2 * Copyright (C) 2005 Junio C Hamano
8 #include "count-delta.h"
10 /* Table of rename/copy destinations */
12 static struct diff_rename_dst {
13 struct diff_filespec *two;
14 struct diff_filepair *pair;
16 static int rename_dst_nr, rename_dst_alloc;
18 static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two,
25 while (last > first) {
26 int next = (last + first) >> 1;
27 struct diff_rename_dst *dst = &(rename_dst[next]);
28 int cmp = strcmp(two->path, dst->two->path);
40 /* insert to make it at "first" */
41 if (rename_dst_alloc <= rename_dst_nr) {
42 rename_dst_alloc = alloc_nr(rename_dst_alloc);
43 rename_dst = xrealloc(rename_dst,
44 rename_dst_alloc * sizeof(*rename_dst));
47 if (first < rename_dst_nr)
48 memmove(rename_dst + first + 1, rename_dst + first,
49 (rename_dst_nr - first - 1) * sizeof(*rename_dst));
50 rename_dst[first].two = two;
51 rename_dst[first].pair = NULL;
52 return &(rename_dst[first]);
55 /* Table of rename/copy src files */
56 static struct diff_rename_src {
57 struct diff_filespec *one;
58 unsigned src_stays : 1;
60 static int rename_src_nr, rename_src_alloc;
62 static struct diff_rename_src *register_rename_src(struct diff_filespec *one,
69 while (last > first) {
70 int next = (last + first) >> 1;
71 struct diff_rename_src *src = &(rename_src[next]);
72 int cmp = strcmp(one->path, src->one->path);
82 /* insert to make it at "first" */
83 if (rename_src_alloc <= rename_src_nr) {
84 rename_src_alloc = alloc_nr(rename_src_alloc);
85 rename_src = xrealloc(rename_src,
86 rename_src_alloc * sizeof(*rename_src));
89 if (first < rename_src_nr)
90 memmove(rename_src + first + 1, rename_src + first,
91 (rename_src_nr - first - 1) * sizeof(*rename_src));
92 rename_src[first].one = one;
93 rename_src[first].src_stays = src_stays;
94 return &(rename_src[first]);
97 static int is_exact_match(struct diff_filespec *src, struct diff_filespec *dst)
99 if (src->sha1_valid && dst->sha1_valid &&
100 !memcmp(src->sha1, dst->sha1, 20))
102 if (diff_populate_filespec(src) || diff_populate_filespec(dst))
103 /* this is an error but will be caught downstream */
105 if (src->size == dst->size &&
106 !memcmp(src->data, dst->data, src->size))
112 int src; /* index in rename_src */
113 int dst; /* index in rename_dst */
117 static int estimate_similarity(struct diff_filespec *src,
118 struct diff_filespec *dst,
121 /* src points at a file that existed in the original tree (or
122 * optionally a file in the destination tree) and dst points
123 * at a newly created file. They may be quite similar, in which
124 * case we want to say src is renamed to dst or src is copied into
125 * dst, and then some edit has been applied to dst.
127 * Compare them and return how similar they are, representing
128 * the score as an integer between 0 and 10000, except
129 * where they match exactly it is considered better than anything
133 unsigned long delta_size, base_size;
136 /* We deal only with regular files. Symlink renames are handled
137 * only when they are exact matches --- in other words, no edits
140 if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
143 delta_size = ((src->size < dst->size) ?
144 (dst->size - src->size) : (src->size - dst->size));
145 base_size = ((src->size < dst->size) ? src->size : dst->size);
147 /* We would not consider edits that change the file size so
148 * drastically. delta_size must be smaller than
149 * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
150 * Note that base_size == 0 case is handled here already
151 * and the final score computation below would not have a
152 * divide-by-zero issue.
154 if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
157 delta = diff_delta(src->data, src->size,
158 dst->data, dst->size,
161 /* A delta that has a lot of literal additions would have
162 * big delta_size no matter what else it does.
164 if (base_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
167 /* Estimate the edit size by interpreting delta. */
168 delta_size = count_delta(delta, delta_size);
170 if (delta_size == UINT_MAX)
174 * Now we will give some score to it. 100% edit gets 0 points
175 * and 0% edit gets MAX_SCORE points.
177 score = MAX_SCORE - (MAX_SCORE * delta_size / base_size);
178 if (score < 0) return 0;
179 if (MAX_SCORE < score) return MAX_SCORE;
183 static void record_rename_pair(struct diff_queue_struct *renq,
184 int dst_index, int src_index, int score)
186 struct diff_filespec *one, *two, *src, *dst;
187 struct diff_filepair *dp;
189 if (rename_dst[dst_index].pair)
190 die("internal error: dst already matched.");
192 src = rename_src[src_index].one;
193 one = alloc_filespec(src->path);
194 fill_filespec(one, src->sha1, src->mode);
196 dst = rename_dst[dst_index].two;
197 two = alloc_filespec(dst->path);
198 fill_filespec(two, dst->sha1, dst->mode);
200 dp = diff_queue(renq, one, two);
201 dp->score = score ? : 1; /* make sure it is at least 1 */
202 dp->source_stays = rename_src[src_index].src_stays;
203 rename_dst[dst_index].pair = dp;
207 * We sort the rename similarity matrix with the score, in descending
208 * order (the most similar first).
210 static int score_compare(const void *a_, const void *b_)
212 const struct diff_score *a = a_, *b = b_;
213 return b->score - a->score;
216 int diff_scoreopt_parse(const char *opt)
218 int diglen, num, scale, i;
219 if (opt[0] != '-' || (opt[1] != 'M' && opt[1] != 'C'))
220 return -1; /* that is not a -M nor -C option */
221 diglen = strspn(opt+2, "0123456789");
222 if (diglen == 0 || strlen(opt+2) != diglen)
223 return 0; /* use default */
224 sscanf(opt+2, "%d", &num);
225 for (i = 0, scale = 1; i < diglen; i++)
228 /* user says num divided by scale and we say internally that
229 * is MAX_SCORE * num / scale.
231 return MAX_SCORE * num / scale;
234 void diffcore_rename(int detect_rename, int minimum_score)
236 struct diff_queue_struct *q = &diff_queued_diff;
237 struct diff_queue_struct renq, outq;
238 struct diff_score *mx;
240 int num_create, num_src, dst_cnt;
243 minimum_score = DEFAULT_MINIMUM_SCORE;
245 renq.nr = renq.alloc = 0;
247 for (i = 0; i < q->nr; i++) {
248 struct diff_filepair *p = q->queue[i];
249 if (!DIFF_FILE_VALID(p->one))
250 if (!DIFF_FILE_VALID(p->two))
251 continue; /* unmerged */
253 locate_rename_dst(p->two, 1);
254 else if (!DIFF_FILE_VALID(p->two))
255 register_rename_src(p->one, 0);
256 else if (detect_rename == DIFF_DETECT_COPY)
257 register_rename_src(p->one, 1);
259 if (rename_dst_nr == 0)
260 goto cleanup; /* nothing to do */
262 /* We really want to cull the candidates list early
263 * with cheap tests in order to avoid doing deltas.
265 for (i = 0; i < rename_dst_nr; i++) {
266 struct diff_filespec *two = rename_dst[i].two;
267 for (j = 0; j < rename_src_nr; j++) {
268 struct diff_filespec *one = rename_src[j].one;
269 if (!is_exact_match(one, two))
271 record_rename_pair(&renq, i, j, MAX_SCORE);
272 break; /* we are done with this entry */
275 diff_debug_queue("done detecting exact", &renq);
277 /* Have we run out the created file pool? If so we can avoid
278 * doing the delta matrix altogether.
280 if (renq.nr == rename_dst_nr)
283 num_create = (rename_dst_nr - renq.nr);
284 num_src = rename_src_nr;
285 mx = xmalloc(sizeof(*mx) * num_create * num_src);
286 for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
287 int base = dst_cnt * num_src;
288 struct diff_filespec *two = rename_dst[i].two;
289 if (rename_dst[i].pair)
290 continue; /* dealt with exact match already. */
291 for (j = 0; j < rename_src_nr; j++) {
292 struct diff_filespec *one = rename_src[j].one;
293 struct diff_score *m = &mx[base+j];
296 m->score = estimate_similarity(one, two,
301 /* cost matrix sorted by most to least similar pair */
302 qsort(mx, num_create * num_src, sizeof(*mx), score_compare);
303 for (i = 0; i < num_create * num_src; i++) {
304 struct diff_rename_dst *dst = &rename_dst[mx[i].dst];
306 continue; /* already done, either exact or fuzzy. */
307 if (mx[i].score < minimum_score)
308 break; /* there is no more usable pair. */
309 record_rename_pair(&renq, mx[i].dst, mx[i].src, mx[i].score);
312 diff_debug_queue("done detecting fuzzy", &renq);
315 /* At this point, we have found some renames and copies and they
316 * are kept in renq. The original list is still in *q.
319 outq.nr = outq.alloc = 0;
320 for (i = 0; i < q->nr; i++) {
321 struct diff_filepair *p = q->queue[i];
322 struct diff_rename_dst *dst = locate_rename_dst(p->two, 0);
323 struct diff_filepair *pair_to_free = NULL;
328 /* renq has rename/copy to produce
329 * this file already, so we do not
330 * emit the creation record in the
333 diff_q(&outq, dst->pair);
337 /* no matching rename/copy source, so record
338 * this as a creation.
342 else if (!diff_unmodified_pair(p))
343 /* all the usual ones need to be kept */
346 /* no need to keep unmodified pairs */
350 diff_free_filepair(pair_to_free);
352 diff_debug_queue("done copying original", &outq);
357 diff_debug_queue("done collapsing", q);
361 rename_dst_nr = rename_dst_alloc = 0;
364 rename_src_nr = rename_src_alloc = 0;