; | |
281 | ||
db106d01 RB |
282 | #define USE_DEFAULT(X) ((X) == 0 || (X) > 100) |
283 | ||
284 | if (USE_DEFAULT(opts->rename_threshold)) | |
285 | opts->rename_threshold = DEFAULT_THRESHOLD; | |
286 | ||
287 | if (USE_DEFAULT(opts->rename_from_rewrite_threshold)) | |
288 | opts->rename_from_rewrite_threshold = DEFAULT_THRESHOLD; | |
289 | ||
290 | if (USE_DEFAULT(opts->copy_threshold)) | |
291 | opts->copy_threshold = DEFAULT_THRESHOLD; | |
292 | ||
293 | if (USE_DEFAULT(opts->break_rewrite_threshold)) | |
294 | opts->break_rewrite_threshold = DEFAULT_BREAK_REWRITE_THRESHOLD; | |
295 | ||
296 | #undef USE_DEFAULT | |
297 | ||
a21cbb12 | 298 | if (!opts->rename_limit) { |
db106d01 RB |
299 | int32_t limit = 0; |
300 | ||
a21cbb12 | 301 | opts->rename_limit = DEFAULT_RENAME_LIMIT; |
db106d01 RB |
302 | |
303 | if (git_config_get_int32(&limit, cfg, "diff.renameLimit") < 0) | |
304 | giterr_clear(); | |
305 | else if (limit > 0) | |
a21cbb12 | 306 | opts->rename_limit = limit; |
db106d01 RB |
307 | } |
308 | ||
f8275890 | 309 | /* assign the internal metric with whitespace flag as payload */ |
9bc8be3d | 310 | if (!opts->metric) { |
f8275890 RB |
311 | opts->metric = git__malloc(sizeof(git_diff_similarity_metric)); |
312 | GITERR_CHECK_ALLOC(opts->metric); | |
313 | ||
0462fba5 ET |
314 | opts->metric->file_signature = git_diff_find_similar__hashsig_for_file; |
315 | opts->metric->buffer_signature = git_diff_find_similar__hashsig_for_buf; | |
316 | opts->metric->free_signature = git_diff_find_similar__hashsig_free; | |
317 | opts->metric->similarity = git_diff_find_similar__calc_similarity; | |
f8275890 | 318 | |
9bc8be3d | 319 | if (opts->flags & GIT_DIFF_FIND_IGNORE_WHITESPACE) |
f8275890 | 320 | opts->metric->payload = (void *)GIT_HASHSIG_IGNORE_WHITESPACE; |
9bc8be3d | 321 | else if (opts->flags & GIT_DIFF_FIND_DONT_IGNORE_WHITESPACE) |
f8275890 | 322 | opts->metric->payload = (void *)GIT_HASHSIG_NORMAL; |
9bc8be3d | 323 | else |
f8275890 | 324 | opts->metric->payload = (void *)GIT_HASHSIG_SMART_WHITESPACE; |
9bc8be3d RB |
325 | } |
326 | ||
db106d01 RB |
327 | return 0; |
328 | } | |
329 | ||
d958e37a RB |
330 | static int apply_splits_and_deletes( |
331 | git_diff_list *diff, size_t expected_size, bool actually_split) | |
db106d01 RB |
332 | { |
333 | git_vector onto = GIT_VECTOR_INIT; | |
334 | size_t i; | |
d958e37a | 335 | git_diff_delta *delta, *deleted; |
db106d01 RB |
336 | |
337 | if (git_vector_init(&onto, expected_size, git_diff_delta__cmp) < 0) | |
338 | return -1; | |
339 | ||
340 | /* build new delta list without TO_DELETE and splitting TO_SPLIT */ | |
341 | git_vector_foreach(&diff->deltas, i, delta) { | |
71a3d27e | 342 | if ((delta->flags & GIT_DIFF_FLAG__TO_DELETE) != 0) |
db106d01 | 343 | continue; |
db106d01 | 344 | |
a21cbb12 | 345 | if ((delta->flags & GIT_DIFF_FLAG__TO_SPLIT) != 0 && actually_split) { |
d958e37a RB |
346 | delta->similarity = 0; |
347 | ||
348 | /* make new record for DELETED side of split */ | |
349 | if (!(deleted = diff_delta__dup(delta, &diff->pool))) | |
11d9f6b3 | 350 | goto on_error; |
db106d01 RB |
351 | |
352 | deleted->status = GIT_DELTA_DELETED; | |
353 | memset(&deleted->new_file, 0, sizeof(deleted->new_file)); | |
354 | deleted->new_file.path = deleted->old_file.path; | |
71a3d27e | 355 | deleted->new_file.flags |= GIT_DIFF_FLAG_VALID_OID; |
db106d01 | 356 | |
11d9f6b3 PK |
357 | if (git_vector_insert(&onto, deleted) < 0) |
358 | goto on_error; | |
db106d01 | 359 | |
9be5be47 RB |
360 | if (diff->new_src == GIT_ITERATOR_TYPE_WORKDIR) |
361 | delta->status = GIT_DELTA_UNTRACKED; | |
362 | else | |
363 | delta->status = GIT_DELTA_ADDED; | |
db106d01 RB |
364 | memset(&delta->old_file, 0, sizeof(delta->old_file)); |
365 | delta->old_file.path = delta->new_file.path; | |
71a3d27e | 366 | delta->old_file.flags |= GIT_DIFF_FLAG_VALID_OID; |
db106d01 RB |
367 | } |
368 | ||
c68b09dc RB |
369 | /* clean up delta before inserting into new list */ |
370 | GIT_DIFF_FLAG__CLEAR_INTERNAL(delta->flags); | |
371 | ||
372 | if (delta->status != GIT_DELTA_COPIED && | |
373 | delta->status != GIT_DELTA_RENAMED && | |
374 | (delta->status != GIT_DELTA_MODIFIED || actually_split)) | |
375 | delta->similarity = 0; | |
376 | ||
377 | /* insert into new list */ | |
11d9f6b3 PK |
378 | if (git_vector_insert(&onto, delta) < 0) |
379 | goto on_error; | |
db106d01 RB |
380 | } |
381 | ||
11d9f6b3 | 382 | /* cannot return an error past this point */ |
c68b09dc RB |
383 | |
384 | /* free deltas from old list that didn't make it to the new one */ | |
a21cbb12 | 385 | git_vector_foreach(&diff->deltas, i, delta) { |
71a3d27e | 386 | if ((delta->flags & GIT_DIFF_FLAG__TO_DELETE) != 0) |
11d9f6b3 | 387 | git__free(delta); |
a21cbb12 RB |
388 | } |
389 | ||
db106d01 | 390 | /* swap new delta list into place */ |
db106d01 RB |
391 | git_vector_swap(&diff->deltas, &onto); |
392 | git_vector_free(&onto); | |
a21cbb12 | 393 | git_vector_sort(&diff->deltas); |
db106d01 RB |
394 | |
395 | return 0; | |
11d9f6b3 PK |
396 | |
397 | on_error: | |
398 | git_vector_foreach(&onto, i, delta) | |
399 | git__free(delta); | |
11d9f6b3 PK |
400 | git_vector_free(&onto); |
401 | ||
402 | return -1; | |
db106d01 RB |
403 | } |
404 | ||
960a04dd RB |
405 | GIT_INLINE(git_diff_file *) similarity_get_file(git_diff_list *diff, size_t idx) |
406 | { | |
407 | git_diff_delta *delta = git_vector_get(&diff->deltas, idx / 2); | |
408 | return (idx & 1) ? &delta->new_file : &delta->old_file; | |
409 | } | |
99ba8f23 | 410 | |
960a04dd RB |
411 | static int similarity_calc( |
412 | git_diff_list *diff, | |
a21cbb12 | 413 | const git_diff_find_options *opts, |
960a04dd RB |
414 | size_t file_idx, |
415 | void **cache) | |
db106d01 | 416 | { |
960a04dd RB |
417 | int error = 0; |
418 | git_diff_file *file = similarity_get_file(diff, file_idx); | |
d958e37a | 419 | git_iterator_type_t src = (file_idx & 1) ? diff->new_src : diff->old_src; |
960a04dd RB |
420 | |
421 | if (src == GIT_ITERATOR_TYPE_WORKDIR) { /* compute hashsig from file */ | |
422 | git_buf path = GIT_BUF_INIT; | |
423 | ||
424 | /* TODO: apply wd-to-odb filters to file data if necessary */ | |
425 | ||
09fae31d RB |
426 | if ((error = git_buf_joinpath( |
427 | &path, git_repository_workdir(diff->repo), file->path)) < 0) | |
428 | return error; | |
429 | ||
430 | /* if path is not a regular file, just skip this item */ | |
431 | if (git_path_isfile(path.ptr)) | |
960a04dd RB |
432 | error = opts->metric->file_signature( |
433 | &cache[file_idx], file, path.ptr, opts->metric->payload); | |
434 | ||
435 | git_buf_free(&path); | |
436 | } else { /* compute hashsig from blob buffer */ | |
437 | git_blob *blob = NULL; | |
8cfd54f0 | 438 | git_off_t blobsize; |
db106d01 | 439 | |
960a04dd | 440 | /* TODO: add max size threshold a la diff? */ |
5e5848eb | 441 | |
09fae31d RB |
442 | if (git_blob_lookup(&blob, diff->repo, &file->oid) < 0) { |
443 | /* if lookup fails, just skip this item in similarity calc */ | |
444 | giterr_clear(); | |
445 | return 0; | |
446 | } | |
960a04dd | 447 | |
8cfd54f0 | 448 | blobsize = git_blob_rawsize(blob); |
18e9efc4 RB |
449 | if (!file->size) |
450 | file->size = blobsize; | |
8cfd54f0 RB |
451 | if (!git__is_sizet(blobsize)) /* ? what to do ? */ |
452 | blobsize = (size_t)-1; | |
453 | ||
960a04dd RB |
454 | error = opts->metric->buffer_signature( |
455 | &cache[file_idx], file, git_blob_rawcontent(blob), | |
8cfd54f0 | 456 | (size_t)blobsize, opts->metric->payload); |
960a04dd RB |
457 | |
458 | git_blob_free(blob); | |
459 | } | |
460 | ||
461 | return error; | |
462 | } | |
463 | ||
a21cbb12 | 464 | #define FLAG_SET(opts,flag_name) (((opts)->flags & flag_name) != 0) |
9be5be47 RB |
465 | |
466 | /* - score < 0 means files cannot be compared | |
467 | * - score >= 100 means files are exact match | |
468 | * - score == 0 means files are completely different | |
469 | */ | |
960a04dd | 470 | static int similarity_measure( |
9be5be47 | 471 | int *score, |
960a04dd | 472 | git_diff_list *diff, |
a21cbb12 | 473 | const git_diff_find_options *opts, |
960a04dd RB |
474 | void **cache, |
475 | size_t a_idx, | |
476 | size_t b_idx) | |
477 | { | |
960a04dd RB |
478 | git_diff_file *a_file = similarity_get_file(diff, a_idx); |
479 | git_diff_file *b_file = similarity_get_file(diff, b_idx); | |
a21cbb12 | 480 | bool exact_match = FLAG_SET(opts, GIT_DIFF_FIND_EXACT_MATCH_ONLY); |
9be5be47 RB |
481 | |
482 | *score = -1; | |
960a04dd | 483 | |
9be5be47 | 484 | /* don't try to compare files of different types */ |
960a04dd RB |
485 | if (GIT_MODE_TYPE(a_file->mode) != GIT_MODE_TYPE(b_file->mode)) |
486 | return 0; | |
487 | ||
a1683f28 | 488 | /* if exact match is requested, force calculation of missing OIDs now */ |
9be5be47 RB |
489 | if (exact_match) { |
490 | if (git_oid_iszero(&a_file->oid) && | |
491 | diff->old_src == GIT_ITERATOR_TYPE_WORKDIR && | |
492 | !git_diff__oid_for_file(diff->repo, a_file->path, | |
493 | a_file->mode, a_file->size, &a_file->oid)) | |
494 | a_file->flags |= GIT_DIFF_FLAG_VALID_OID; | |
495 | ||
496 | if (git_oid_iszero(&b_file->oid) && | |
497 | diff->new_src == GIT_ITERATOR_TYPE_WORKDIR && | |
498 | !git_diff__oid_for_file(diff->repo, b_file->path, | |
499 | b_file->mode, b_file->size, &b_file->oid)) | |
500 | b_file->flags |= GIT_DIFF_FLAG_VALID_OID; | |
501 | } | |
502 | ||
503 | /* check OID match as a quick test */ | |
504 | if (git_oid__cmp(&a_file->oid, &b_file->oid) == 0) { | |
505 | *score = 100; | |
506 | return 0; | |
507 | } | |
508 | ||
509 | /* don't calculate signatures if we are doing exact match */ | |
510 | if (exact_match) { | |
511 | *score = 0; | |
512 | return 0; | |
513 | } | |
db106d01 | 514 | |
f5c4d022 | 515 | /* check if file sizes are nowhere near each other */ |
18e9efc4 RB |
516 | if (a_file->size > 127 && |
517 | b_file->size > 127 && | |
518 | (a_file->size > (b_file->size << 4) || | |
519 | b_file->size > (a_file->size << 4))) | |
520 | return 0; | |
521 | ||
960a04dd RB |
522 | /* update signature cache if needed */ |
523 | if (!cache[a_idx] && similarity_calc(diff, opts, a_idx, cache) < 0) | |
524 | return -1; | |
525 | if (!cache[b_idx] && similarity_calc(diff, opts, b_idx, cache) < 0) | |
526 | return -1; | |
1fed6b07 | 527 | |
aa408cbf ET |
528 | /* some metrics may not wish to process this file (too big / too small) */ |
529 | if (!cache[a_idx] || !cache[b_idx]) | |
530 | return 0; | |
db106d01 | 531 | |
960a04dd | 532 | /* compare signatures */ |
9be5be47 RB |
533 | return opts->metric->similarity( |
534 | score, cache[a_idx], cache[b_idx], opts->metric->payload); | |
db106d01 RB |
535 | } |
536 | ||
a21cbb12 | 537 | static int calc_self_similarity( |
9be5be47 | 538 | git_diff_list *diff, |
a21cbb12 RB |
539 | const git_diff_find_options *opts, |
540 | size_t delta_idx, | |
541 | void **cache) | |
9be5be47 | 542 | { |
a21cbb12 RB |
543 | int error, similarity = -1; |
544 | git_diff_delta *delta = GIT_VECTOR_GET(&diff->deltas, delta_idx); | |
545 | ||
546 | if ((delta->flags & GIT_DIFF_FLAG__HAS_SELF_SIMILARITY) != 0) | |
547 | return 0; | |
548 | ||
549 | error = similarity_measure( | |
550 | &similarity, diff, opts, cache, 2 * delta_idx, 2 * delta_idx + 1); | |
551 | if (error < 0) | |
552 | return error; | |
553 | ||
554 | if (similarity >= 0) { | |
555 | delta->similarity = (uint32_t)similarity; | |
556 | delta->flags |= GIT_DIFF_FLAG__HAS_SELF_SIMILARITY; | |
557 | } | |
558 | ||
559 | return 0; | |
560 | } | |
561 | ||
562 | static bool is_rename_target( | |
563 | git_diff_list *diff, | |
564 | const git_diff_find_options *opts, | |
565 | size_t delta_idx, | |
566 | void **cache) | |
567 | { | |
568 | git_diff_delta *delta = GIT_VECTOR_GET(&diff->deltas, delta_idx); | |
569 | ||
570 | /* skip things that aren't plain blobs */ | |
571 | if (!GIT_MODE_ISBLOB(delta->new_file.mode)) | |
572 | return false; | |
573 | ||
574 | /* only consider ADDED, RENAMED, COPIED, and split MODIFIED as | |
575 | * targets; maybe include UNTRACKED and IGNORED if requested. | |
576 | */ | |
577 | switch (delta->status) { | |
578 | case GIT_DELTA_UNMODIFIED: | |
579 | case GIT_DELTA_DELETED: | |
580 | return false; | |
581 | ||
582 | case GIT_DELTA_MODIFIED: | |
583 | if (!FLAG_SET(opts, GIT_DIFF_FIND_REWRITES) && | |
584 | !FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES)) | |
585 | return false; | |
586 | ||
587 | if (calc_self_similarity(diff, opts, delta_idx, cache) < 0) | |
588 | return false; | |
589 | ||
590 | if (FLAG_SET(opts, GIT_DIFF_BREAK_REWRITES) && | |
591 | delta->similarity < opts->break_rewrite_threshold) { | |
592 | delta->flags |= GIT_DIFF_FLAG__TO_SPLIT; | |
593 | break; | |
594 | } | |
595 | if (FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES) && | |
596 | delta->similarity < opts->rename_from_rewrite_threshold) | |
597 | break; | |
598 | ||
599 | return false; | |
600 | ||
601 | case GIT_DELTA_UNTRACKED: | |
a21cbb12 RB |
602 | if (!FLAG_SET(opts, GIT_DIFF_FIND_FOR_UNTRACKED)) |
603 | return false; | |
604 | break; | |
605 | ||
d55bed1a ET |
606 | case GIT_DELTA_IGNORED: |
607 | return false; | |
608 | ||
a21cbb12 RB |
609 | default: /* all other status values should be checked */ |
610 | break; | |
611 | } | |
612 | ||
613 | delta->flags |= GIT_DIFF_FLAG__IS_RENAME_TARGET; | |
614 | return true; | |
615 | } | |
616 | ||
617 | static bool is_rename_source( | |
618 | git_diff_list *diff, | |
619 | const git_diff_find_options *opts, | |
620 | size_t delta_idx, | |
621 | void **cache) | |
622 | { | |
623 | git_diff_delta *delta = GIT_VECTOR_GET(&diff->deltas, delta_idx); | |
624 | ||
625 | /* skip things that aren't blobs */ | |
626 | if (!GIT_MODE_ISBLOB(delta->old_file.mode)) | |
627 | return false; | |
628 | ||
629 | switch (delta->status) { | |
630 | case GIT_DELTA_ADDED: | |
631 | case GIT_DELTA_UNTRACKED: | |
632 | case GIT_DELTA_IGNORED: | |
633 | return false; | |
634 | ||
635 | case GIT_DELTA_DELETED: | |
636 | case GIT_DELTA_TYPECHANGE: | |
637 | break; | |
638 | ||
639 | case GIT_DELTA_UNMODIFIED: | |
640 | if (!FLAG_SET(opts, GIT_DIFF_FIND_COPIES_FROM_UNMODIFIED)) | |
641 | return false; | |
642 | break; | |
643 | ||
644 | default: /* MODIFIED, RENAMED, COPIED */ | |
645 | /* if we're finding copies, this could be a source */ | |
646 | if (FLAG_SET(opts, GIT_DIFF_FIND_COPIES)) | |
647 | break; | |
648 | ||
649 | /* otherwise, this is only a source if we can split it */ | |
650 | if (!FLAG_SET(opts, GIT_DIFF_FIND_REWRITES) && | |
651 | !FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES)) | |
652 | return false; | |
653 | ||
654 | if (calc_self_similarity(diff, opts, delta_idx, cache) < 0) | |
655 | return false; | |
656 | ||
657 | if (FLAG_SET(opts, GIT_DIFF_BREAK_REWRITES) && | |
658 | delta->similarity < opts->break_rewrite_threshold) { | |
659 | delta->flags |= GIT_DIFF_FLAG__TO_SPLIT; | |
660 | break; | |
661 | } | |
662 | ||
663 | if (FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES) && | |
664 | delta->similarity < opts->rename_from_rewrite_threshold) | |
665 | break; | |
666 | ||
667 | return false; | |
668 | } | |
669 | ||
670 | delta->flags |= GIT_DIFF_FLAG__IS_RENAME_SOURCE; | |
671 | return true; | |
672 | } | |
673 | ||
674 | GIT_INLINE(bool) delta_is_split(git_diff_delta *delta) | |
675 | { | |
676 | return (delta->status == GIT_DELTA_TYPECHANGE || | |
677 | (delta->flags & GIT_DIFF_FLAG__TO_SPLIT) != 0); | |
678 | } | |
679 | ||
680 | GIT_INLINE(bool) delta_is_new_only(git_diff_delta *delta) | |
681 | { | |
682 | return (delta->status == GIT_DELTA_ADDED || | |
683 | delta->status == GIT_DELTA_UNTRACKED || | |
684 | delta->status == GIT_DELTA_IGNORED); | |
9be5be47 | 685 | } |
db106d01 | 686 | |
e4acc3ba RB |
687 | GIT_INLINE(void) delta_make_rename( |
688 | git_diff_delta *to, const git_diff_delta *from, uint32_t similarity) | |
689 | { | |
690 | to->status = GIT_DELTA_RENAMED; | |
691 | to->similarity = similarity; | |
692 | memcpy(&to->old_file, &from->old_file, sizeof(to->old_file)); | |
693 | to->flags &= ~GIT_DIFF_FLAG__TO_SPLIT; | |
694 | } | |
695 | ||
d958e37a RB |
696 | typedef struct { |
697 | uint32_t idx; | |
698 | uint32_t similarity; | |
699 | } diff_find_match; | |
700 | ||
db106d01 RB |
701 | int git_diff_find_similar( |
702 | git_diff_list *diff, | |
703 | git_diff_find_options *given_opts) | |
704 | { | |
e4acc3ba | 705 | size_t i, j, sigcache_size; |
960a04dd | 706 | int error = 0, similarity; |
db106d01 RB |
707 | git_diff_delta *from, *to; |
708 | git_diff_find_options opts; | |
e4acc3ba RB |
709 | size_t num_srcs = 0, num_tgts = 0, tried_srcs = 0, tried_tgts = 0; |
710 | size_t num_rewrites = 0, num_updates = 0, num_bumped = 0; | |
711 | void **sigcache; /* cache of similarity metric file signatures */ | |
712 | diff_find_match *match_srcs = NULL, *match_tgts = NULL, *best_match; | |
713 | git_diff_file swap; | |
db106d01 | 714 | |
960a04dd RB |
715 | if ((error = normalize_find_opts(diff, &opts, given_opts)) < 0) |
716 | return error; | |
db106d01 | 717 | |
a21cbb12 | 718 | /* TODO: maybe abort if deltas.length > rename_limit ??? */ |
d958e37a RB |
719 | if (!git__is_uint32(diff->deltas.length)) |
720 | return 0; | |
960a04dd | 721 | |
e4acc3ba RB |
722 | sigcache_size = diff->deltas.length * 2; /* keep size b/c diff may change */ |
723 | sigcache = git__calloc(sigcache_size, sizeof(void *)); | |
724 | GITERR_CHECK_ALLOC(sigcache); | |
725 | ||
726 | /* Label rename sources and targets | |
727 | * | |
728 | * This will also set self-similarity scores for MODIFIED files and | |
729 | * mark them for splitting if break-rewrites is enabled | |
730 | */ | |
731 | git_vector_foreach(&diff->deltas, i, to) { | |
732 | if (is_rename_source(diff, &opts, i, sigcache)) | |
733 | ++num_srcs; | |
734 | ||
735 | if (is_rename_target(diff, &opts, i, sigcache)) | |
736 | ++num_tgts; | |
737 | } | |
960a04dd | 738 | |
e4acc3ba RB |
739 | /* if there are no candidate srcs or tgts, we're done */ |
740 | if (!num_srcs || !num_tgts) | |
741 | goto cleanup; | |
960a04dd | 742 | |
e4acc3ba RB |
743 | match_tgts = git__calloc(diff->deltas.length, sizeof(diff_find_match)); |
744 | GITERR_CHECK_ALLOC(match_tgts); | |
745 | match_srcs = git__calloc(diff->deltas.length, sizeof(diff_find_match)); | |
746 | GITERR_CHECK_ALLOC(match_srcs); | |
db106d01 | 747 | |
e4acc3ba RB |
748 | /* |
749 | * Find best-fit matches for rename / copy candidates | |
750 | */ | |
d958e37a | 751 | |
e4acc3ba RB |
752 | find_best_matches: |
753 | tried_tgts = num_bumped = 0; | |
db106d01 | 754 | |
e4acc3ba | 755 | git_vector_foreach(&diff->deltas, i, to) { |
a21cbb12 | 756 | /* skip things that are not rename targets */ |
e4acc3ba | 757 | if ((to->flags & GIT_DIFF_FLAG__IS_RENAME_TARGET) == 0) |
d958e37a RB |
758 | continue; |
759 | ||
e4acc3ba | 760 | tried_srcs = 0; |
db106d01 | 761 | |
e4acc3ba | 762 | git_vector_foreach(&diff->deltas, j, from) { |
a21cbb12 | 763 | /* skip things that are not rename sources */ |
e4acc3ba | 764 | if ((from->flags & GIT_DIFF_FLAG__IS_RENAME_SOURCE) == 0) |
960a04dd RB |
765 | continue; |
766 | ||
d958e37a | 767 | /* calculate similarity for this pair and find best match */ |
e4acc3ba RB |
768 | if (i == j) |
769 | similarity = -1; /* don't measure self-similarity here */ | |
770 | else if ((error = similarity_measure( | |
771 | &similarity, diff, &opts, sigcache, 2 * j, 2 * i + 1)) < 0) | |
960a04dd | 772 | goto cleanup; |
a21cbb12 | 773 | |
e4acc3ba RB |
774 | /* if this pairing is better for the src and the tgt, keep it */ |
775 | if (similarity > 0 && | |
776 | match_tgts[i].similarity < (uint32_t)similarity && | |
777 | match_srcs[j].similarity < (uint32_t)similarity) | |
778 | { | |
779 | if (match_tgts[i].similarity > 0) { | |
780 | match_tgts[match_srcs[j].idx].similarity = 0; | |
781 | match_srcs[match_tgts[i].idx].similarity = 0; | |
782 | ++num_bumped; | |
783 | } | |
784 | ||
785 | match_tgts[i].similarity = (uint32_t)similarity; | |
786 | match_tgts[i].idx = (uint32_t)j; | |
a21cbb12 | 787 | |
e4acc3ba RB |
788 | match_srcs[j].similarity = (uint32_t)similarity; |
789 | match_srcs[j].idx = (uint32_t)i; | |
db106d01 | 790 | } |
e4acc3ba RB |
791 | |
792 | if (++tried_srcs >= num_srcs) | |
793 | break; | |
794 | ||
795 | /* cap on maximum targets we'll examine (per "to" file) */ | |
796 | if (tried_srcs > opts.rename_limit) | |
797 | break; | |
db106d01 | 798 | } |
e4acc3ba RB |
799 | |
800 | if (++tried_tgts >= num_tgts) | |
801 | break; | |
db106d01 RB |
802 | } |
803 | ||
e4acc3ba RB |
804 | if (num_bumped > 0) /* try again if we bumped some items */ |
805 | goto find_best_matches; | |
806 | ||
807 | /* | |
808 | * Rewrite the diffs with renames / copies | |
809 | */ | |
810 | ||
811 | tried_tgts = 0; | |
db106d01 | 812 | |
a21cbb12 | 813 | git_vector_foreach(&diff->deltas, i, to) { |
e4acc3ba RB |
814 | /* skip things that are not rename targets */ |
815 | if ((to->flags & GIT_DIFF_FLAG__IS_RENAME_TARGET) == 0) | |
a21cbb12 | 816 | continue; |
690bf41c | 817 | |
e4acc3ba RB |
818 | /* check if this delta was the target of a similarity */ |
819 | best_match = &match_tgts[i]; | |
820 | if (!best_match->similarity) | |
821 | continue; | |
d958e37a | 822 | |
e4acc3ba RB |
823 | j = best_match->idx; |
824 | from = GIT_VECTOR_GET(&diff->deltas, j); | |
d958e37a | 825 | |
a21cbb12 RB |
826 | /* possible scenarios: |
827 | * 1. from DELETE to ADD/UNTRACK/IGNORE = RENAME | |
828 | * 2. from DELETE to SPLIT/TYPECHANGE = RENAME + DELETE | |
829 | * 3. from SPLIT/TYPECHANGE to ADD/UNTRACK/IGNORE = ADD + RENAME | |
830 | * 4. from SPLIT/TYPECHANGE to SPLIT/TYPECHANGE = RENAME + SPLIT | |
831 | * 5. from OTHER to ADD/UNTRACK/IGNORE = OTHER + COPY | |
db106d01 RB |
832 | */ |
833 | ||
834 | if (from->status == GIT_DELTA_DELETED) { | |
db106d01 | 835 | |
a21cbb12 | 836 | if (delta_is_new_only(to)) { |
db106d01 | 837 | |
e4acc3ba | 838 | if (best_match->similarity < opts.rename_threshold) |
a21cbb12 | 839 | continue; |
960a04dd | 840 | |
e4acc3ba | 841 | delta_make_rename(to, from, best_match->similarity); |
d958e37a | 842 | |
e4acc3ba | 843 | from->flags |= GIT_DIFF_FLAG__TO_DELETE; |
a21cbb12 RB |
844 | num_rewrites++; |
845 | } else { | |
49f70f2c | 846 | assert(delta_is_split(to)); |
960a04dd | 847 | |
e4acc3ba | 848 | if (best_match->similarity < opts.rename_from_rewrite_threshold) |
a21cbb12 | 849 | continue; |
db106d01 | 850 | |
e4acc3ba | 851 | memcpy(&swap, &to->old_file, sizeof(swap)); |
db106d01 | 852 | |
e4acc3ba RB |
853 | delta_make_rename(to, from, best_match->similarity); |
854 | num_rewrites--; | |
855 | ||
856 | from->status = GIT_DELTA_DELETED; | |
857 | memcpy(&from->old_file, &swap, sizeof(from->old_file)); | |
858 | memset(&from->new_file, 0, sizeof(from->new_file)); | |
859 | from->new_file.path = from->old_file.path; | |
860 | from->new_file.flags |= GIT_DIFF_FLAG_VALID_OID; | |
db106d01 | 861 | |
d958e37a | 862 | num_updates++; |
db106d01 RB |
863 | } |
864 | } | |
865 | ||
a21cbb12 | 866 | else if (delta_is_split(from)) { |
a21cbb12 RB |
867 | |
868 | if (delta_is_new_only(to)) { | |
db106d01 | 869 | |
e4acc3ba | 870 | if (best_match->similarity < opts.rename_threshold) |
a21cbb12 | 871 | continue; |
d958e37a | 872 | |
e4acc3ba | 873 | delta_make_rename(to, from, best_match->similarity); |
a21cbb12 | 874 | |
e4acc3ba | 875 | from->status = (diff->new_src == GIT_ITERATOR_TYPE_WORKDIR) ? |
a21cbb12 | 876 | GIT_DELTA_UNTRACKED : GIT_DELTA_ADDED; |
e4acc3ba RB |
877 | memset(&from->old_file, 0, sizeof(from->old_file)); |
878 | from->old_file.path = from->new_file.path; | |
879 | from->old_file.flags |= GIT_DIFF_FLAG_VALID_OID; | |
880 | ||
881 | from->flags &= ~GIT_DIFF_FLAG__TO_SPLIT; | |
882 | num_rewrites--; | |
a21cbb12 RB |
883 | |
884 | num_updates++; | |
885 | } else { | |
886 | assert(delta_is_split(from)); | |
887 | ||
e4acc3ba | 888 | if (best_match->similarity < opts.rename_from_rewrite_threshold) |
a21cbb12 RB |
889 | continue; |
890 | ||
e4acc3ba | 891 | memcpy(&swap, &to->old_file, sizeof(swap)); |
a21cbb12 | 892 | |
e4acc3ba RB |
893 | delta_make_rename(to, from, best_match->similarity); |
894 | num_rewrites--; | |
895 | num_updates++; | |
a21cbb12 | 896 | |
e4acc3ba | 897 | memcpy(&from->old_file, &swap, sizeof(from->old_file)); |
a21cbb12 | 898 | |
e4acc3ba RB |
899 | /* if we've just swapped the new element into the correct |
900 | * place, clear the SPLIT flag | |
67db583d | 901 | */ |
e4acc3ba RB |
902 | if (match_tgts[j].idx == i && |
903 | match_tgts[j].similarity > | |
67db583d RB |
904 | opts.rename_from_rewrite_threshold) { |
905 | ||
e4acc3ba RB |
906 | from->status = GIT_DELTA_RENAMED; |
907 | from->similarity = match_tgts[j].similarity; | |
908 | match_tgts[j].similarity = 0; | |
67db583d RB |
909 | from->flags &= ~GIT_DIFF_FLAG__TO_SPLIT; |
910 | num_rewrites--; | |
911 | } | |
e4acc3ba RB |
912 | /* otherwise, if we just overwrote a source, update mapping */ |
913 | else if (j > i && match_srcs[i].similarity > 0) { | |
c4ac556e | 914 | match_tgts[match_srcs[i].idx].idx = (uint32_t)j; |
e4acc3ba | 915 | } |
67db583d | 916 | |
a21cbb12 RB |
917 | num_updates++; |
918 | } | |
919 | } | |
920 | ||
921 | else if (delta_is_new_only(to)) { | |
922 | if (!FLAG_SET(&opts, GIT_DIFF_FIND_COPIES) || | |
e4acc3ba | 923 | best_match->similarity < opts.copy_threshold) |
a21cbb12 RB |
924 | continue; |
925 | ||
e4acc3ba RB |
926 | to->status = GIT_DELTA_COPIED; |
927 | to->similarity = best_match->similarity; | |
a21cbb12 RB |
928 | memcpy(&to->old_file, &from->old_file, sizeof(to->old_file)); |
929 | ||
930 | num_updates++; | |
931 | } | |
db106d01 RB |
932 | } |
933 | ||
e4acc3ba RB |
934 | /* |
935 | * Actually split and delete entries as needed | |
936 | */ | |
937 | ||
a21cbb12 | 938 | if (num_rewrites > 0 || num_updates > 0) |
960a04dd | 939 | error = apply_splits_and_deletes( |
d958e37a | 940 | diff, diff->deltas.length - num_rewrites, |
a21cbb12 | 941 | FLAG_SET(&opts, GIT_DIFF_BREAK_REWRITES)); |
d958e37a | 942 | |
960a04dd | 943 | cleanup: |
e4acc3ba RB |
944 | git__free(match_srcs); |
945 | git__free(match_tgts); | |
db106d01 | 946 | |
e4acc3ba RB |
947 | for (i = 0; i < sigcache_size; ++i) { |
948 | if (sigcache[i] != NULL) | |
949 | opts.metric->free_signature(sigcache[i], opts.metric->payload); | |
db106d01 | 950 | } |
e4acc3ba | 951 | git__free(sigcache); |
db106d01 | 952 | |
f8275890 RB |
953 | if (!given_opts || !given_opts->metric) |
954 | git__free(opts.metric); | |
955 | ||
960a04dd | 956 | return error; |
db106d01 RB |
957 | } |
958 | ||
959 | #undef FLAG_SET |