diffcore-rename: somewhat optimized.
[git/git.git] / diffcore-rename.c
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "diff.h"
6 #include "diffcore.h"
7
8 /* Table of rename/copy destinations */
9
10 static struct diff_rename_dst {
11 struct diff_filespec *two;
12 struct diff_filepair *pair;
13 } *rename_dst;
14 static int rename_dst_nr, rename_dst_alloc;
15
16 static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two,
17 int insert_ok)
18 {
19 int first, last;
20
21 first = 0;
22 last = rename_dst_nr;
23 while (last > first) {
24 int next = (last + first) >> 1;
25 struct diff_rename_dst *dst = &(rename_dst[next]);
26 int cmp = strcmp(two->path, dst->two->path);
27 if (!cmp)
28 return dst;
29 if (cmp < 0) {
30 last = next;
31 continue;
32 }
33 first = next+1;
34 }
35 /* not found */
36 if (!insert_ok)
37 return NULL;
38 /* insert to make it at "first" */
39 if (rename_dst_alloc <= rename_dst_nr) {
40 rename_dst_alloc = alloc_nr(rename_dst_alloc);
41 rename_dst = xrealloc(rename_dst,
42 rename_dst_alloc * sizeof(*rename_dst));
43 }
44 rename_dst_nr++;
45 if (first < rename_dst_nr)
46 memmove(rename_dst + first + 1, rename_dst + first,
47 (rename_dst_nr - first - 1) * sizeof(*rename_dst));
48 rename_dst[first].two = alloc_filespec(two->path);
49 fill_filespec(rename_dst[first].two, two->sha1, two->mode);
50 rename_dst[first].pair = NULL;
51 return &(rename_dst[first]);
52 }
53
54 /* Table of rename/copy src files */
55 static struct diff_rename_src {
56 struct diff_filespec *one;
57 unsigned src_path_left : 1;
58 } *rename_src;
59 static int rename_src_nr, rename_src_alloc;
60
61 static struct diff_rename_src *register_rename_src(struct diff_filespec *one,
62 int src_path_left)
63 {
64 int first, last;
65
66 first = 0;
67 last = rename_src_nr;
68 while (last > first) {
69 int next = (last + first) >> 1;
70 struct diff_rename_src *src = &(rename_src[next]);
71 int cmp = strcmp(one->path, src->one->path);
72 if (!cmp)
73 return src;
74 if (cmp < 0) {
75 last = next;
76 continue;
77 }
78 first = next+1;
79 }
80
81 /* insert to make it at "first" */
82 if (rename_src_alloc <= rename_src_nr) {
83 rename_src_alloc = alloc_nr(rename_src_alloc);
84 rename_src = xrealloc(rename_src,
85 rename_src_alloc * sizeof(*rename_src));
86 }
87 rename_src_nr++;
88 if (first < rename_src_nr)
89 memmove(rename_src + first + 1, rename_src + first,
90 (rename_src_nr - first - 1) * sizeof(*rename_src));
91 rename_src[first].one = one;
92 rename_src[first].src_path_left = src_path_left;
93 return &(rename_src[first]);
94 }
95
96 static int is_exact_match(struct diff_filespec *src, struct diff_filespec *dst)
97 {
98 if (src->sha1_valid && dst->sha1_valid &&
99 !memcmp(src->sha1, dst->sha1, 20))
100 return 1;
101 if (diff_populate_filespec(src, 1) || diff_populate_filespec(dst, 1))
102 return 0;
103 if (src->size != dst->size)
104 return 0;
105 if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
106 return 0;
107 if (src->size == dst->size &&
108 !memcmp(src->data, dst->data, src->size))
109 return 1;
110 return 0;
111 }
112
113 struct diff_score {
114 int src; /* index in rename_src */
115 int dst; /* index in rename_dst */
116 int score;
117 };
118
119 static int estimate_similarity(struct diff_filespec *src,
120 struct diff_filespec *dst,
121 int minimum_score)
122 {
123 /* src points at a file that existed in the original tree (or
124 * optionally a file in the destination tree) and dst points
125 * at a newly created file. They may be quite similar, in which
126 * case we want to say src is renamed to dst or src is copied into
127 * dst, and then some edit has been applied to dst.
128 *
129 * Compare them and return how similar they are, representing
130 * the score as an integer between 0 and MAX_SCORE.
131 *
132 * When there is an exact match, it is considered a better
133 * match than anything else; the destination does not even
134 * call into this function in that case.
135 */
136 unsigned long delta_size, base_size, src_copied, literal_added;
137 unsigned long delta_limit;
138 int score;
139
140 /* We deal only with regular files. Symlink renames are handled
141 * only when they are exact matches --- in other words, no edits
142 * after renaming.
143 */
144 if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
145 return 0;
146
147 delta_size = ((src->size < dst->size) ?
148 (dst->size - src->size) : (src->size - dst->size));
149 base_size = ((src->size < dst->size) ? src->size : dst->size);
150
151 /* We would not consider edits that change the file size so
152 * drastically. delta_size must be smaller than
153 * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
154 *
155 * Note that base_size == 0 case is handled here already
156 * and the final score computation below would not have a
157 * divide-by-zero issue.
158 */
159 if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
160 return 0;
161
162 if (diff_populate_filespec(src, 0) || diff_populate_filespec(dst, 0))
163 return 0; /* error but caught downstream */
164
165
166 delta_limit = base_size * (MAX_SCORE-minimum_score) / MAX_SCORE;
167 if (diffcore_count_changes(src->data, src->size,
168 dst->data, dst->size,
169 &src->cnt_data, &dst->cnt_data,
170 delta_limit,
171 &src_copied, &literal_added))
172 return 0;
173
174 /* How similar are they?
175 * what percentage of material in dst are from source?
176 */
177 if (dst->size < src_copied)
178 score = MAX_SCORE;
179 else if (!dst->size)
180 score = 0; /* should not happen */
181 else
182 score = src_copied * MAX_SCORE / dst->size;
183 return score;
184 }
185
186 static void record_rename_pair(int dst_index, int src_index, int score)
187 {
188 struct diff_filespec *one, *two, *src, *dst;
189 struct diff_filepair *dp;
190
191 if (rename_dst[dst_index].pair)
192 die("internal error: dst already matched.");
193
194 src = rename_src[src_index].one;
195 one = alloc_filespec(src->path);
196 fill_filespec(one, src->sha1, src->mode);
197
198 dst = rename_dst[dst_index].two;
199 two = alloc_filespec(dst->path);
200 fill_filespec(two, dst->sha1, dst->mode);
201
202 dp = diff_queue(NULL, one, two);
203 dp->score = score;
204 dp->source_stays = rename_src[src_index].src_path_left;
205 rename_dst[dst_index].pair = dp;
206 }
207
208 /*
209 * We sort the rename similarity matrix with the score, in descending
210 * order (the most similar first).
211 */
212 static int score_compare(const void *a_, const void *b_)
213 {
214 const struct diff_score *a = a_, *b = b_;
215 return b->score - a->score;
216 }
217
218 static int compute_stays(struct diff_queue_struct *q,
219 struct diff_filespec *one)
220 {
221 int i;
222 for (i = 0; i < q->nr; i++) {
223 struct diff_filepair *p = q->queue[i];
224 if (strcmp(one->path, p->two->path))
225 continue;
226 if (DIFF_PAIR_RENAME(p)) {
227 return 0; /* something else is renamed into this */
228 }
229 }
230 return 1;
231 }
232
233 void diffcore_rename(struct diff_options *options)
234 {
235 int detect_rename = options->detect_rename;
236 int minimum_score = options->rename_score;
237 int rename_limit = options->rename_limit;
238 struct diff_queue_struct *q = &diff_queued_diff;
239 struct diff_queue_struct outq;
240 struct diff_score *mx;
241 int i, j, rename_count;
242 int num_create, num_src, dst_cnt;
243
244 if (!minimum_score)
245 minimum_score = DEFAULT_RENAME_SCORE;
246 rename_count = 0;
247
248 for (i = 0; i < q->nr; i++) {
249 struct diff_filepair *p = q->queue[i];
250 if (!DIFF_FILE_VALID(p->one))
251 if (!DIFF_FILE_VALID(p->two))
252 continue; /* unmerged */
253 else
254 locate_rename_dst(p->two, 1);
255 else if (!DIFF_FILE_VALID(p->two)) {
256 /* If the source is a broken "delete", and
257 * they did not really want to get broken,
258 * that means the source actually stays.
259 */
260 int stays = (p->broken_pair && !p->score);
261 register_rename_src(p->one, stays);
262 }
263 else if (detect_rename == DIFF_DETECT_COPY)
264 register_rename_src(p->one, 1);
265 }
266 if (rename_dst_nr == 0 || rename_src_nr == 0 ||
267 (0 < rename_limit && rename_limit < rename_dst_nr))
268 goto cleanup; /* nothing to do */
269
270 /* We really want to cull the candidates list early
271 * with cheap tests in order to avoid doing deltas.
272 */
273 for (i = 0; i < rename_dst_nr; i++) {
274 struct diff_filespec *two = rename_dst[i].two;
275 for (j = 0; j < rename_src_nr; j++) {
276 struct diff_filespec *one = rename_src[j].one;
277 if (!is_exact_match(one, two))
278 continue;
279 record_rename_pair(i, j, MAX_SCORE);
280 rename_count++;
281 break; /* we are done with this entry */
282 }
283 }
284
285 /* Have we run out the created file pool? If so we can avoid
286 * doing the delta matrix altogether.
287 */
288 if (rename_count == rename_dst_nr)
289 goto cleanup;
290
291 if (minimum_score == MAX_SCORE)
292 goto cleanup;
293
294 num_create = (rename_dst_nr - rename_count);
295 num_src = rename_src_nr;
296 mx = xmalloc(sizeof(*mx) * num_create * num_src);
297 for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
298 int base = dst_cnt * num_src;
299 struct diff_filespec *two = rename_dst[i].two;
300 if (rename_dst[i].pair)
301 continue; /* dealt with exact match already. */
302 for (j = 0; j < rename_src_nr; j++) {
303 struct diff_filespec *one = rename_src[j].one;
304 struct diff_score *m = &mx[base+j];
305 m->src = j;
306 m->dst = i;
307 m->score = estimate_similarity(one, two,
308 minimum_score);
309 }
310 free(two->cnt_data);
311 two->cnt_data = NULL;
312 dst_cnt++;
313 }
314 /* cost matrix sorted by most to least similar pair */
315 qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
316 for (i = 0; i < num_create * num_src; i++) {
317 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
318 if (dst->pair)
319 continue; /* already done, either exact or fuzzy. */
320 if (mx[i].score < minimum_score)
321 break; /* there is no more usable pair. */
322 record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
323 rename_count++;
324 }
325 free(mx);
326
327 cleanup:
328 /* At this point, we have found some renames and copies and they
329 * are recorded in rename_dst. The original list is still in *q.
330 */
331 outq.queue = NULL;
332 outq.nr = outq.alloc = 0;
333 for (i = 0; i < q->nr; i++) {
334 struct diff_filepair *p = q->queue[i];
335 struct diff_filepair *pair_to_free = NULL;
336
337 if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
338 /*
339 * Creation
340 *
341 * We would output this create record if it has
342 * not been turned into a rename/copy already.
343 */
344 struct diff_rename_dst *dst =
345 locate_rename_dst(p->two, 0);
346 if (dst && dst->pair) {
347 diff_q(&outq, dst->pair);
348 pair_to_free = p;
349 }
350 else
351 /* no matching rename/copy source, so
352 * record this as a creation.
353 */
354 diff_q(&outq, p);
355 }
356 else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
357 /*
358 * Deletion
359 *
360 * We would output this delete record if:
361 *
362 * (1) this is a broken delete and the counterpart
363 * broken create remains in the output; or
364 * (2) this is not a broken delete, and rename_dst
365 * does not have a rename/copy to move p->one->path
366 * out of existence.
367 *
368 * Otherwise, the counterpart broken create
369 * has been turned into a rename-edit; or
370 * delete did not have a matching create to
371 * begin with.
372 */
373 if (DIFF_PAIR_BROKEN(p)) {
374 /* broken delete */
375 struct diff_rename_dst *dst =
376 locate_rename_dst(p->one, 0);
377 if (dst && dst->pair)
378 /* counterpart is now rename/copy */
379 pair_to_free = p;
380 }
381 else {
382 for (j = 0; j < rename_dst_nr; j++) {
383 if (!rename_dst[j].pair)
384 continue;
385 if (strcmp(rename_dst[j].pair->
386 one->path,
387 p->one->path))
388 continue;
389 break;
390 }
391 if (j < rename_dst_nr)
392 /* this path remains */
393 pair_to_free = p;
394 }
395
396 if (pair_to_free)
397 ;
398 else
399 diff_q(&outq, p);
400 }
401 else if (!diff_unmodified_pair(p))
402 /* all the usual ones need to be kept */
403 diff_q(&outq, p);
404 else
405 /* no need to keep unmodified pairs */
406 pair_to_free = p;
407
408 if (pair_to_free)
409 diff_free_filepair(pair_to_free);
410 }
411 diff_debug_queue("done copying original", &outq);
412
413 free(q->queue);
414 *q = outq;
415 diff_debug_queue("done collapsing", q);
416
417 /* We need to see which rename source really stays here;
418 * earlier we only checked if the path is left in the result,
419 * but even if a path remains in the result, if that is coming
420 * from copying something else on top of it, then the original
421 * source is lost and does not stay.
422 */
423 for (i = 0; i < q->nr; i++) {
424 struct diff_filepair *p = q->queue[i];
425 if (DIFF_PAIR_RENAME(p) && p->source_stays) {
426 /* If one appears as the target of a rename-copy,
427 * then mark p->source_stays = 0; otherwise
428 * leave it as is.
429 */
430 p->source_stays = compute_stays(q, p->one);
431 }
432 }
433
434 for (i = 0; i < rename_dst_nr; i++) {
435 diff_free_filespec_data(rename_dst[i].two);
436 free(rename_dst[i].two);
437 }
438
439 free(rename_dst);
440 rename_dst = NULL;
441 rename_dst_nr = rename_dst_alloc = 0;
442 free(rename_src);
443 rename_src = NULL;
444 rename_src_nr = rename_src_alloc = 0;
445 return;
446 }