blame.c 50.9 KB
Newer Older
1 2
#include "cache.h"
#include "refs.h"
3
#include "object-store.h"
4
#include "cache-tree.h"
5 6 7
#include "mergesort.h"
#include "diff.h"
#include "diffcore.h"
8
#include "tag.h"
9
#include "blame.h"
10
#include "alloc.h"
11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
#include "commit-slab.h"

define_commit_slab(blame_suspects, struct blame_origin *);
static struct blame_suspects blame_suspects;

struct blame_origin *get_blame_suspects(struct commit *commit)
{
	struct blame_origin **result;

	result = blame_suspects_peek(&blame_suspects, commit);

	return result ? *result : NULL;
}

static void set_blame_suspects(struct commit *commit, struct blame_origin *origin)
{
	*blame_suspects_at(&blame_suspects, commit) = origin;
}
29 30 31 32 33 34 35 36 37

void blame_origin_decref(struct blame_origin *o)
{
	if (o && --o->refcnt <= 0) {
		struct blame_origin *p, *l = NULL;
		if (o->previous)
			blame_origin_decref(o->previous);
		free(o->file.ptr);
		/* Should be present exactly once in commit chain */
38
		for (p = get_blame_suspects(o->commit); p; l = p, p = p->next) {
39 40 41 42
			if (p == o) {
				if (l)
					l->next = p->next;
				else
43
					set_blame_suspects(o->commit, p->next);
44 45 46 47 48 49 50 51 52 53 54 55 56 57
				free(o);
				return;
			}
		}
		die("internal error in blame_origin_decref");
	}
}

/*
 * Given a commit and a path in it, create a new origin structure.
 * The callers that add blame to the scoreboard should use
 * get_origin() to obtain shared, refcounted copy instead of calling
 * this function directly.
 */
58
static struct blame_origin *make_origin(struct commit *commit, const char *path)
59 60 61 62 63
{
	struct blame_origin *o;
	FLEX_ALLOC_STR(o, path, path);
	o->commit = commit;
	o->refcnt = 1;
64 65
	o->next = get_blame_suspects(commit);
	set_blame_suspects(commit, o);
66 67 68 69 70 71 72
	return o;
}

/*
 * Locate an existing origin or create a new one.
 * This moves the origin to front position in the commit util list.
 */
73
static struct blame_origin *get_origin(struct commit *commit, const char *path)
74 75 76
{
	struct blame_origin *o, *l;

77
	for (o = get_blame_suspects(commit), l = NULL; o; l = o, o = o->next) {
78 79 80 81
		if (!strcmp(o->path, path)) {
			/* bump to front */
			if (l) {
				l->next = o->next;
82 83
				o->next = get_blame_suspects(commit);
				set_blame_suspects(commit, o);
84 85 86 87 88 89
			}
			return blame_origin_incref(o);
		}
	}
	return make_origin(commit, path);
}
90 91 92



93
static void verify_working_tree_path(struct repository *r,
94
				     struct commit *work_tree, const char *path)
95 96 97 98 99 100 101 102 103
{
	struct commit_list *parents;
	int pos;

	for (parents = work_tree->parents; parents; parents = parents->next) {
		const struct object_id *commit_oid = &parents->item->object.oid;
		struct object_id blob_oid;
		unsigned mode;

104
		if (!get_tree_entry(commit_oid, path, &blob_oid, &mode) &&
105
		    oid_object_info(r, &blob_oid, NULL) == OBJ_BLOB)
106 107 108
			return;
	}

109
	pos = index_name_pos(r->index, path, strlen(path));
110 111
	if (pos >= 0)
		; /* path is in the index */
112 113
	else if (-1 - pos < r->index->cache_nr &&
		 !strcmp(r->index->cache[-1 - pos]->name, path))
114 115 116 117 118
		; /* path is in the index, unmerged */
	else
		die("no such path '%s' in HEAD", path);
}

119 120 121
static struct commit_list **append_parent(struct repository *r,
					  struct commit_list **tail,
					  const struct object_id *oid)
122 123 124
{
	struct commit *parent;

125
	parent = lookup_commit_reference(r, oid);
126 127 128 129 130
	if (!parent)
		die("no such commit %s", oid_to_hex(oid));
	return &commit_list_insert(parent, tail)->next;
}

131 132
static void append_merge_parents(struct repository *r,
				 struct commit_list **tail)
133 134 135 136
{
	int merge_head;
	struct strbuf line = STRBUF_INIT;

137
	merge_head = open(git_path_merge_head(r), O_RDONLY);
138 139 140
	if (merge_head < 0) {
		if (errno == ENOENT)
			return;
141
		die("cannot open '%s' for reading",
142
		    git_path_merge_head(r));
143 144 145 146 147
	}

	while (!strbuf_getwholeline_fd(&line, merge_head, '\n')) {
		struct object_id oid;
		if (line.len < GIT_SHA1_HEXSZ || get_oid_hex(line.buf, &oid))
148
			die("unknown line in '%s': %s",
149 150
			    git_path_merge_head(r), line.buf);
		tail = append_parent(r, tail, &oid);
151 152 153 154 155 156 157 158 159 160
	}
	close(merge_head);
	strbuf_release(&line);
}

/*
 * This isn't as simple as passing sb->buf and sb->len, because we
 * want to transfer ownership of the buffer to the commit (so we
 * must use detach).
 */
161 162 163
static void set_commit_buffer_from_strbuf(struct repository *r,
					  struct commit *c,
					  struct strbuf *sb)
164 165 166
{
	size_t len;
	void *buf = strbuf_detach(sb, &len);
167
	set_commit_buffer(r, c, buf, len);
168 169 170 171 172 173
}

/*
 * Prepare a dummy commit that represents the work tree (or staged) item.
 * Note that annotating work tree item never works in the reverse.
 */
174
static struct commit *fake_working_tree_commit(struct repository *r,
175
					       struct diff_options *opt,
176 177
					       const char *path,
					       const char *contents_from)
178 179 180 181 182 183 184 185
{
	struct commit *commit;
	struct blame_origin *origin;
	struct commit_list **parent_tail, *parent;
	struct object_id head_oid;
	struct strbuf buf = STRBUF_INIT;
	const char *ident;
	time_t now;
186
	int len;
187 188 189 190
	struct cache_entry *ce;
	unsigned mode;
	struct strbuf msg = STRBUF_INIT;

191
	repo_read_index(r);
192
	time(&now);
193
	commit = alloc_commit_node(r);
194 195 196 197
	commit->object.parsed = 1;
	commit->date = now;
	parent_tail = &commit->parents;

198
	if (!resolve_ref_unsafe("HEAD", RESOLVE_REF_READING, &head_oid, NULL))
199 200
		die("no such ref: HEAD");

201 202
	parent_tail = append_parent(r, parent_tail, &head_oid);
	append_merge_parents(r, parent_tail);
203
	verify_working_tree_path(r, commit, path);
204 205 206

	origin = make_origin(commit, path);

207 208
	ident = fmt_ident("Not Committed Yet", "not.committed.yet",
			WANT_BLANK_IDENT, NULL, 0);
209 210 211 212 213 214 215 216 217 218 219
	strbuf_addstr(&msg, "tree 0000000000000000000000000000000000000000\n");
	for (parent = commit->parents; parent; parent = parent->next)
		strbuf_addf(&msg, "parent %s\n",
			    oid_to_hex(&parent->item->object.oid));
	strbuf_addf(&msg,
		    "author %s\n"
		    "committer %s\n\n"
		    "Version of %s from %s\n",
		    ident, ident, path,
		    (!contents_from ? path :
		     (!strcmp(contents_from, "-") ? "standard input" : contents_from)));
220
	set_commit_buffer_from_strbuf(r, commit, &msg);
221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241

	if (!contents_from || strcmp("-", contents_from)) {
		struct stat st;
		const char *read_from;
		char *buf_ptr;
		unsigned long buf_len;

		if (contents_from) {
			if (stat(contents_from, &st) < 0)
				die_errno("Cannot stat '%s'", contents_from);
			read_from = contents_from;
		}
		else {
			if (lstat(path, &st) < 0)
				die_errno("Cannot lstat '%s'", path);
			read_from = path;
		}
		mode = canon_mode(st.st_mode);

		switch (st.st_mode & S_IFMT) {
		case S_IFREG:
242
			if (opt->flags.allow_textconv &&
243
			    textconv_object(r, read_from, mode, &null_oid, 0, &buf_ptr, &buf_len))
244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261
				strbuf_attach(&buf, buf_ptr, buf_len, buf_len + 1);
			else if (strbuf_read_file(&buf, read_from, st.st_size) != st.st_size)
				die_errno("cannot open or read '%s'", read_from);
			break;
		case S_IFLNK:
			if (strbuf_readlink(&buf, read_from, st.st_size) < 0)
				die_errno("cannot readlink '%s'", read_from);
			break;
		default:
			die("unsupported file type %s", read_from);
		}
	}
	else {
		/* Reading from stdin */
		mode = 0;
		if (strbuf_read(&buf, 0, 0) < 0)
			die_errno("failed to read from stdin");
	}
262
	convert_to_git(r->index, path, buf.buf, buf.len, &buf, 0);
263 264
	origin->file.ptr = buf.buf;
	origin->file.size = buf.len;
265
	pretend_object_file(buf.buf, buf.len, OBJ_BLOB, &origin->blob_oid);
266 267 268 269 270 271 272

	/*
	 * Read the current index, replace the path entry with
	 * origin->blob_sha1 without mucking with its mode or type
	 * bits; we are not going to write this index out -- we just
	 * want to run "diff-index --cached".
	 */
273
	discard_index(r->index);
274
	repo_read_index(r);
275 276 277

	len = strlen(path);
	if (!mode) {
278
		int pos = index_name_pos(r->index, path, len);
279
		if (0 <= pos)
280
			mode = r->index->cache[pos]->ce_mode;
281 282 283 284
		else
			/* Let's not bother reading from HEAD tree */
			mode = S_IFREG | 0644;
	}
285
	ce = make_empty_cache_entry(r->index, len);
286 287 288 289 290
	oidcpy(&ce->oid, &origin->blob_oid);
	memcpy(ce->name, path, len);
	ce->ce_flags = create_ce_flags(0);
	ce->ce_namelen = len;
	ce->ce_mode = create_ce_mode(mode);
291
	add_index_entry(r->index, ce,
292
			ADD_CACHE_OK_TO_ADD | ADD_CACHE_OK_TO_REPLACE);
293

294
	cache_tree_invalidate_path(r->index, path);
295 296 297

	return commit;
}
298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325



static int diff_hunks(mmfile_t *file_a, mmfile_t *file_b,
		      xdl_emit_hunk_consume_func_t hunk_func, void *cb_data, int xdl_opts)
{
	xpparam_t xpp = {0};
	xdemitconf_t xecfg = {0};
	xdemitcb_t ecb = {NULL};

	xpp.flags = xdl_opts;
	xecfg.hunk_func = hunk_func;
	ecb.priv = cb_data;
	return xdi_diff(file_a, file_b, &xpp, &xecfg, &ecb);
}

/*
 * Given an origin, prepare mmfile_t structure to be used by the
 * diff machinery
 */
static void fill_origin_blob(struct diff_options *opt,
			     struct blame_origin *o, mmfile_t *file, int *num_read_blob)
{
	if (!o->file.ptr) {
		enum object_type type;
		unsigned long file_size;

		(*num_read_blob)++;
326
		if (opt->flags.allow_textconv &&
327 328
		    textconv_object(opt->repo, o->path, o->mode,
				    &o->blob_oid, 1, &file->ptr, &file_size))
329 330
			;
		else
331 332
			file->ptr = read_object_file(&o->blob_oid, &type,
						     &file_size);
333 334 335 336 337 338 339 340 341 342 343 344 345 346
		file->size = file_size;

		if (!file->ptr)
			die("Cannot read blob %s for path %s",
			    oid_to_hex(&o->blob_oid),
			    o->path);
		o->file = *file;
	}
	else
		*file = o->file;
}

static void drop_origin_blob(struct blame_origin *o)
{
347
	FREE_AND_NULL(o->file.ptr);
348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441
}

/*
 * Any merge of blames happens on lists of blames that arrived via
 * different parents in a single suspect.  In this case, we want to
 * sort according to the suspect line numbers as opposed to the final
 * image line numbers.  The function body is somewhat longish because
 * it avoids unnecessary writes.
 */

static struct blame_entry *blame_merge(struct blame_entry *list1,
				       struct blame_entry *list2)
{
	struct blame_entry *p1 = list1, *p2 = list2,
		**tail = &list1;

	if (!p1)
		return p2;
	if (!p2)
		return p1;

	if (p1->s_lno <= p2->s_lno) {
		do {
			tail = &p1->next;
			if ((p1 = *tail) == NULL) {
				*tail = p2;
				return list1;
			}
		} while (p1->s_lno <= p2->s_lno);
	}
	for (;;) {
		*tail = p2;
		do {
			tail = &p2->next;
			if ((p2 = *tail) == NULL)  {
				*tail = p1;
				return list1;
			}
		} while (p1->s_lno > p2->s_lno);
		*tail = p1;
		do {
			tail = &p1->next;
			if ((p1 = *tail) == NULL) {
				*tail = p2;
				return list1;
			}
		} while (p1->s_lno <= p2->s_lno);
	}
}

static void *get_next_blame(const void *p)
{
	return ((struct blame_entry *)p)->next;
}

static void set_next_blame(void *p1, void *p2)
{
	((struct blame_entry *)p1)->next = p2;
}

/*
 * Final image line numbers are all different, so we don't need a
 * three-way comparison here.
 */

static int compare_blame_final(const void *p1, const void *p2)
{
	return ((struct blame_entry *)p1)->lno > ((struct blame_entry *)p2)->lno
		? 1 : -1;
}

static int compare_blame_suspect(const void *p1, const void *p2)
{
	const struct blame_entry *s1 = p1, *s2 = p2;
	/*
	 * to allow for collating suspects, we sort according to the
	 * respective pointer value as the primary sorting criterion.
	 * The actual relation is pretty unimportant as long as it
	 * establishes a total order.  Comparing as integers gives us
	 * that.
	 */
	if (s1->suspect != s2->suspect)
		return (intptr_t)s1->suspect > (intptr_t)s2->suspect ? 1 : -1;
	if (s1->s_lno == s2->s_lno)
		return 0;
	return s1->s_lno > s2->s_lno ? 1 : -1;
}

void blame_sort_final(struct blame_scoreboard *sb)
{
	sb->ent = llist_mergesort(sb->ent, get_next_blame, set_next_blame,
				  compare_blame_final);
}

442 443 444 445 446 447 448
static int compare_commits_by_reverse_commit_date(const void *a,
						  const void *b,
						  void *c)
{
	return -compare_commits_by_commit_date(a, b, c);
}

449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509
/*
 * For debugging -- origin is refcounted, and this asserts that
 * we do not underflow.
 */
static void sanity_check_refcnt(struct blame_scoreboard *sb)
{
	int baa = 0;
	struct blame_entry *ent;

	for (ent = sb->ent; ent; ent = ent->next) {
		/* Nobody should have zero or negative refcnt */
		if (ent->suspect->refcnt <= 0) {
			fprintf(stderr, "%s in %s has negative refcnt %d\n",
				ent->suspect->path,
				oid_to_hex(&ent->suspect->commit->object.oid),
				ent->suspect->refcnt);
			baa = 1;
		}
	}
	if (baa)
		sb->on_sanity_fail(sb, baa);
}

/*
 * If two blame entries that are next to each other came from
 * contiguous lines in the same origin (i.e. <commit, path> pair),
 * merge them together.
 */
void blame_coalesce(struct blame_scoreboard *sb)
{
	struct blame_entry *ent, *next;

	for (ent = sb->ent; ent && (next = ent->next); ent = next) {
		if (ent->suspect == next->suspect &&
		    ent->s_lno + ent->num_lines == next->s_lno) {
			ent->num_lines += next->num_lines;
			ent->next = next->next;
			blame_origin_decref(next->suspect);
			free(next);
			ent->score = 0;
			next = ent; /* again */
		}
	}

	if (sb->debug) /* sanity */
		sanity_check_refcnt(sb);
}

/*
 * Merge the given sorted list of blames into a preexisting origin.
 * If there were no previous blames to that commit, it is entered into
 * the commit priority queue of the score board.
 */

static void queue_blames(struct blame_scoreboard *sb, struct blame_origin *porigin,
			 struct blame_entry *sorted)
{
	if (porigin->suspects)
		porigin->suspects = blame_merge(porigin->suspects, sorted);
	else {
		struct blame_origin *o;
510
		for (o = get_blame_suspects(porigin->commit); o; o = o->next) {
511 512 513 514 515 516 517 518 519 520
			if (o->suspects) {
				porigin->suspects = sorted;
				return;
			}
		}
		porigin->suspects = sorted;
		prio_queue_put(&sb->commits, porigin->commit);
	}
}

521 522 523 524 525 526 527 528 529
/*
 * Fill the blob_sha1 field of an origin if it hasn't, so that later
 * call to fill_origin_blob() can use it to locate the data.  blob_sha1
 * for an origin is also used to pass the blame for the entire file to
 * the parent to detect the case where a child's blob is identical to
 * that of its parent's.
 *
 * This also fills origin->mode for corresponding tree path.
 */
530
static int fill_blob_sha1_and_mode(struct repository *r,
531
				   struct blame_origin *origin)
532 533 534
{
	if (!is_null_oid(&origin->blob_oid))
		return 0;
535
	if (get_tree_entry(&origin->commit->object.oid, origin->path, &origin->blob_oid, &origin->mode))
536
		goto error_out;
537
	if (oid_object_info(r, &origin->blob_oid, NULL) != OBJ_BLOB)
538 539 540 541 542 543 544 545
		goto error_out;
	return 0;
 error_out:
	oidclr(&origin->blob_oid);
	origin->mode = S_IFINVALID;
	return -1;
}

546 547 548 549
/*
 * We have an origin -- check if the same path exists in the
 * parent and return an origin structure to represent it.
 */
550 551 552
static struct blame_origin *find_origin(struct repository *r,
					struct commit *parent,
					struct blame_origin *origin)
553 554 555 556 557 558
{
	struct blame_origin *porigin;
	struct diff_options diff_opts;
	const char *paths[2];

	/* First check any existing origins */
559
	for (porigin = get_blame_suspects(parent); porigin; porigin = porigin->next)
560 561 562 563 564 565 566 567 568 569 570 571
		if (!strcmp(porigin->path, origin->path)) {
			/*
			 * The same path between origin and its parent
			 * without renaming -- the most common case.
			 */
			return blame_origin_incref (porigin);
		}

	/* See if the origin->path is different between parent
	 * and origin first.  Most of the time they are the
	 * same and diff-tree is fairly efficient about this.
	 */
572
	repo_diff_setup(r, &diff_opts);
573
	diff_opts.flags.recursive = 1;
574 575 576 577 578 579 580 581 582 583 584
	diff_opts.detect_rename = 0;
	diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
	paths[0] = origin->path;
	paths[1] = NULL;

	parse_pathspec(&diff_opts.pathspec,
		       PATHSPEC_ALL_MAGIC & ~PATHSPEC_LITERAL,
		       PATHSPEC_LITERAL_PATH, "", paths);
	diff_setup_done(&diff_opts);

	if (is_null_oid(&origin->commit->object.oid))
585
		do_diff_cache(get_commit_tree_oid(parent), &diff_opts);
586
	else
587 588
		diff_tree_oid(get_commit_tree_oid(parent),
			      get_commit_tree_oid(origin->commit),
589
			      "", &diff_opts);
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638
	diffcore_std(&diff_opts);

	if (!diff_queued_diff.nr) {
		/* The path is the same as parent */
		porigin = get_origin(parent, origin->path);
		oidcpy(&porigin->blob_oid, &origin->blob_oid);
		porigin->mode = origin->mode;
	} else {
		/*
		 * Since origin->path is a pathspec, if the parent
		 * commit had it as a directory, we will see a whole
		 * bunch of deletion of files in the directory that we
		 * do not care about.
		 */
		int i;
		struct diff_filepair *p = NULL;
		for (i = 0; i < diff_queued_diff.nr; i++) {
			const char *name;
			p = diff_queued_diff.queue[i];
			name = p->one->path ? p->one->path : p->two->path;
			if (!strcmp(name, origin->path))
				break;
		}
		if (!p)
			die("internal error in blame::find_origin");
		switch (p->status) {
		default:
			die("internal error in blame::find_origin (%c)",
			    p->status);
		case 'M':
			porigin = get_origin(parent, origin->path);
			oidcpy(&porigin->blob_oid, &p->one->oid);
			porigin->mode = p->one->mode;
			break;
		case 'A':
		case 'T':
			/* Did not exist in parent, or type changed */
			break;
		}
	}
	diff_flush(&diff_opts);
	clear_pathspec(&diff_opts.pathspec);
	return porigin;
}

/*
 * We have an origin -- find the path that corresponds to it in its
 * parent and return an origin structure to represent it.
 */
639 640 641
static struct blame_origin *find_rename(struct repository *r,
					struct commit *parent,
					struct blame_origin *origin)
642 643 644 645 646
{
	struct blame_origin *porigin = NULL;
	struct diff_options diff_opts;
	int i;

647
	repo_diff_setup(r, &diff_opts);
648
	diff_opts.flags.recursive = 1;
649 650 651 652 653 654
	diff_opts.detect_rename = DIFF_DETECT_RENAME;
	diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
	diff_opts.single_follow = origin->path;
	diff_setup_done(&diff_opts);

	if (is_null_oid(&origin->commit->object.oid))
655
		do_diff_cache(get_commit_tree_oid(parent), &diff_opts);
656
	else
657 658
		diff_tree_oid(get_commit_tree_oid(parent),
			      get_commit_tree_oid(origin->commit),
659
			      "", &diff_opts);
660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030
	diffcore_std(&diff_opts);

	for (i = 0; i < diff_queued_diff.nr; i++) {
		struct diff_filepair *p = diff_queued_diff.queue[i];
		if ((p->status == 'R' || p->status == 'C') &&
		    !strcmp(p->two->path, origin->path)) {
			porigin = get_origin(parent, p->one->path);
			oidcpy(&porigin->blob_oid, &p->one->oid);
			porigin->mode = p->one->mode;
			break;
		}
	}
	diff_flush(&diff_opts);
	clear_pathspec(&diff_opts.pathspec);
	return porigin;
}

/*
 * Append a new blame entry to a given output queue.
 */
static void add_blame_entry(struct blame_entry ***queue,
			    const struct blame_entry *src)
{
	struct blame_entry *e = xmalloc(sizeof(*e));
	memcpy(e, src, sizeof(*e));
	blame_origin_incref(e->suspect);

	e->next = **queue;
	**queue = e;
	*queue = &e->next;
}

/*
 * src typically is on-stack; we want to copy the information in it to
 * a malloced blame_entry that gets added to the given queue.  The
 * origin of dst loses a refcnt.
 */
static void dup_entry(struct blame_entry ***queue,
		      struct blame_entry *dst, struct blame_entry *src)
{
	blame_origin_incref(src->suspect);
	blame_origin_decref(dst->suspect);
	memcpy(dst, src, sizeof(*src));
	dst->next = **queue;
	**queue = dst;
	*queue = &dst->next;
}

const char *blame_nth_line(struct blame_scoreboard *sb, long lno)
{
	return sb->final_buf + sb->lineno[lno];
}

/*
 * It is known that lines between tlno to same came from parent, and e
 * has an overlap with that range.  it also is known that parent's
 * line plno corresponds to e's line tlno.
 *
 *                <---- e ----->
 *                   <------>
 *                   <------------>
 *             <------------>
 *             <------------------>
 *
 * Split e into potentially three parts; before this chunk, the chunk
 * to be blamed for the parent, and after that portion.
 */
static void split_overlap(struct blame_entry *split,
			  struct blame_entry *e,
			  int tlno, int plno, int same,
			  struct blame_origin *parent)
{
	int chunk_end_lno;
	memset(split, 0, sizeof(struct blame_entry [3]));

	if (e->s_lno < tlno) {
		/* there is a pre-chunk part not blamed on parent */
		split[0].suspect = blame_origin_incref(e->suspect);
		split[0].lno = e->lno;
		split[0].s_lno = e->s_lno;
		split[0].num_lines = tlno - e->s_lno;
		split[1].lno = e->lno + tlno - e->s_lno;
		split[1].s_lno = plno;
	}
	else {
		split[1].lno = e->lno;
		split[1].s_lno = plno + (e->s_lno - tlno);
	}

	if (same < e->s_lno + e->num_lines) {
		/* there is a post-chunk part not blamed on parent */
		split[2].suspect = blame_origin_incref(e->suspect);
		split[2].lno = e->lno + (same - e->s_lno);
		split[2].s_lno = e->s_lno + (same - e->s_lno);
		split[2].num_lines = e->s_lno + e->num_lines - same;
		chunk_end_lno = split[2].lno;
	}
	else
		chunk_end_lno = e->lno + e->num_lines;
	split[1].num_lines = chunk_end_lno - split[1].lno;

	/*
	 * if it turns out there is nothing to blame the parent for,
	 * forget about the splitting.  !split[1].suspect signals this.
	 */
	if (split[1].num_lines < 1)
		return;
	split[1].suspect = blame_origin_incref(parent);
}

/*
 * split_overlap() divided an existing blame e into up to three parts
 * in split.  Any assigned blame is moved to queue to
 * reflect the split.
 */
static void split_blame(struct blame_entry ***blamed,
			struct blame_entry ***unblamed,
			struct blame_entry *split,
			struct blame_entry *e)
{
	if (split[0].suspect && split[2].suspect) {
		/* The first part (reuse storage for the existing entry e) */
		dup_entry(unblamed, e, &split[0]);

		/* The last part -- me */
		add_blame_entry(unblamed, &split[2]);

		/* ... and the middle part -- parent */
		add_blame_entry(blamed, &split[1]);
	}
	else if (!split[0].suspect && !split[2].suspect)
		/*
		 * The parent covers the entire area; reuse storage for
		 * e and replace it with the parent.
		 */
		dup_entry(blamed, e, &split[1]);
	else if (split[0].suspect) {
		/* me and then parent */
		dup_entry(unblamed, e, &split[0]);
		add_blame_entry(blamed, &split[1]);
	}
	else {
		/* parent and then me */
		dup_entry(blamed, e, &split[1]);
		add_blame_entry(unblamed, &split[2]);
	}
}

/*
 * After splitting the blame, the origins used by the
 * on-stack blame_entry should lose one refcnt each.
 */
static void decref_split(struct blame_entry *split)
{
	int i;

	for (i = 0; i < 3; i++)
		blame_origin_decref(split[i].suspect);
}

/*
 * reverse_blame reverses the list given in head, appending tail.
 * That allows us to build lists in reverse order, then reverse them
 * afterwards.  This can be faster than building the list in proper
 * order right away.  The reason is that building in proper order
 * requires writing a link in the _previous_ element, while building
 * in reverse order just requires placing the list head into the
 * _current_ element.
 */

static struct blame_entry *reverse_blame(struct blame_entry *head,
					 struct blame_entry *tail)
{
	while (head) {
		struct blame_entry *next = head->next;
		head->next = tail;
		tail = head;
		head = next;
	}
	return tail;
}

/*
 * Process one hunk from the patch between the current suspect for
 * blame_entry e and its parent.  This first blames any unfinished
 * entries before the chunk (which is where target and parent start
 * differing) on the parent, and then splits blame entries at the
 * start and at the end of the difference region.  Since use of -M and
 * -C options may lead to overlapping/duplicate source line number
 * ranges, all we can rely on from sorting/merging is the order of the
 * first suspect line number.
 */
static void blame_chunk(struct blame_entry ***dstq, struct blame_entry ***srcq,
			int tlno, int offset, int same,
			struct blame_origin *parent)
{
	struct blame_entry *e = **srcq;
	struct blame_entry *samep = NULL, *diffp = NULL;

	while (e && e->s_lno < tlno) {
		struct blame_entry *next = e->next;
		/*
		 * current record starts before differing portion.  If
		 * it reaches into it, we need to split it up and
		 * examine the second part separately.
		 */
		if (e->s_lno + e->num_lines > tlno) {
			/* Move second half to a new record */
			int len = tlno - e->s_lno;
			struct blame_entry *n = xcalloc(1, sizeof (struct blame_entry));
			n->suspect = e->suspect;
			n->lno = e->lno + len;
			n->s_lno = e->s_lno + len;
			n->num_lines = e->num_lines - len;
			e->num_lines = len;
			e->score = 0;
			/* Push new record to diffp */
			n->next = diffp;
			diffp = n;
		} else
			blame_origin_decref(e->suspect);
		/* Pass blame for everything before the differing
		 * chunk to the parent */
		e->suspect = blame_origin_incref(parent);
		e->s_lno += offset;
		e->next = samep;
		samep = e;
		e = next;
	}
	/*
	 * As we don't know how much of a common stretch after this
	 * diff will occur, the currently blamed parts are all that we
	 * can assign to the parent for now.
	 */

	if (samep) {
		**dstq = reverse_blame(samep, **dstq);
		*dstq = &samep->next;
	}
	/*
	 * Prepend the split off portions: everything after e starts
	 * after the blameable portion.
	 */
	e = reverse_blame(diffp, e);

	/*
	 * Now retain records on the target while parts are different
	 * from the parent.
	 */
	samep = NULL;
	diffp = NULL;
	while (e && e->s_lno < same) {
		struct blame_entry *next = e->next;

		/*
		 * If current record extends into sameness, need to split.
		 */
		if (e->s_lno + e->num_lines > same) {
			/*
			 * Move second half to a new record to be
			 * processed by later chunks
			 */
			int len = same - e->s_lno;
			struct blame_entry *n = xcalloc(1, sizeof (struct blame_entry));
			n->suspect = blame_origin_incref(e->suspect);
			n->lno = e->lno + len;
			n->s_lno = e->s_lno + len;
			n->num_lines = e->num_lines - len;
			e->num_lines = len;
			e->score = 0;
			/* Push new record to samep */
			n->next = samep;
			samep = n;
		}
		e->next = diffp;
		diffp = e;
		e = next;
	}
	**srcq = reverse_blame(diffp, reverse_blame(samep, e));
	/* Move across elements that are in the unblamable portion */
	if (diffp)
		*srcq = &diffp->next;
}

struct blame_chunk_cb_data {
	struct blame_origin *parent;
	long offset;
	struct blame_entry **dstq;
	struct blame_entry **srcq;
};

/* diff chunks are from parent to target */
static int blame_chunk_cb(long start_a, long count_a,
			  long start_b, long count_b, void *data)
{
	struct blame_chunk_cb_data *d = data;
	if (start_a - start_b != d->offset)
		die("internal error in blame::blame_chunk_cb");
	blame_chunk(&d->dstq, &d->srcq, start_b, start_a - start_b,
		    start_b + count_b, d->parent);
	d->offset = start_a + count_a - (start_b + count_b);
	return 0;
}

/*
 * We are looking at the origin 'target' and aiming to pass blame
 * for the lines it is suspected to its parent.  Run diff to find
 * which lines came from parent and pass blame for them.
 */
static void pass_blame_to_parent(struct blame_scoreboard *sb,
				 struct blame_origin *target,
				 struct blame_origin *parent)
{
	mmfile_t file_p, file_o;
	struct blame_chunk_cb_data d;
	struct blame_entry *newdest = NULL;

	if (!target->suspects)
		return; /* nothing remains for this target */

	d.parent = parent;
	d.offset = 0;
	d.dstq = &newdest; d.srcq = &target->suspects;

	fill_origin_blob(&sb->revs->diffopt, parent, &file_p, &sb->num_read_blob);
	fill_origin_blob(&sb->revs->diffopt, target, &file_o, &sb->num_read_blob);
	sb->num_get_patch++;

	if (diff_hunks(&file_p, &file_o, blame_chunk_cb, &d, sb->xdl_opts))
		die("unable to generate diff (%s -> %s)",
		    oid_to_hex(&parent->commit->object.oid),
		    oid_to_hex(&target->commit->object.oid));
	/* The rest are the same as the parent */
	blame_chunk(&d.dstq, &d.srcq, INT_MAX, d.offset, INT_MAX, parent);
	*d.dstq = NULL;
	queue_blames(sb, parent, newdest);

	return;
}

/*
 * The lines in blame_entry after splitting blames many times can become
 * very small and trivial, and at some point it becomes pointless to
 * blame the parents.  E.g. "\t\t}\n\t}\n\n" appears everywhere in any
 * ordinary C program, and it is not worth to say it was copied from
 * totally unrelated file in the parent.
 *
 * Compute how trivial the lines in the blame_entry are.
 */
unsigned blame_entry_score(struct blame_scoreboard *sb, struct blame_entry *e)
{
	unsigned score;
	const char *cp, *ep;

	if (e->score)
		return e->score;

	score = 1;
	cp = blame_nth_line(sb, e->lno);
	ep = blame_nth_line(sb, e->lno + e->num_lines);
	while (cp < ep) {
		unsigned ch = *((unsigned char *)cp);
		if (isalnum(ch))
			score++;
		cp++;
	}
	e->score = score;
	return score;
}

/*
1031 1032 1033
 * best_so_far[] and potential[] are both a split of an existing blame_entry
 * that passes blame to the parent.  Maintain best_so_far the best split so
 * far, by comparing potential and best_so_far and copying potential into
1034 1035 1036 1037
 * bst_so_far as needed.
 */
static void copy_split_if_better(struct blame_scoreboard *sb,
				 struct blame_entry *best_so_far,
1038
				 struct blame_entry *potential)
1039 1040 1041
{
	int i;

1042
	if (!potential[1].suspect)
1043 1044
		return;
	if (best_so_far[1].suspect) {
1045 1046
		if (blame_entry_score(sb, &potential[1]) <
		    blame_entry_score(sb, &best_so_far[1]))
1047 1048 1049 1050
			return;
	}

	for (i = 0; i < 3; i++)
1051
		blame_origin_incref(potential[i].suspect);
1052
	decref_split(best_so_far);
1053
	memcpy(best_so_far, potential, sizeof(struct blame_entry[3]));
1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079
}

/*
 * We are looking at a part of the final image represented by
 * ent (tlno and same are offset by ent->s_lno).
 * tlno is where we are looking at in the final image.
 * up to (but not including) same match preimage.
 * plno is where we are looking at in the preimage.
 *
 * <-------------- final image ---------------------->
 *       <------ent------>
 *         ^tlno ^same
 *    <---------preimage----->
 *         ^plno
 *
 * All line numbers are 0-based.
 */
static void handle_split(struct blame_scoreboard *sb,
			 struct blame_entry *ent,
			 int tlno, int plno, int same,
			 struct blame_origin *parent,
			 struct blame_entry *split)
{
	if (ent->num_lines <= tlno)
		return;
	if (tlno < same) {
1080
		struct blame_entry potential[3];
1081 1082
		tlno += ent->s_lno;
		same += ent->s_lno;
1083 1084 1085
		split_overlap(potential, ent, tlno, plno, same, parent);
		copy_split_if_better(sb, split, potential);
		decref_split(potential);
1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
	}
}

struct handle_split_cb_data {
	struct blame_scoreboard *sb;
	struct blame_entry *ent;
	struct blame_origin *parent;
	struct blame_entry *split;
	long plno;
	long tlno;
};

static int handle_split_cb(long start_a, long count_a,
			   long start_b, long count_b, void *data)
{
	struct handle_split_cb_data *d = data;
	handle_split(d->sb, d->ent, d->tlno, d->plno, start_b, d->parent,
		     d->split);
	d->plno = start_a + count_a;
	d->tlno = start_b + count_b;
	return 0;
}

/*
 * Find the lines from parent that are the same as ent so that
 * we can pass blames to it.  file_p has the blob contents for
 * the parent.
 */
static void find_copy_in_blob(struct blame_scoreboard *sb,
			      struct blame_entry *ent,
			      struct blame_origin *parent,
			      struct blame_entry *split,
			      mmfile_t *file_p)
{
	const char *cp;
	mmfile_t file_o;
	struct handle_split_cb_data d;

	memset(&d, 0, sizeof(d));
	d.sb = sb; d.ent = ent; d.parent = parent; d.split = split;
	/*
	 * Prepare mmfile that contains only the lines in ent.
	 */
	cp = blame_nth_line(sb, ent->lno);
	file_o.ptr = (char *) cp;
	file_o.size = blame_nth_line(sb, ent->lno + ent->num_lines) - cp;

	/*
	 * file_o is a part of final image we are annotating.
	 * file_p partially may match that image.
	 */
	memset(split, 0, sizeof(struct blame_entry [3]));
	if (diff_hunks(file_p, &file_o, handle_split_cb, &d, sb->xdl_opts))
		die("unable to generate diff (%s)",
		    oid_to_hex(&parent->commit->object.oid));
	/* remainder, if any, all match the preimage */
	handle_split(sb, ent, d.tlno, d.plno, ent->num_lines, parent, split);
}

/* Move all blame entries from list *source that have a score smaller
 * than score_min to the front of list *small.
 * Returns a pointer to the link pointing to the old head of the small list.
 */

static struct blame_entry **filter_small(struct blame_scoreboard *sb,
					 struct blame_entry **small,
					 struct blame_entry **source,
					 unsigned score_min)
{
	struct blame_entry *p = *source;
	struct blame_entry *oldsmall = *small;
	while (p) {
		if (blame_entry_score(sb, p) <= score_min) {
			*small = p;
			small = &p->next;
			p = *small;
		} else {
			*source = p;
			source = &p->next;
			p = *source;
		}
	}
	*small = oldsmall;
	*source = NULL;
	return small;
}

/*
 * See if lines currently target is suspected for can be attributed to
 * parent.
 */
static void find_move_in_parent(struct blame_scoreboard *sb,
				struct blame_entry ***blamed,
				struct blame_entry **toosmall,
				struct blame_origin *target,
				struct blame_origin *parent)
{
	struct blame_entry *e, split[3];
	struct blame_entry *unblamed = target->suspects;
	struct blame_entry *leftover = NULL;
	mmfile_t file_p;

	if (!unblamed)
		return; /* nothing remains for this target */

	fill_origin_blob(&sb->revs->diffopt, parent, &file_p, &sb->num_read_blob);
	if (!file_p.ptr)
		return;

	/* At each iteration, unblamed has a NULL-terminated list of
	 * entries that have not yet been tested for blame.  leftover
	 * contains the reversed list of entries that have been tested
	 * without being assignable to the parent.
	 */
	do {
		struct blame_entry **unblamedtail = &unblamed;
		struct blame_entry *next;
		for (e = unblamed; e; e = next) {
			next = e->next;
			find_copy_in_blob(sb, e, parent, split, &file_p);
			if (split[1].suspect &&
			    sb->move_score < blame_entry_score(sb, &split[1])) {
				split_blame(blamed, &unblamedtail, split, e);
			} else {
				e->next = leftover;
				leftover = e;
			}
			decref_split(split);
		}
		*unblamedtail = NULL;
		toosmall = filter_small(sb, toosmall, &unblamed, sb->move_score);
	} while (unblamed);
	target->suspects = reverse_blame(leftover, NULL);
}

struct blame_list {
	struct blame_entry *ent;
	struct blame_entry split[3];
};

/*
 * Count the number of entries the target is suspected for,
 * and prepare a list of entry and the best split.
 */
static struct blame_list *setup_blame_list(struct blame_entry *unblamed,
					   int *num_ents_p)
{
	struct blame_entry *e;
	int num_ents, i;
	struct blame_list *blame_list = NULL;

	for (e = unblamed, num_ents = 0; e; e = e->next)
		num_ents++;
	if (num_ents) {
		blame_list = xcalloc(num_ents, sizeof(struct blame_list));
		for (e = unblamed, i = 0; e; e = e->next)
			blame_list[i++].ent = e;
	}
	*num_ents_p = num_ents;
	return blame_list;
}

/*
 * For lines target is suspected for, see if we can find code movement
 * across file boundary from the parent commit.  porigin is the path
 * in the parent we already tried.
 */
static void find_copy_in_parent(struct blame_scoreboard *sb,
				struct blame_entry ***blamed,
				struct blame_entry **toosmall,
				struct blame_origin *target,
				struct commit *parent,
				struct blame_origin *porigin,
				int opt)
{
	struct diff_options diff_opts;
	int i, j;
	struct blame_list *blame_list;
	int num_ents;
	struct blame_entry *unblamed = target->suspects;
	struct blame_entry *leftover = NULL;

	if (!unblamed)
		return; /* nothing remains for this target */

1271
	repo_diff_setup(sb->repo, &diff_opts);
1272
	diff_opts.flags.recursive = 1;
1273 1274 1275 1276 1277 1278 1279
	diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;

	diff_setup_done(&diff_opts);

	/* Try "find copies harder" on new path if requested;
	 * we do not want to use diffcore_rename() actually to
	 * match things up; find_copies_harder is set only to
1280
	 * force diff_tree_oid() to feed all filepairs to diff_queue,
1281 1282 1283 1284 1285 1286
	 * and this code needs to be after diff_setup_done(), which
	 * usually makes find-copies-harder imply copy detection.
	 */
	if ((opt & PICKAXE_BLAME_COPY_HARDEST)
	    || ((opt & PICKAXE_BLAME_COPY_HARDER)
		&& (!porigin || strcmp(target->path, porigin->path))))
1287
		diff_opts.flags.find_copies_harder = 1;
1288 1289

	if (is_null_oid(&target->commit->object.oid))
1290
		do_diff_cache(get_commit_tree_oid(parent), &diff_opts);
1291
	else
1292 1293
		diff_tree_oid(get_commit_tree_oid(parent),
			      get_commit_tree_oid(target->commit),
1294
			      "", &diff_opts);
1295

1296
	if (!diff_opts.flags.find_copies_harder)
1297 1298 1299 1300 1301 1302 1303 1304 1305 1306
		diffcore_std(&diff_opts);

	do {
		struct blame_entry **unblamedtail = &unblamed;
		blame_list = setup_blame_list(unblamed, &num_ents);

		for (i = 0; i < diff_queued_diff.nr; i++) {
			struct diff_filepair *p = diff_queued_diff.queue[i];
			struct blame_origin *norigin;
			mmfile_t file_p;
1307
			struct blame_entry potential[3];
1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325

			if (!DIFF_FILE_VALID(p->one))
				continue; /* does not exist in parent */
			if (S_ISGITLINK(p->one->mode))
				continue; /* ignore git links */
			if (porigin && !strcmp(p->one->path, porigin->path))
				/* find_move already dealt with this path */
				continue;

			norigin = get_origin(parent, p->one->path);
			oidcpy(&norigin->blob_oid, &p->one->oid);
			norigin->mode = p->one->mode;
			fill_origin_blob(&sb->revs->diffopt, norigin, &file_p, &sb->num_read_blob);
			if (!file_p.ptr)
				continue;

			for (j = 0; j < num_ents; j++) {
				find_copy_in_blob(sb, blame_list[j].ent,
1326
						  norigin, potential, &file_p);
1327
				copy_split_if_better(sb, blame_list[j].split,
1328 1329
						     potential);
				decref_split(potential);
1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452
			}
			blame_origin_decref(norigin);
		}

		for (j = 0; j < num_ents; j++) {
			struct blame_entry *split = blame_list[j].split;
			if (split[1].suspect &&
			    sb->copy_score < blame_entry_score(sb, &split[1])) {
				split_blame(blamed, &unblamedtail, split,
					    blame_list[j].ent);
			} else {
				blame_list[j].ent->next = leftover;
				leftover = blame_list[j].ent;
			}
			decref_split(split);
		}
		free(blame_list);
		*unblamedtail = NULL;
		toosmall = filter_small(sb, toosmall, &unblamed, sb->copy_score);
	} while (unblamed);
	target->suspects = reverse_blame(leftover, NULL);
	diff_flush(&diff_opts);
	clear_pathspec(&diff_opts.pathspec);
}

/*
 * The blobs of origin and porigin exactly match, so everything
 * origin is suspected for can be blamed on the parent.
 */
static void pass_whole_blame(struct blame_scoreboard *sb,
			     struct blame_origin *origin, struct blame_origin *porigin)
{
	struct blame_entry *e, *suspects;

	if (!porigin->file.ptr && origin->file.ptr) {
		/* Steal its file */
		porigin->file = origin->file;
		origin->file.ptr = NULL;
	}
	suspects = origin->suspects;
	origin->suspects = NULL;
	for (e = suspects; e; e = e->next) {
		blame_origin_incref(porigin);
		blame_origin_decref(e->suspect);
		e->suspect = porigin;
	}
	queue_blames(sb, porigin, suspects);
}

/*
 * We pass blame from the current commit to its parents.  We keep saying
 * "parent" (and "porigin"), but what we mean is to find scapegoat to
 * exonerate ourselves.
 */
static struct commit_list *first_scapegoat(struct rev_info *revs, struct commit *commit,
					int reverse)
{
	if (!reverse) {
		if (revs->first_parent_only &&
		    commit->parents &&
		    commit->parents->next) {
			free_commit_list(commit->parents->next);
			commit->parents->next = NULL;
		}
		return commit->parents;
	}
	return lookup_decoration(&revs->children, &commit->object);
}

static int num_scapegoats(struct rev_info *revs, struct commit *commit, int reverse)
{
	struct commit_list *l = first_scapegoat(revs, commit, reverse);
	return commit_list_count(l);
}

/* Distribute collected unsorted blames to the respected sorted lists
 * in the various origins.
 */
static void distribute_blame(struct blame_scoreboard *sb, struct blame_entry *blamed)
{
	blamed = llist_mergesort(blamed, get_next_blame, set_next_blame,
				 compare_blame_suspect);
	while (blamed)
	{
		struct blame_origin *porigin = blamed->suspect;
		struct blame_entry *suspects = NULL;
		do {
			struct blame_entry *next = blamed->next;
			blamed->next = suspects;
			suspects = blamed;
			blamed = next;
		} while (blamed && blamed->suspect == porigin);
		suspects = reverse_blame(suspects, NULL);
		queue_blames(sb, porigin, suspects);
	}
}

#define MAXSG 16

static void pass_blame(struct blame_scoreboard *sb, struct blame_origin *origin, int opt)
{
	struct rev_info *revs = sb->revs;
	int i, pass, num_sg;
	struct commit *commit = origin->commit;
	struct commit_list *sg;
	struct blame_origin *sg_buf[MAXSG];
	struct blame_origin *porigin, **sg_origin = sg_buf;
	struct blame_entry *toosmall = NULL;
	struct blame_entry *blames, **blametail = &blames;

	num_sg = num_scapegoats(revs, commit, sb->reverse);
	if (!num_sg)
		goto finish;
	else if (num_sg < ARRAY_SIZE(sg_buf))
		memset(sg_buf, 0, sizeof(sg_buf));
	else
		sg_origin = xcalloc(num_sg, sizeof(*sg_origin));

	/*
	 * The first pass looks for unrenamed path to optimize for
	 * common cases, then we look for renames in the second pass.
	 */
	for (pass = 0; pass < 2 - sb->no_whole_file_rename; pass++) {
1453
		struct blame_origin *(*find)(struct repository *, struct commit *, struct blame_origin *);
1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465
		find = pass ? find_rename : find_origin;

		for (i = 0, sg = first_scapegoat(revs, commit, sb->reverse);
		     i < num_sg && sg;
		     sg = sg->next, i++) {
			struct commit *p = sg->item;
			int j, same;

			if (sg_origin[i])
				continue;
			if (parse_commit(p))
				continue;
1466
			porigin = find(sb->repo, p, origin);
1467 1468
			if (!porigin)
				continue;
1469
			if (oideq(&porigin->blob_oid, &origin->blob_oid)) {
1470 1471 1472 1473 1474 1475
				pass_whole_blame(sb, origin, porigin);
				blame_origin_decref(porigin);
				goto finish;
			}
			for (j = same = 0; j < i; j++)
				if (sg_origin[j] &&
1476
				    oideq(&sg_origin[j]->blob_oid, &porigin->blob_oid)) {
1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564
					same = 1;
					break;
				}
			if (!same)
				sg_origin[i] = porigin;
			else
				blame_origin_decref(porigin);
		}
	}

	sb->num_commits++;
	for (i = 0, sg = first_scapegoat(revs, commit, sb->reverse);
	     i < num_sg && sg;
	     sg = sg->next, i++) {
		struct blame_origin *porigin = sg_origin[i];
		if (!porigin)
			continue;
		if (!origin->previous) {
			blame_origin_incref(porigin);
			origin->previous = porigin;
		}
		pass_blame_to_parent(sb, origin, porigin);
		if (!origin->suspects)
			goto finish;
	}

	/*
	 * Optionally find moves in parents' files.
	 */
	if (opt & PICKAXE_BLAME_MOVE) {
		filter_small(sb, &toosmall, &origin->suspects, sb->move_score);
		if (origin->suspects) {
			for (i = 0, sg = first_scapegoat(revs, commit, sb->reverse);
			     i < num_sg && sg;
			     sg = sg->next, i++) {
				struct blame_origin *porigin = sg_origin[i];
				if (!porigin)
					continue;
				find_move_in_parent(sb, &blametail, &toosmall, origin, porigin);
				if (!origin->suspects)
					break;
			}
		}
	}

	/*
	 * Optionally find copies from parents' files.
	 */
	if (opt & PICKAXE_BLAME_COPY) {
		if (sb->copy_score > sb->move_score)
			filter_small(sb, &toosmall, &origin->suspects, sb->copy_score);
		else if (sb->copy_score < sb->move_score) {
			origin->suspects = blame_merge(origin->suspects, toosmall);
			toosmall = NULL;
			filter_small(sb, &toosmall, &origin->suspects, sb->copy_score);
		}
		if (!origin->suspects)
			goto finish;

		for (i = 0, sg = first_scapegoat(revs, commit, sb->reverse);
		     i < num_sg && sg;
		     sg = sg->next, i++) {
			struct blame_origin *porigin = sg_origin[i];
			find_copy_in_parent(sb, &blametail, &toosmall,
					    origin, sg->item, porigin, opt);
			if (!origin->suspects)
				goto finish;
		}
	}

finish:
	*blametail = NULL;
	distribute_blame(sb, blames);
	/*
	 * prepend toosmall to origin->suspects
	 *
	 * There is no point in sorting: this ends up on a big
	 * unsorted list in the caller anyway.
	 */
	if (toosmall) {
		struct blame_entry **tail = &toosmall;
		while (*tail)
			tail = &(*tail)->next;
		*tail = origin->suspects;
		origin->suspects = toosmall;
	}
	for (i = 0; i < num_sg; i++) {
		if (sg_origin[i]) {
1565 1566
			if (!sg_origin[i]->suspects)
				drop_origin_blob(sg_origin[i]);
1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585
			blame_origin_decref(sg_origin[i]);
		}
	}
	drop_origin_blob(origin);
	if (sg_buf != sg_origin)
		free(sg_origin);
}

/*
 * The main loop -- while we have blobs with lines whose true origin
 * is still unknown, pick one blob, and allow its lines to pass blames
 * to its parents. */
void assign_blame(struct blame_scoreboard *sb, int opt)
{
	struct rev_info *revs = sb->revs;
	struct commit *commit = prio_queue_get(&sb->commits);

	while (commit) {
		struct blame_entry *ent;
1586
		struct blame_origin *suspect = get_blame_suspects(commit);
1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641

		/* find one suspect to break down */
		while (suspect && !suspect->suspects)
			suspect = suspect->next;

		if (!suspect) {
			commit = prio_queue_get(&sb->commits);
			continue;
		}

		assert(commit == suspect->commit);

		/*
		 * We will use this suspect later in the loop,
		 * so hold onto it in the meantime.
		 */
		blame_origin_incref(suspect);
		parse_commit(commit);
		if (sb->reverse ||
		    (!(commit->object.flags & UNINTERESTING) &&
		     !(revs->max_age != -1 && commit->date < revs->max_age)))
			pass_blame(sb, suspect, opt);
		else {
			commit->object.flags |= UNINTERESTING;
			if (commit->object.parsed)
				mark_parents_uninteresting(commit);
		}
		/* treat root commit as boundary */
		if (!commit->parents && !sb->show_root)
			commit->object.flags |= UNINTERESTING;

		/* Take responsibility for the remaining entries */
		ent = suspect->suspects;
		if (ent) {
			suspect->guilty = 1;
			for (;;) {
				struct blame_entry *next = ent->next;
				if (sb->found_guilty_entry)
					sb->found_guilty_entry(ent, sb->found_guilty_entry_data);
				if (next) {
					ent = next;
					continue;
				}
				ent->next = sb->ent;
				sb->ent = suspect->suspects;
				suspect->suspects = NULL;
				break;
			}
		}
		blame_origin_decref(suspect);

		if (sb->debug) /* sanity */
			sanity_check_refcnt(sb);
	}
}
1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687

static const char *get_next_line(const char *start, const char *end)
{
	const char *nl = memchr(start, '\n', end - start);
	return nl ? nl + 1 : end;
}

/*
 * To allow quick access to the contents of nth line in the
 * final image, prepare an index in the scoreboard.
 */
static int prepare_lines(struct blame_scoreboard *sb)
{
	const char *buf = sb->final_buf;
	unsigned long len = sb->final_buf_size;
	const char *end = buf + len;
	const char *p;
	int *lineno;
	int num = 0;

	for (p = buf; p < end; p = get_next_line(p, end))
		num++;

	ALLOC_ARRAY(sb->lineno, num + 1);
	lineno = sb->lineno;

	for (p = buf; p < end; p = get_next_line(p, end))
		*lineno++ = p - buf;

	*lineno = len;

	sb->num_lines = num;
	return sb->num_lines;
}

static struct commit *find_single_final(struct rev_info *revs,
					const char **name_p)
{
	int i;
	struct commit *found = NULL;
	const char *name = NULL;

	for (i = 0; i < revs->pending.nr; i++) {
		struct object *obj = revs->pending.objects[i].item;
		if (obj->flags & UNINTERESTING)
			continue;
1688
		obj = deref_tag(revs->repo, obj, NULL, 0);
1689 1690 1691 1692 1693 1694 1695 1696 1697
		if (obj->type != OBJ_COMMIT)
			die("Non commit %s?", revs->pending.objects[i].name);
		if (found)
			die("More than one commit to dig from %s and %s?",
			    revs->pending.objects[i].name, name);
		found = (struct commit *)obj;
		name = revs->pending.objects[i].name;
	}
	if (name_p)
1698
		*name_p = xstrdup_or_null(name);
1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711
	return found;
}

static struct commit *dwim_reverse_initial(struct rev_info *revs,
					   const char **name_p)
{
	/*
	 * DWIM "git blame --reverse ONE -- PATH" as
	 * "git blame --reverse ONE..HEAD -- PATH" but only do so
	 * when it makes sense.
	 */
	struct object *obj;
	struct commit *head_commit;
1712
	struct object_id head_oid;
1713 1714 1715 1716 1717 1718

	if (revs->pending.nr != 1)
		return NULL;

	/* Is that sole rev a committish? */
	obj = revs->pending.objects[0].item;
1719
	obj = deref_tag(revs->repo, obj, NULL, 0);
1720 1721 1722 1723
	if (obj->type != OBJ_COMMIT)
		return NULL;

	/* Do we have HEAD? */
1724
	if (!resolve_ref_unsafe("HEAD", RESOLVE_REF_READING, &head_oid, NULL))
1725
		return NULL;
1726
	head_commit = lookup_commit_reference_gently(revs->repo,
1727
						     &head_oid, 1);
1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754
	if (!head_commit)
		return NULL;

	/* Turn "ONE" into "ONE..HEAD" then */
	obj->flags |= UNINTERESTING;
	add_pending_object(revs, &head_commit->object, "HEAD");

	if (name_p)
		*name_p = revs->pending.objects[0].name;
	return (struct commit *)obj;
}

static struct commit *find_single_initial(struct rev_info *revs,
					  const char **name_p)
{
	int i;
	struct commit *found = NULL;
	const char *name = NULL;

	/*
	 * There must be one and only one negative commit, and it must be
	 * the boundary.
	 */
	for (i = 0; i < revs->pending.nr; i++) {
		struct object *obj = revs->pending.objects[i].item;
		if (!(obj->flags & UNINTERESTING))
			continue;
1755
		obj = deref_tag(revs->repo, obj, NULL, 0);
1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770
		if (obj->type != OBJ_COMMIT)
			die("Non commit %s?", revs->pending.objects[i].name);
		if (found)
			die("More than one commit to dig up from, %s and %s?",
			    revs->pending.objects[i].name, name);
		found = (struct commit *) obj;
		name = revs->pending.objects[i].name;
	}

	if (!name)
		found = dwim_reverse_initial(revs, &name);
	if (!name)
		die("No commit to dig up from?");

	if (name_p)
1771
		*name_p = xstrdup(name);
1772 1773 1774 1775 1776 1777 1778 1779 1780 1781
	return found;
}

void init_scoreboard(struct blame_scoreboard *sb)
{
	memset(sb, 0, sizeof(struct blame_scoreboard));
	sb->move_score = BLAME_DEFAULT_MOVE_SCORE;
	sb->copy_score = BLAME_DEFAULT_COPY_SCORE;
}

1782 1783 1784
void setup_scoreboard(struct blame_scoreboard *sb,
		      const char *path,
		      struct blame_origin **orig)
1785 1786 1787 1788 1789 1790
{
	const char *final_commit_name = NULL;
	struct blame_origin *o;
	struct commit *final_commit = NULL;
	enum object_type type;

1791 1792
	init_blame_suspects(&blame_suspects);

1793 1794 1795
	if (sb->reverse && sb->contents_from)
		die(_("--contents and --reverse do not blend well."));

1796 1797 1798
	if (!sb->repo)
		BUG("repo is NULL");

1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819
	if (!sb->reverse) {
		sb->final = find_single_final(sb->revs, &final_commit_name);
		sb->commits.compare = compare_commits_by_commit_date;
	} else {
		sb->final = find_single_initial(sb->revs, &final_commit_name);
		sb->commits.compare = compare_commits_by_reverse_commit_date;
	}

	if (sb->final && sb->contents_from)
		die(_("cannot use --contents with final commit object name"));

	if (sb->reverse && sb->revs->first_parent_only)
		sb->revs->children.name = NULL;

	if (!sb->final) {
		/*
		 * "--not A B -- path" without anything positive;
		 * do not default to HEAD, but use the working tree
		 * or "--contents".
		 */
		setup_work_tree();
1820 1821
		sb->final = fake_working_tree_commit(sb->repo,
						     &sb->revs->diffopt,
1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844
						     path, sb->contents_from);
		add_pending_object(sb->revs, &(sb->final->object), ":");
	}

	if (sb->reverse && sb->revs->first_parent_only) {
		final_commit = find_single_final(sb->revs, NULL);
		if (!final_commit)
			die(_("--reverse and --first-parent together require specified latest commit"));
	}

	/*
	 * If we have bottom, this will mark the ancestors of the
	 * bottom commits we would reach while traversing as
	 * uninteresting.
	 */
	if (prepare_revision_walk(sb->revs))
		die(_("revision walk setup failed"));

	if (sb->reverse && sb->revs->first_parent_only) {
		struct commit *c = final_commit;

		sb->revs->children.name = "children";
		while (c->parents &&
1845
		       !oideq(&c->object.oid, &sb->final->object.oid)) {
1846 1847 1848 1849 1850
			struct commit_list *l = xcalloc(1, sizeof(*l));

			l->item = c;
			if (add_decoration(&sb->revs->children,
					   &c->parents->item->object, l))
1851
				BUG("not unique item in first-parent chain");
1852 1853 1854
			c = c->parents->item;
		}

1855
		if (!oideq(&c->object.oid, &sb->final->object.oid))
1856 1857 1858 1859
			die(_("--reverse --first-parent together require range along first-parent chain"));
	}

	if (is_null_oid(&sb->final->object.oid)) {
1860
		o = get_blame_suspects(sb->final);
1861 1862 1863 1864 1865
		sb->final_buf = xmemdupz(o->file.ptr, o->file.size);
		sb->final_buf_size = o->file.size;
	}
	else {
		o = get_origin(sb->final, path);
1866
		if (fill_blob_sha1_and_mode(sb->repo, o))
1867 1868
			die(_("no such path %s in %s"), path, final_commit_name);

1869
		if (sb->revs->diffopt.flags.allow_textconv &&
1870
		    textconv_object(sb->repo, path, o->mode, &o->blob_oid, 1, (char **) &sb->final_buf,
1871 1872 1873
				    &sb->final_buf_size))
			;
		else
1874 1875
			sb->final_buf = read_object_file(&o->blob_oid, &type,
							 &sb->final_buf_size);
1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886

		if (!sb->final_buf)
			die(_("cannot read blob %s for path %s"),
			    oid_to_hex(&o->blob_oid),
			    path);
	}
	sb->num_read_blob++;
	prepare_lines(sb);

	if (orig)
		*orig = o;
1887 1888

	free((char *)final_commit_name);
1889
}
1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905



struct blame_entry *blame_entry_prepend(struct blame_entry *head,
					long start, long end,
					struct blame_origin *o)
{
	struct blame_entry *new_head = xcalloc(1, sizeof(struct blame_entry));
	new_head->lno = start;
	new_head->num_lines = end - start;
	new_head->suspect = o;
	new_head->s_lno = start;
	new_head->next = head;
	blame_origin_incref(o);
	return new_head;
}