diffcore-rename.c 17.8 KB
Newer Older
1 2 3 4 5 6
/*
 * Copyright (C) 2005 Junio C Hamano
 */
#include "cache.h"
#include "diff.h"
#include "diffcore.h"
7
#include "object-store.h"
8
#include "hashmap.h"
9
#include "progress.h"
10

11 12 13 14 15 16 17
/* Table of rename/copy destinations */

static struct diff_rename_dst {
	struct diff_filespec *two;
	struct diff_filepair *pair;
} *rename_dst;
static int rename_dst_nr, rename_dst_alloc;
18

19
static int find_rename_dst(struct diff_filespec *two)
20
{
21 22 23 24 25 26 27 28 29
	int first, last;

	first = 0;
	last = rename_dst_nr;
	while (last > first) {
		int next = (last + first) >> 1;
		struct diff_rename_dst *dst = &(rename_dst[next]);
		int cmp = strcmp(two->path, dst->two->path);
		if (!cmp)
30
			return next;
31 32 33 34 35 36
		if (cmp < 0) {
			last = next;
			continue;
		}
		first = next+1;
	}
37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56
	return -first - 1;
}

static struct diff_rename_dst *locate_rename_dst(struct diff_filespec *two)
{
	int ofs = find_rename_dst(two);
	return ofs < 0 ? NULL : &rename_dst[ofs];
}

/*
 * Returns 0 on success, -1 if we found a duplicate.
 */
static int add_rename_dst(struct diff_filespec *two)
{
	int first = find_rename_dst(two);

	if (first >= 0)
		return -1;
	first = -first - 1;

57
	/* insert to make it at "first" */
58
	ALLOC_GROW(rename_dst, rename_dst_nr + 1, rename_dst_alloc);
59 60
	rename_dst_nr++;
	if (first < rename_dst_nr)
Gábor Szeder's avatar
Gábor Szeder committed
61 62
		MOVE_ARRAY(rename_dst + first + 1, rename_dst + first,
			   rename_dst_nr - first - 1);
Junio C Hamano's avatar
Junio C Hamano committed
63
	rename_dst[first].two = alloc_filespec(two->path);
64
	fill_filespec(rename_dst[first].two, &two->oid, two->oid_valid,
65
		      two->mode);
66
	rename_dst[first].pair = NULL;
67
	return 0;
68 69
}

70
/* Table of rename/copy src files */
71
static struct diff_rename_src {
72
	struct diff_filepair *p;
73
	unsigned short score; /* to remember the break score */
74 75
} *rename_src;
static int rename_src_nr, rename_src_alloc;
76

77
static struct diff_rename_src *register_rename_src(struct diff_filepair *p)
78 79
{
	int first, last;
80 81
	struct diff_filespec *one = p->one;
	unsigned short score = p->score;
82 83 84 85 86 87

	first = 0;
	last = rename_src_nr;
	while (last > first) {
		int next = (last + first) >> 1;
		struct diff_rename_src *src = &(rename_src[next]);
88
		int cmp = strcmp(one->path, src->p->one->path);
89 90 91 92 93 94 95 96
		if (!cmp)
			return src;
		if (cmp < 0) {
			last = next;
			continue;
		}
		first = next+1;
	}
97

98
	/* insert to make it at "first" */
99
	ALLOC_GROW(rename_src, rename_src_nr + 1, rename_src_alloc);
100 101
	rename_src_nr++;
	if (first < rename_src_nr)
Gábor Szeder's avatar
Gábor Szeder committed
102 103
		MOVE_ARRAY(rename_src + first + 1, rename_src + first,
			   rename_src_nr - first - 1);
104
	rename_src[first].p = p;
105
	rename_src[first].score = score;
106
	return &(rename_src[first]);
107 108
}

109 110 111 112 113 114 115 116 117 118 119 120 121 122 123
static int basename_same(struct diff_filespec *src, struct diff_filespec *dst)
{
	int src_len = strlen(src->path), dst_len = strlen(dst->path);
	while (src_len && dst_len) {
		char c1 = src->path[--src_len];
		char c2 = dst->path[--dst_len];
		if (c1 != c2)
			return 0;
		if (c1 == '/')
			return 1;
	}
	return (!src_len || src->path[src_len - 1] == '/') &&
		(!dst_len || dst->path[dst_len - 1] == '/');
}

124
struct diff_score {
125 126
	int src; /* index in rename_src */
	int dst; /* index in rename_dst */
127 128
	unsigned short score;
	short name_score;
129 130
};

131 132
static int estimate_similarity(struct repository *r,
			       struct diff_filespec *src,
133 134 135 136 137 138 139 140 141 142
			       struct diff_filespec *dst,
			       int minimum_score)
{
	/* src points at a file that existed in the original tree (or
	 * optionally a file in the destination tree) and dst points
	 * at a newly created file.  They may be quite similar, in which
	 * case we want to say src is renamed to dst or src is copied into
	 * dst, and then some edit has been applied to dst.
	 *
	 * Compare them and return how similar they are, representing
143 144 145 146 147
	 * the score as an integer between 0 and MAX_SCORE.
	 *
	 * When there is an exact match, it is considered a better
	 * match than anything else; the destination does not even
	 * call into this function in that case.
148
	 */
149
	unsigned long max_size, delta_size, base_size, src_copied, literal_added;
150 151
	int score;

152 153 154 155 156 157 158
	/* We deal only with regular files.  Symlink renames are handled
	 * only when they are exact matches --- in other words, no edits
	 * after renaming.
	 */
	if (!S_ISREG(src->mode) || !S_ISREG(dst->mode))
		return 0;

159 160 161 162 163 164 165 166 167
	/*
	 * Need to check that source and destination sizes are
	 * filled in before comparing them.
	 *
	 * If we already have "cnt_data" filled in, we know it's
	 * all good (avoid checking the size for zero, as that
	 * is a possible size - we really should have a flag to
	 * say whether the size is valid or not!)
	 */
168
	if (!src->cnt_data &&
169
	    diff_populate_filespec(r, src, CHECK_SIZE_ONLY))
170
		return 0;
171
	if (!dst->cnt_data &&
172
	    diff_populate_filespec(r, dst, CHECK_SIZE_ONLY))
173 174
		return 0;

175
	max_size = ((src->size > dst->size) ? src->size : dst->size);
176
	base_size = ((src->size < dst->size) ? src->size : dst->size);
177
	delta_size = max_size - base_size;
178

179 180
	/* We would not consider edits that change the file size so
	 * drastically.  delta_size must be smaller than
181
	 * (MAX_SCORE-minimum_score)/MAX_SCORE * min(src->size, dst->size).
182
	 *
183 184 185
	 * Note that base_size == 0 case is handled here already
	 * and the final score computation below would not have a
	 * divide-by-zero issue.
186
	 */
187
	if (max_size * (MAX_SCORE-minimum_score) < delta_size * MAX_SCORE)
188 189
		return 0;

190
	if (!src->cnt_data && diff_populate_filespec(r, src, 0))
191
		return 0;
192
	if (!dst->cnt_data && diff_populate_filespec(r, dst, 0))
193 194
		return 0;

195
	if (diffcore_count_changes(r, src, dst,
196
				   &src->cnt_data, &dst->cnt_data,
197
				   &src_copied, &literal_added))
198
		return 0;
199

200 201
	/* How similar are they?
	 * what percentage of material in dst are from source?
202
	 */
203
	if (!dst->size)
204
		score = 0; /* should not happen */
205
	else
206
		score = (int)(src_copied * MAX_SCORE / max_size);
207 208 209
	return score;
}

Junio C Hamano's avatar
Junio C Hamano committed
210
static void record_rename_pair(int dst_index, int src_index, int score)
211
{
212
	struct diff_filespec *src, *dst;
213
	struct diff_filepair *dp;
214

215 216
	if (rename_dst[dst_index].pair)
		die("internal error: dst already matched.");
217

218
	src = rename_src[src_index].p->one;
219
	src->rename_used++;
220
	src->count++;
221

222
	dst = rename_dst[dst_index].two;
223
	dst->count++;
224

225
	dp = diff_queue(NULL, src, dst);
226
	dp->renamed_pair = 1;
227 228 229 230
	if (!strcmp(src->path, dst->path))
		dp->score = rename_src[src_index].score;
	else
		dp->score = score;
231
	rename_dst[dst_index].pair = dp;
232 233 234 235
}

/*
 * We sort the rename similarity matrix with the score, in descending
236
 * order (the most similar first).
237 238 239 240
 */
static int score_compare(const void *a_, const void *b_)
{
	const struct diff_score *a = a_, *b = b_;
241

242 243 244 245 246 247
	/* sink the unused ones to the bottom */
	if (a->dst < 0)
		return (0 <= b->dst);
	else if (b->dst < 0)
		return -1;

248 249 250
	if (a->score == b->score)
		return b->name_score - a->name_score;

251 252 253
	return b->score - a->score;
}

254
struct file_similarity {
255
	struct hashmap_entry entry;
256
	int index;
257 258 259
	struct diff_filespec *filespec;
};

260 261
static unsigned int hash_filespec(struct repository *r,
				  struct diff_filespec *filespec)
262
{
263
	if (!filespec->oid_valid) {
264
		if (diff_populate_filespec(r, filespec, 0))
265
			return 0;
266 267
		hash_object_file(filespec->data, filespec->size, "blob",
				 &filespec->oid);
268
	}
269
	return sha1hash(filespec->oid.hash);
270 271
}

272
static int find_identical_files(struct hashmap *srcs,
273
				int dst_index,
274
				struct diff_options *options)
275 276 277
{
	int renames = 0;

278
	struct diff_filespec *target = rename_dst[dst_index].two;
279
	struct file_similarity *p, *best = NULL;
280 281 282
	int i = 100, best_score = -1;

	/*
283
	 * Find the best source match for specified destination.
284
	 */
285 286 287
	p = hashmap_get_from_hash(srcs,
				  hash_filespec(options->repo, target),
				  NULL);
288
	for (; p; p = hashmap_get_next(srcs, p)) {
289 290 291 292
		int score;
		struct diff_filespec *source = p->filespec;

		/* False hash collision? */
293
		if (!oideq(&source->oid, &target->oid))
294 295 296 297
			continue;
		/* Non-regular files? If so, the modes must match! */
		if (!S_ISREG(source->mode) || !S_ISREG(target->mode)) {
			if (source->mode != target->mode)
298
				continue;
299
		}
300 301 302 303 304 305 306 307 308 309
		/* Give higher scores to sources that haven't been used already */
		score = !source->rename_used;
		if (source->rename_used && options->detect_rename != DIFF_DETECT_COPY)
			continue;
		score += basename_same(source, target);
		if (score > best_score) {
			best = p;
			best_score = score;
			if (score == 2)
				break;
310
		}
311 312 313 314 315 316

		/* Too many identical alternatives? Pick one */
		if (!--i)
			break;
	}
	if (best) {
317
		record_rename_pair(dst_index, best->index, MAX_SCORE);
318 319
		renames++;
	}
320 321 322
	return renames;
}

323 324 325
static void insert_file_table(struct repository *r,
			      struct hashmap *table, int index,
			      struct diff_filespec *filespec)
326 327 328 329 330 331
{
	struct file_similarity *entry = xmalloc(sizeof(*entry));

	entry->index = index;
	entry->filespec = filespec;

332
	hashmap_entry_init(entry, hash_filespec(r, filespec));
333
	hashmap_add(table, entry);
334 335
}

336 337 338 339 340 341 342
/*
 * Find exact renames first.
 *
 * The first round matches up the up-to-date entries,
 * and then during the second round we try to match
 * cache-dirty entries as well.
 */
343
static int find_exact_renames(struct diff_options *options)
344
{
345
	int i, renames = 0;
346
	struct hashmap file_table;
347

348 349 350
	/* Add all sources to the hash table in reverse order, because
	 * later on they will be retrieved in LIFO order.
	 */
351
	hashmap_init(&file_table, NULL, NULL, rename_src_nr);
352
	for (i = rename_src_nr-1; i >= 0; i--)
353 354 355
		insert_file_table(options->repo,
				  &file_table, i,
				  rename_src[i].p->one);
356

357
	/* Walk the destinations and find best source match */
358
	for (i = 0; i < rename_dst_nr; i++)
359
		renames += find_identical_files(&file_table, i, options);
360

361 362
	/* Free the hash data structure and entries */
	hashmap_free(&file_table, 1);
363

364
	return renames;
365 366
}

367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382
#define NUM_CANDIDATE_PER_DST 4
static void record_if_better(struct diff_score m[], struct diff_score *o)
{
	int i, worst;

	/* find the worst one */
	worst = 0;
	for (i = 1; i < NUM_CANDIDATE_PER_DST; i++)
		if (score_compare(&m[i], &m[worst]) > 0)
			worst = i;

	/* is it better than the worst one? */
	if (score_compare(&m[worst], o) > 0)
		m[worst] = *o;
}

383 384 385 386 387 388
/*
 * Returns:
 * 0 if we are under the limit;
 * 1 if we need to disable inexact rename detection;
 * 2 if we would be under the limit if we were given -C instead of -C -C.
 */
389 390 391 392 393
static int too_many_rename_candidates(int num_create,
				      struct diff_options *options)
{
	int rename_limit = options->rename_limit;
	int num_src = rename_src_nr;
394
	int i;
395 396 397 398 399 400 401 402 403

	options->needed_rename_limit = 0;

	/*
	 * This basically does a test for the rename matrix not
	 * growing larger than a "rename_limit" square matrix, ie:
	 *
	 *    num_create * num_src > rename_limit * rename_limit
	 */
404 405
	if (rename_limit <= 0)
		rename_limit = 32767;
406
	if ((num_create <= rename_limit || num_src <= rename_limit) &&
407 408
	    ((uint64_t)num_create * (uint64_t)num_src
	     <= (uint64_t)rename_limit * (uint64_t)rename_limit))
409 410 411 412
		return 0;

	options->needed_rename_limit =
		num_src > num_create ? num_src : num_create;
413 414

	/* Are we running under -C -C? */
415
	if (!options->flags.find_copies_harder)
416 417 418 419 420 421 422 423 424
		return 1;

	/* Would we bust the limit if we were running under -C? */
	for (num_src = i = 0; i < rename_src_nr; i++) {
		if (diff_unmodified_pair(rename_src[i].p))
			continue;
		num_src++;
	}
	if ((num_create <= rename_limit || num_src <= rename_limit) &&
425 426
	    ((uint64_t)num_create * (uint64_t)num_src
	     <= (uint64_t)rename_limit * (uint64_t)rename_limit))
427
		return 2;
428 429 430
	return 1;
}

431 432 433 434 435 436 437 438 439 440 441 442 443
static int find_renames(struct diff_score *mx, int dst_cnt, int minimum_score, int copies)
{
	int count = 0, i;

	for (i = 0; i < dst_cnt * NUM_CANDIDATE_PER_DST; i++) {
		struct diff_rename_dst *dst;

		if ((mx[i].dst < 0) ||
		    (mx[i].score < minimum_score))
			break; /* there is no more usable pair. */
		dst = &rename_dst[mx[i].dst];
		if (dst->pair)
			continue; /* already done, either exact or fuzzy. */
444
		if (!copies && rename_src[mx[i].src].p->one->rename_used)
445 446 447 448 449 450 451
			continue;
		record_rename_pair(mx[i].dst, mx[i].src, mx[i].score);
		count++;
	}
	return count;
}

452
void diffcore_rename(struct diff_options *options)
453
{
454 455
	int detect_rename = options->detect_rename;
	int minimum_score = options->rename_score;
456
	struct diff_queue_struct *q = &diff_queued_diff;
Junio C Hamano's avatar
Junio C Hamano committed
457
	struct diff_queue_struct outq;
458
	struct diff_score *mx;
459
	int i, j, rename_count, skip_unmodified = 0;
460
	int num_create, dst_cnt;
461
	struct progress *progress = NULL;
462

463
	if (!minimum_score)
464
		minimum_score = DEFAULT_RENAME_SCORE;
465 466

	for (i = 0; i < q->nr; i++) {
467
		struct diff_filepair *p = q->queue[i];
468
		if (!DIFF_FILE_VALID(p->one)) {
469
			if (!DIFF_FILE_VALID(p->two))
470
				continue; /* unmerged */
471 472 473
			else if (options->single_follow &&
				 strcmp(options->single_follow, p->two->path))
				continue; /* not interested */
474
			else if (!options->flags.rename_empty &&
475
				 is_empty_blob_oid(&p->two->oid))
476
				continue;
477 478 479 480 481 482
			else if (add_rename_dst(p->two) < 0) {
				warning("skipping rename detection, detected"
					" duplicate destination '%s'",
					p->two->path);
				goto cleanup;
			}
483
		}
484
		else if (!options->flags.rename_empty &&
485
			 is_empty_blob_oid(&p->one->oid))
486
			continue;
487
		else if (!DIFF_PAIR_UNMERGED(p) && !DIFF_FILE_VALID(p->two)) {
488 489
			/*
			 * If the source is a broken "delete", and
490 491
			 * they did not really want to get broken,
			 * that means the source actually stays.
492 493 494 495 496
			 * So we increment the "rename_used" score
			 * by one, to indicate ourselves as a user
			 */
			if (p->broken_pair && !p->score)
				p->one->rename_used++;
497
			register_rename_src(p);
498 499 500 501 502
		}
		else if (detect_rename == DIFF_DETECT_COPY) {
			/*
			 * Increment the "rename_used" score by
			 * one, to indicate ourselves as a user.
503
			 */
504
			p->one->rename_used++;
505
			register_rename_src(p);
506
		}
507
	}
508
	if (rename_dst_nr == 0 || rename_src_nr == 0)
509 510
		goto cleanup; /* nothing to do */

511 512 513 514
	/*
	 * We really want to cull the candidates list early
	 * with cheap tests in order to avoid doing deltas.
	 */
515
	rename_count = find_exact_renames(options);
516

517 518 519 520 521 522 523 524 525 526 527 528 529 530
	/* Did we only want exact renames? */
	if (minimum_score == MAX_SCORE)
		goto cleanup;

	/*
	 * Calculate how many renames are left (but all the source
	 * files still remain as options for rename/copies!)
	 */
	num_create = (rename_dst_nr - rename_count);

	/* All done? */
	if (!num_create)
		goto cleanup;

531 532
	switch (too_many_rename_candidates(num_create, options)) {
	case 1:
533
		goto cleanup;
534 535 536 537 538 539 540
	case 2:
		options->degraded_cc_to_c = 1;
		skip_unmodified = 1;
		break;
	default:
		break;
	}
541

542
	if (options->show_rename_progress) {
543
		progress = start_delayed_progress(
544
				_("Performing inexact rename detection"),
545
				(uint64_t)rename_dst_nr * (uint64_t)rename_src_nr);
546 547
	}

548
	mx = xcalloc(st_mult(NUM_CANDIDATE_PER_DST, num_create), sizeof(*mx));
549 550
	for (dst_cnt = i = 0; i < rename_dst_nr; i++) {
		struct diff_filespec *two = rename_dst[i].two;
551 552
		struct diff_score *m;

553
		if (rename_dst[i].pair)
554
			continue; /* dealt with exact match already. */
555 556 557 558 559

		m = &mx[dst_cnt * NUM_CANDIDATE_PER_DST];
		for (j = 0; j < NUM_CANDIDATE_PER_DST; j++)
			m[j].dst = -1;

560
		for (j = 0; j < rename_src_nr; j++) {
561
			struct diff_filespec *one = rename_src[j].p->one;
562
			struct diff_score this_src;
563 564 565 566 567

			if (skip_unmodified &&
			    diff_unmodified_pair(rename_src[j].p))
				continue;

568 569
			this_src.score = estimate_similarity(options->repo,
							     one, two,
570 571 572 573 574
							     minimum_score);
			this_src.name_score = basename_same(one, two);
			this_src.dst = i;
			this_src.src = j;
			record_if_better(m, &this_src);
575 576 577 578
			/*
			 * Once we run estimate_similarity,
			 * We do not need the text anymore.
			 */
579
			diff_free_filespec_blob(one);
580
			diff_free_filespec_blob(two);
581 582
		}
		dst_cnt++;
583
		display_progress(progress, (uint64_t)(i+1)*(uint64_t)rename_src_nr);
584
	}
585
	stop_progress(&progress);
586

587
	/* cost matrix sorted by most to least similar pair */
René Scharfe's avatar
René Scharfe committed
588
	QSORT(mx, dst_cnt * NUM_CANDIDATE_PER_DST, score_compare);
589

590 591 592
	rename_count += find_renames(mx, dst_cnt, minimum_score, 0);
	if (detect_rename == DIFF_DETECT_COPY)
		rename_count += find_renames(mx, dst_cnt, minimum_score, 1);
593 594
	free(mx);

595
 cleanup:
596
	/* At this point, we have found some renames and copies and they
Junio C Hamano's avatar
Junio C Hamano committed
597
	 * are recorded in rename_dst.  The original list is still in *q.
598
	 */
Bo Yang's avatar
Bo Yang committed
599
	DIFF_QUEUE_CLEAR(&outq);
600
	for (i = 0; i < q->nr; i++) {
601 602 603
		struct diff_filepair *p = q->queue[i];
		struct diff_filepair *pair_to_free = NULL;

604 605 606 607
		if (DIFF_PAIR_UNMERGED(p)) {
			diff_q(&outq, p);
		}
		else if (!DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
608 609 610 611 612 613
			/*
			 * Creation
			 *
			 * We would output this create record if it has
			 * not been turned into a rename/copy already.
			 */
614
			struct diff_rename_dst *dst = locate_rename_dst(p->two);
615
			if (dst && dst->pair) {
616 617 618 619
				diff_q(&outq, dst->pair);
				pair_to_free = p;
			}
			else
620 621
				/* no matching rename/copy source, so
				 * record this as a creation.
622 623
				 */
				diff_q(&outq, p);
624
		}
625 626 627 628
		else if (DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two)) {
			/*
			 * Deletion
			 *
629 630 631 632
			 * We would output this delete record if:
			 *
			 * (1) this is a broken delete and the counterpart
			 *     broken create remains in the output; or
Junio C Hamano's avatar
Junio C Hamano committed
633 634 635
			 * (2) this is not a broken delete, and rename_dst
			 *     does not have a rename/copy to move p->one->path
			 *     out of existence.
636 637 638 639 640
			 *
			 * Otherwise, the counterpart broken create
			 * has been turned into a rename-edit; or
			 * delete did not have a matching create to
			 * begin with.
641
			 */
642 643
			if (DIFF_PAIR_BROKEN(p)) {
				/* broken delete */
644
				struct diff_rename_dst *dst = locate_rename_dst(p->one);
645 646 647 648 649
				if (dst && dst->pair)
					/* counterpart is now rename/copy */
					pair_to_free = p;
			}
			else {
650
				if (p->one->rename_used)
651 652 653
					/* this path remains */
					pair_to_free = p;
			}
654 655 656 657 658 659

			if (pair_to_free)
				;
			else
				diff_q(&outq, p);
		}
660
		else if (!diff_unmodified_pair(p))
661
			/* all the usual ones need to be kept */
662
			diff_q(&outq, p);
663 664 665 666
		else
			/* no need to keep unmodified pairs */
			pair_to_free = p;

667 668
		if (pair_to_free)
			diff_free_filepair(pair_to_free);
669
	}
670
	diff_debug_queue("done copying original", &outq);
671

672 673 674
	free(q->queue);
	*q = outq;
	diff_debug_queue("done collapsing", q);
675

676 677
	for (i = 0; i < rename_dst_nr; i++)
		free_filespec(rename_dst[i].two);
Junio C Hamano's avatar
Junio C Hamano committed
678

679
	FREE_AND_NULL(rename_dst);
680
	rename_dst_nr = rename_dst_alloc = 0;
681
	FREE_AND_NULL(rename_src);
682
	rename_src_nr = rename_src_alloc = 0;
683 684
	return;
}