[thirdparty/git.git] / builtin / pack-redundant.c

/*
*
* Copyright 2005, Lukas Sandstrom <lukass@etek.chalmers.se>
*
* This file is licensed under the GPL v2.
*
*/

#include "builtin.h"

#define BLKSIZE 512

static const char pack_redundant_usage[] =
"git pack-redundant [--verbose] [--alt-odb] (--all | <filename.pack>...)";

static int load_all_packs, verbose, alt_odb;

struct llist_item {
	struct llist_item *next;
	const unsigned char *sha1;
};
static struct llist {
	struct llist_item *front;
	struct llist_item *back;
	size_t size;
} *all_objects; /* all objects which must be present in local packfiles */

static struct pack_list {
	struct pack_list *next;
	struct packed_git *pack;
	struct llist *unique_objects;
	struct llist *all_objects;
} *local_packs = NULL, *altodb_packs = NULL;

struct pll {
	struct pll *next;
	struct pack_list *pl;
};

static struct llist_item *free_nodes;

static inline void llist_item_put(struct llist_item *item)
{
	item->next = free_nodes;
	free_nodes = item;
}

static inline struct llist_item *llist_item_get(void)
{
	struct llist_item *new;
	if ( free_nodes ) {
		new = free_nodes;
		free_nodes = free_nodes->next;
	} else {
		int i = 1;
		ALLOC_ARRAY(new, BLKSIZE);
		for (; i < BLKSIZE; i++)
			llist_item_put(&new[i]);
	}
	return new;
}

static void llist_free(struct llist *list)
{
	while ((list->back = list->front)) {
		list->front = list->front->next;
		llist_item_put(list->back);
	}
	free(list);
}

static inline void llist_init(struct llist **list)
{
	*list = xmalloc(sizeof(struct llist));
	(*list)->front = (*list)->back = NULL;
	(*list)->size = 0;
}

static struct llist * llist_copy(struct llist *list)
{
	struct llist *ret;
	struct llist_item *new, *old, *prev;

	llist_init(&ret);

	if ((ret->size = list->size) == 0)
		return ret;

	new = ret->front = llist_item_get();
	new->sha1 = list->front->sha1;

	old = list->front->next;
	while (old) {
		prev = new;
		new = llist_item_get();
		prev->next = new;
		new->sha1 = old->sha1;
		old = old->next;
	}
	new->next = NULL;
	ret->back = new;

	return ret;
}

static inline struct llist_item *llist_insert(struct llist *list,
					      struct llist_item *after,
					       const unsigned char *sha1)
{
	struct llist_item *new = llist_item_get();
	new->sha1 = sha1;
	new->next = NULL;

	if (after != NULL) {
		new->next = after->next;
		after->next = new;
		if (after == list->back)
			list->back = new;
	} else {/* insert in front */
		if (list->size == 0)
			list->back = new;
		else
			new->next = list->front;
		list->front = new;
	}
	list->size++;
	return new;
}

static inline struct llist_item *llist_insert_back(struct llist *list,
						   const unsigned char *sha1)
{
	return llist_insert(list, list->back, sha1);
}

static inline struct llist_item *llist_insert_sorted_unique(struct llist *list,
			const unsigned char *sha1, struct llist_item *hint)
{
	struct llist_item *prev = NULL, *l;

	l = (hint == NULL) ? list->front : hint;
	while (l) {
		int cmp = hashcmp(l->sha1, sha1);
		if (cmp > 0) { /* we insert before this entry */
			return llist_insert(list, prev, sha1);
		}
		if (!cmp) { /* already exists */
			return l;
		}
		prev = l;
		l = l->next;
	}
	/* insert at the end */
	return llist_insert_back(list, sha1);
}

/* returns a pointer to an item in front of sha1 */
static inline struct llist_item * llist_sorted_remove(struct llist *list, const unsigned char *sha1, struct llist_item *hint)
{
	struct llist_item *prev, *l;

redo_from_start:
	l = (hint == NULL) ? list->front : hint;
	prev = NULL;
	while (l) {
		int cmp = hashcmp(l->sha1, sha1);
		if (cmp > 0) /* not in list, since sorted */
			return prev;
		if (!cmp) { /* found */
			if (prev == NULL) {
				if (hint != NULL && hint != list->front) {
					/* we don't know the previous element */
					hint = NULL;
					goto redo_from_start;
				}
				list->front = l->next;
			} else
				prev->next = l->next;
			if (l == list->back)
				list->back = prev;
			llist_item_put(l);
			list->size--;
			return prev;
		}
		prev = l;
		l = l->next;
	}
	return prev;
}

/* computes A\B */
static void llist_sorted_difference_inplace(struct llist *A,
				     struct llist *B)
{
	struct llist_item *hint, *b;

	hint = NULL;
	b = B->front;

	while (b) {
		hint = llist_sorted_remove(A, b->sha1, hint);
		b = b->next;
	}
}

static inline struct pack_list * pack_list_insert(struct pack_list **pl,
					   struct pack_list *entry)
{
	struct pack_list *p = xmalloc(sizeof(struct pack_list));
	memcpy(p, entry, sizeof(struct pack_list));
	p->next = *pl;
	*pl = p;
	return p;
}

static inline size_t pack_list_size(struct pack_list *pl)
{
	size_t ret = 0;
	while (pl) {
		ret++;
		pl = pl->next;
	}
	return ret;
}

static struct pack_list * pack_list_difference(const struct pack_list *A,
					       const struct pack_list *B)
{
	struct pack_list *ret;
	const struct pack_list *pl;

	if (A == NULL)
		return NULL;

	pl = B;
	while (pl != NULL) {
		if (A->pack == pl->pack)
			return pack_list_difference(A->next, B);
		pl = pl->next;
	}
	ret = xmalloc(sizeof(struct pack_list));
	memcpy(ret, A, sizeof(struct pack_list));
	ret->next = pack_list_difference(A->next, B);
	return ret;
}

static void cmp_two_packs(struct pack_list *p1, struct pack_list *p2)
{
	unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
	const unsigned char *p1_base, *p2_base;
	struct llist_item *p1_hint = NULL, *p2_hint = NULL;

	p1_base = p1->pack->index_data;
	p2_base = p2->pack->index_data;
	p1_base += 256 * 4 + ((p1->pack->index_version < 2) ? 4 : 8);
	p2_base += 256 * 4 + ((p2->pack->index_version < 2) ? 4 : 8);
	p1_step = (p1->pack->index_version < 2) ? 24 : 20;
	p2_step = (p2->pack->index_version < 2) ? 24 : 20;

	while (p1_off < p1->pack->num_objects * p1_step &&
	       p2_off < p2->pack->num_objects * p2_step)
	{
		int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
		/* cmp ~ p1 - p2 */
		if (cmp == 0) {
			p1_hint = llist_sorted_remove(p1->unique_objects,
					p1_base + p1_off, p1_hint);
			p2_hint = llist_sorted_remove(p2->unique_objects,
					p1_base + p1_off, p2_hint);
			p1_off += p1_step;
			p2_off += p2_step;
			continue;
		}
		if (cmp < 0) { /* p1 has the object, p2 doesn't */
			p1_off += p1_step;
		} else { /* p2 has the object, p1 doesn't */
			p2_off += p2_step;
		}
	}
}

static void pll_free(struct pll *l)
{
	struct pll *old;
	struct pack_list *opl;

	while (l) {
		old = l;
		while (l->pl) {
			opl = l->pl;
			l->pl = opl->next;
			free(opl);
		}
		l = l->next;
		free(old);
	}
}

/* all the permutations have to be free()d at the same time,
 * since they refer to each other
 */
static struct pll * get_permutations(struct pack_list *list, int n)
{
	struct pll *subset, *ret = NULL, *new_pll = NULL;

	if (list == NULL || pack_list_size(list) < n || n == 0)
		return NULL;

	if (n == 1) {
		while (list) {
			new_pll = xmalloc(sizeof(*new_pll));
			new_pll->pl = NULL;
			pack_list_insert(&new_pll->pl, list);
			new_pll->next = ret;
			ret = new_pll;
			list = list->next;
		}
		return ret;
	}

	while (list->next) {
		subset = get_permutations(list->next, n - 1);
		while (subset) {
			new_pll = xmalloc(sizeof(*new_pll));
			new_pll->pl = subset->pl;
			pack_list_insert(&new_pll->pl, list);
			new_pll->next = ret;
			ret = new_pll;
			subset = subset->next;
		}
		list = list->next;
	}
	return ret;
}

static int is_superset(struct pack_list *pl, struct llist *list)
{
	struct llist *diff;

	diff = llist_copy(list);

	while (pl) {
		llist_sorted_difference_inplace(diff, pl->all_objects);
		if (diff->size == 0) { /* we're done */
			llist_free(diff);
			return 1;
		}
		pl = pl->next;
	}
	llist_free(diff);
	return 0;
}

static size_t sizeof_union(struct packed_git *p1, struct packed_git *p2)
{
	size_t ret = 0;
	unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
	const unsigned char *p1_base, *p2_base;

	p1_base = p1->index_data;
	p2_base = p2->index_data;
	p1_base += 256 * 4 + ((p1->index_version < 2) ? 4 : 8);
	p2_base += 256 * 4 + ((p2->index_version < 2) ? 4 : 8);
	p1_step = (p1->index_version < 2) ? 24 : 20;
	p2_step = (p2->index_version < 2) ? 24 : 20;

	while (p1_off < p1->num_objects * p1_step &&
	       p2_off < p2->num_objects * p2_step)
	{
		int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
		/* cmp ~ p1 - p2 */
		if (cmp == 0) {
			ret++;
			p1_off += p1_step;
			p2_off += p2_step;
			continue;
		}
		if (cmp < 0) { /* p1 has the object, p2 doesn't */
			p1_off += p1_step;
		} else { /* p2 has the object, p1 doesn't */
			p2_off += p2_step;
		}
	}
	return ret;
}

/* another O(n^2) function ... */
static size_t get_pack_redundancy(struct pack_list *pl)
{
	struct pack_list *subset;
	size_t ret = 0;

	if (pl == NULL)
		return 0;

	while ((subset = pl->next)) {
		while (subset) {
			ret += sizeof_union(pl->pack, subset->pack);
			subset = subset->next;
		}
		pl = pl->next;
	}
	return ret;
}

static inline off_t pack_set_bytecount(struct pack_list *pl)
{
	off_t ret = 0;
	while (pl) {
		ret += pl->pack->pack_size;
		ret += pl->pack->index_size;
		pl = pl->next;
	}
	return ret;
}

static void minimize(struct pack_list **min)
{
	struct pack_list *pl, *unique = NULL,
		*non_unique = NULL, *min_perm = NULL;
	struct pll *perm, *perm_all, *perm_ok = NULL, *new_perm;
	struct llist *missing;
	off_t min_perm_size = 0, perm_size;
	int n;

	pl = local_packs;
	while (pl) {
		if (pl->unique_objects->size)
			pack_list_insert(&unique, pl);
		else
			pack_list_insert(&non_unique, pl);
		pl = pl->next;
	}
	/* find out which objects are missing from the set of unique packs */
	missing = llist_copy(all_objects);
	pl = unique;
	while (pl) {
		llist_sorted_difference_inplace(missing, pl->all_objects);
		pl = pl->next;
	}

	/* return if there are no objects missing from the unique set */
	if (missing->size == 0) {
		*min = unique;
		free(missing);
		return;
	}

	/* find the permutations which contain all missing objects */
	for (n = 1; n <= pack_list_size(non_unique) && !perm_ok; n++) {
		perm_all = perm = get_permutations(non_unique, n);
		while (perm) {
			if (is_superset(perm->pl, missing)) {
				new_perm = xmalloc(sizeof(struct pll));
				memcpy(new_perm, perm, sizeof(struct pll));
				new_perm->next = perm_ok;
				perm_ok = new_perm;
			}
			perm = perm->next;
		}
		if (perm_ok)
			break;
		pll_free(perm_all);
	}
	if (perm_ok == NULL)
		die("Internal error: No complete sets found!");

	/* find the permutation with the smallest size */
	perm = perm_ok;
	while (perm) {
		perm_size = pack_set_bytecount(perm->pl);
		if (!min_perm_size || min_perm_size > perm_size) {
			min_perm_size = perm_size;
			min_perm = perm->pl;
		}
		perm = perm->next;
	}
	*min = min_perm;
	/* add the unique packs to the list */
	pl = unique;
	while (pl) {
		pack_list_insert(min, pl);
		pl = pl->next;
	}
}

static void load_all_objects(void)
{
	struct pack_list *pl = local_packs;
	struct llist_item *hint, *l;

	llist_init(&all_objects);

	while (pl) {
		hint = NULL;
		l = pl->all_objects->front;
		while (l) {
			hint = llist_insert_sorted_unique(all_objects,
							  l->sha1, hint);
			l = l->next;
		}
		pl = pl->next;
	}
	/* remove objects present in remote packs */
	pl = altodb_packs;
	while (pl) {
		llist_sorted_difference_inplace(all_objects, pl->all_objects);
		pl = pl->next;
	}
}

/* this scales like O(n^2) */
static void cmp_local_packs(void)
{
	struct pack_list *subset, *pl = local_packs;

	while ((subset = pl)) {
		while ((subset = subset->next))
			cmp_two_packs(pl, subset);
		pl = pl->next;
	}
}

static void scan_alt_odb_packs(void)
{
	struct pack_list *local, *alt;

	alt = altodb_packs;
	while (alt) {
		local = local_packs;
		while (local) {
			llist_sorted_difference_inplace(local->unique_objects,
							alt->all_objects);
			local = local->next;
		}
		llist_sorted_difference_inplace(all_objects, alt->all_objects);
		alt = alt->next;
	}
}

static struct pack_list * add_pack(struct packed_git *p)
{
	struct pack_list l;
	unsigned long off = 0, step;
	const unsigned char *base;

	if (!p->pack_local && !(alt_odb || verbose))
		return NULL;

	l.pack = p;
	llist_init(&l.all_objects);

	if (open_pack_index(p))
		return NULL;

	base = p->index_data;
	base += 256 * 4 + ((p->index_version < 2) ? 4 : 8);
	step = (p->index_version < 2) ? 24 : 20;
	while (off < p->num_objects * step) {
		llist_insert_back(l.all_objects, base + off);
		off += step;
	}
	/* this list will be pruned in cmp_two_packs later */
	l.unique_objects = llist_copy(l.all_objects);
	if (p->pack_local)
		return pack_list_insert(&local_packs, &l);
	else
		return pack_list_insert(&altodb_packs, &l);
}

static struct pack_list * add_pack_file(const char *filename)
{
	struct packed_git *p = packed_git;

	if (strlen(filename) < 40)
		die("Bad pack filename: %s", filename);

	while (p) {
		if (strstr(p->pack_name, filename))
			return add_pack(p);
		p = p->next;
	}
	die("Filename %s not found in packed_git", filename);
}

static void load_all(void)
{
	struct packed_git *p = packed_git;

	while (p) {
		add_pack(p);
		p = p->next;
	}
}

int cmd_pack_redundant(int argc, const char **argv, const char *prefix)
{
	int i;
	struct pack_list *min, *red, *pl;
	struct llist *ignore;
	unsigned char *sha1;
	char buf[42]; /* 40 byte sha1 + \n + \0 */

	if (argc == 2 && !strcmp(argv[1], "-h"))
		usage(pack_redundant_usage);

	for (i = 1; i < argc; i++) {
		const char *arg = argv[i];
		if (!strcmp(arg, "--")) {
			i++;
			break;
		}
		if (!strcmp(arg, "--all")) {
			load_all_packs = 1;
			continue;
		}
		if (!strcmp(arg, "--verbose")) {
			verbose = 1;
			continue;
		}
		if (!strcmp(arg, "--alt-odb")) {
			alt_odb = 1;
			continue;
		}
		if (*arg == '-')
			usage(pack_redundant_usage);
		else
			break;
	}

	prepare_packed_git();

	if (load_all_packs)
		load_all();
	else
		while (*(argv + i) != NULL)
			add_pack_file(*(argv + i++));

	if (local_packs == NULL)
		die("Zero packs found!");

	load_all_objects();

	cmp_local_packs();
	if (alt_odb)
		scan_alt_odb_packs();

	/* ignore objects given on stdin */
	llist_init(&ignore);
	if (!isatty(0)) {
		while (fgets(buf, sizeof(buf), stdin)) {
			sha1 = xmalloc(20);
			if (get_sha1_hex(buf, sha1))
				die("Bad sha1 on stdin: %s", buf);
			llist_insert_sorted_unique(ignore, sha1, NULL);
		}
	}
	llist_sorted_difference_inplace(all_objects, ignore);
	pl = local_packs;
	while (pl) {
		llist_sorted_difference_inplace(pl->unique_objects, ignore);
		pl = pl->next;
	}

	minimize(&min);

	if (verbose) {
		fprintf(stderr, "There are %lu packs available in alt-odbs.\n",
			(unsigned long)pack_list_size(altodb_packs));
		fprintf(stderr, "The smallest (bytewise) set of packs is:\n");
		pl = min;
		while (pl) {
			fprintf(stderr, "\t%s\n", pl->pack->pack_name);
			pl = pl->next;
		}
		fprintf(stderr, "containing %lu duplicate objects "
				"with a total size of %lukb.\n",
			(unsigned long)get_pack_redundancy(min),
			(unsigned long)pack_set_bytecount(min)/1024);
		fprintf(stderr, "A total of %lu unique objects were considered.\n",
			(unsigned long)all_objects->size);
		fprintf(stderr, "Redundant packs (with indexes):\n");
	}
	pl = red = pack_list_difference(local_packs, min);
	while (pl) {
		printf("%s\n%s\n",
		       sha1_pack_index_name(pl->pack->sha1),
		       pl->pack->pack_name);
		pl = pl->next;
	}
	if (verbose)
		fprintf(stderr, "%luMB of redundant packs in total.\n",
			(unsigned long)pack_set_bytecount(red)/(1024*1024));

	return 0;
}
Commit	Line	Data
c283ab21 LS	1	/*
	2	*
	3	* Copyright 2005, Lukas Sandstrom <lukass@etek.chalmers.se>
	4	*
	5	* This file is licensed under the GPL v2.
	6	*
	7	*/
	8
c2e86add	9	#include "builtin.h"
c283ab21	10
6d016c9c LS	11	#define BLKSIZE 512
6d016c9c LS	12
9bc0f32c	13	static const char pack_redundant_usage[] =
9c9b4f2f	14	"git pack-redundant [--verbose] [--alt-odb] (--all \| <filename.pack>...)";
c283ab21	15
96f1e58f	16	static int load_all_packs, verbose, alt_odb;
c283ab21 LS	17
	18	struct llist_item {
	19	struct llist_item *next;
42873078	20	const unsigned char *sha1;
c283ab21	21	};
bd22c904	22	static struct llist {
c283ab21 LS	23	struct llist_item *front;
	24	struct llist_item *back;
	25	size_t size;
1c3039e8	26	} all_objects; / all objects which must be present in local packfiles */
c283ab21	27
bd22c904	28	static struct pack_list {
c283ab21 LS	29	struct pack_list *next;
	30	struct packed_git *pack;
	31	struct llist *unique_objects;
	32	struct llist *all_objects;
1c3039e8	33	} local_packs = NULL, altodb_packs = NULL;
c283ab21 LS	34
	35	struct pll {
	36	struct pll *next;
	37	struct pack_list *pl;
	38	};
	39
96f1e58f	40	static struct llist_item *free_nodes;
60435f68	41
6d016c9c LS	42	static inline void llist_item_put(struct llist_item *item)
	43	{
	44	item->next = free_nodes;
	45	free_nodes = item;
	46	}
	47
962554c6	48	static inline struct llist_item *llist_item_get(void)
60435f68 AR	49	{
	50	struct llist_item *new;
	51	if ( free_nodes ) {
	52	new = free_nodes;
	53	free_nodes = free_nodes->next;
6d016c9c LS	54	} else {
6d016c9c LS	55	int i = 1;
b32fa95f	56	ALLOC_ARRAY(new, BLKSIZE);
eeefa7c9	57	for (; i < BLKSIZE; i++)
6d016c9c	58	llist_item_put(&new[i]);
6d016c9c	59	}
60435f68 AR	60	return new;
	61	}
	62
60435f68	63	static void llist_free(struct llist *list)
c283ab21	64	{
eeefa7c9	65	while ((list->back = list->front)) {
c283ab21	66	list->front = list->front->next;
60435f68	67	llist_item_put(list->back);
c283ab21 LS	68	}
	69	free(list);
	70	}
	71
bd22c904	72	static inline void llist_init(struct llist **list)
c283ab21 LS	73	{
	74	*list = xmalloc(sizeof(struct llist));
	75	(list)->front = (list)->back = NULL;
	76	(*list)->size = 0;
	77	}
	78
bd22c904	79	static struct llist * llist_copy(struct llist *list)
c283ab21 LS	80	{
	81	struct llist *ret;
	82	struct llist_item new, old, *prev;
a6080a0a	83
c283ab21 LS	84	llist_init(&ret);
	85
	86	if ((ret->size = list->size) == 0)
	87	return ret;
	88
60435f68	89	new = ret->front = llist_item_get();
c283ab21 LS	90	new->sha1 = list->front->sha1;
	91
	92	old = list->front->next;
	93	while (old) {
	94	prev = new;
60435f68	95	new = llist_item_get();
c283ab21 LS	96	prev->next = new;
	97	new->sha1 = old->sha1;
	98	old = old->next;
	99	}
	100	new->next = NULL;
	101	ret->back = new;
a6080a0a	102
c283ab21 LS	103	return ret;
	104	}
	105
42873078 NP	106	static inline struct llist_item llist_insert(struct llist list,
	107	struct llist_item *after,
	108	const unsigned char *sha1)
c283ab21	109	{
60435f68	110	struct llist_item *new = llist_item_get();
c283ab21 LS	111	new->sha1 = sha1;
	112	new->next = NULL;
	113
	114	if (after != NULL) {
	115	new->next = after->next;
	116	after->next = new;
	117	if (after == list->back)
	118	list->back = new;
	119	} else {/* insert in front */
	120	if (list->size == 0)
	121	list->back = new;
	122	else
	123	new->next = list->front;
	124	list->front = new;
	125	}
	126	list->size++;
	127	return new;
	128	}
	129
42873078 NP	130	static inline struct llist_item llist_insert_back(struct llist list,
42873078 NP	131	const unsigned char *sha1)
c283ab21 LS	132	{
	133	return llist_insert(list, list->back, sha1);
	134	}
	135
42873078 NP	136	static inline struct llist_item llist_insert_sorted_unique(struct llist list,
42873078 NP	137	const unsigned char sha1, struct llist_item hint)
c283ab21 LS	138	{
	139	struct llist_item prev = NULL, l;
	140
	141	l = (hint == NULL) ? list->front : hint;
	142	while (l) {
a89fccd2	143	int cmp = hashcmp(l->sha1, sha1);
c283ab21 LS	144	if (cmp > 0) { /* we insert before this entry */
	145	return llist_insert(list, prev, sha1);
	146	}
eeefa7c9	147	if (!cmp) { /* already exists */
c283ab21 LS	148	return l;
	149	}
	150	prev = l;
	151	l = l->next;
	152	}
	153	/* insert at the end */
	154	return llist_insert_back(list, sha1);
	155	}
	156
c283ab21	157	/* returns a pointer to an item in front of sha1 */
b99a394c	158	static inline struct llist_item * llist_sorted_remove(struct llist list, const unsigned char sha1, struct llist_item *hint)
c283ab21 LS	159	{
	160	struct llist_item prev, l;
	161
	162	redo_from_start:
	163	l = (hint == NULL) ? list->front : hint;
	164	prev = NULL;
	165	while (l) {
a89fccd2	166	int cmp = hashcmp(l->sha1, sha1);
c283ab21 LS	167	if (cmp > 0) /* not in list, since sorted */
c283ab21 LS	168	return prev;
eeefa7c9	169	if (!cmp) { /* found */
c283ab21 LS	170	if (prev == NULL) {
	171	if (hint != NULL && hint != list->front) {
	172	/* we don't know the previous element */
	173	hint = NULL;
	174	goto redo_from_start;
	175	}
	176	list->front = l->next;
	177	} else
	178	prev->next = l->next;
	179	if (l == list->back)
	180	list->back = prev;
60435f68	181	llist_item_put(l);
c283ab21 LS	182	list->size--;
	183	return prev;
	184	}
	185	prev = l;
	186	l = l->next;
	187	}
	188	return prev;
	189	}
	190
1a41e743	191	/* computes A\B */
bd22c904	192	static void llist_sorted_difference_inplace(struct llist *A,
1a41e743 LS	193	struct llist *B)
	194	{
	195	struct llist_item hint, b;
	196
	197	hint = NULL;
	198	b = B->front;
	199
	200	while (b) {
	201	hint = llist_sorted_remove(A, b->sha1, hint);
	202	b = b->next;
	203	}
	204	}
	205
bd22c904	206	static inline struct pack_list * pack_list_insert(struct pack_list **pl,
c283ab21 LS	207	struct pack_list *entry)
	208	{
	209	struct pack_list *p = xmalloc(sizeof(struct pack_list));
	210	memcpy(p, entry, sizeof(struct pack_list));
	211	p->next = *pl;
	212	*pl = p;
	213	return p;
	214	}
	215
bd22c904	216	static inline size_t pack_list_size(struct pack_list *pl)
1c3039e8 LS	217	{
1c3039e8 LS	218	size_t ret = 0;
eeefa7c9	219	while (pl) {
1c3039e8 LS	220	ret++;
	221	pl = pl->next;
	222	}
	223	return ret;
	224	}
	225
d1ab1577 AR	226	static struct pack_list * pack_list_difference(const struct pack_list *A,
d1ab1577 AR	227	const struct pack_list *B)
c283ab21	228	{
d1ab1577 AR	229	struct pack_list *ret;
d1ab1577 AR	230	const struct pack_list *pl;
c283ab21 LS	231
	232	if (A == NULL)
	233	return NULL;
	234
	235	pl = B;
	236	while (pl != NULL) {
	237	if (A->pack == pl->pack)
	238	return pack_list_difference(A->next, B);
	239	pl = pl->next;
	240	}
	241	ret = xmalloc(sizeof(struct pack_list));
	242	memcpy(ret, A, sizeof(struct pack_list));
	243	ret->next = pack_list_difference(A->next, B);
	244	return ret;
	245	}
	246
bd22c904	247	static void cmp_two_packs(struct pack_list p1, struct pack_list p2)
c283ab21	248	{
8c681e07	249	unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
42873078	250	const unsigned char p1_base, p2_base;
c283ab21	251	struct llist_item p1_hint = NULL, p2_hint = NULL;
1d7f171c	252
42873078 NP	253	p1_base = p1->pack->index_data;
42873078 NP	254	p2_base = p2->pack->index_data;
8c681e07 NP	255	p1_base += 256 * 4 + ((p1->pack->index_version < 2) ? 4 : 8);
	256	p2_base += 256 * 4 + ((p2->pack->index_version < 2) ? 4 : 8);
	257	p1_step = (p1->pack->index_version < 2) ? 24 : 20;
	258	p2_step = (p2->pack->index_version < 2) ? 24 : 20;
c283ab21	259
8c681e07 NP	260	while (p1_off < p1->pack->num_objects * p1_step &&
8c681e07 NP	261	p2_off < p2->pack->num_objects * p2_step)
c283ab21	262	{
a89fccd2	263	int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
c283ab21 LS	264	/* cmp ~ p1 - p2 */
	265	if (cmp == 0) {
	266	p1_hint = llist_sorted_remove(p1->unique_objects,
	267	p1_base + p1_off, p1_hint);
	268	p2_hint = llist_sorted_remove(p2->unique_objects,
	269	p1_base + p1_off, p2_hint);
8c681e07 NP	270	p1_off += p1_step;
8c681e07 NP	271	p2_off += p2_step;
c283ab21 LS	272	continue;
	273	}
	274	if (cmp < 0) { /* p1 has the object, p2 doesn't */
8c681e07	275	p1_off += p1_step;
c283ab21	276	} else { /* p2 has the object, p1 doesn't */
8c681e07	277	p2_off += p2_step;
c283ab21 LS	278	}
	279	}
	280	}
	281
962554c6	282	static void pll_free(struct pll *l)
751a71e2	283	{
6d016c9c LS	284	struct pll *old;
	285	struct pack_list *opl;
	286
	287	while (l) {
	288	old = l;
	289	while (l->pl) {
	290	opl = l->pl;
	291	l->pl = opl->next;
	292	free(opl);
751a71e2	293	}
6d016c9c LS	294	l = l->next;
6d016c9c LS	295	free(old);
751a71e2	296	}
751a71e2 LS	297	}
751a71e2 LS	298
c283ab21 LS	299	/* all the permutations have to be free()d at the same time,
	300	* since they refer to each other
	301	*/
6d016c9c	302	static struct pll * get_permutations(struct pack_list *list, int n)
c283ab21	303	{
bdd478d6	304	struct pll subset, ret = NULL, *new_pll = NULL;
6d016c9c LS	305
6d016c9c LS	306	if (list == NULL \|\| pack_list_size(list) < n \|\| n == 0)
c283ab21 LS	307	return NULL;
c283ab21 LS	308
6d016c9c LS	309	if (n == 1) {
6d016c9c LS	310	while (list) {
bdd478d6	311	new_pll = xmalloc(sizeof(*new_pll));
6d016c9c LS	312	new_pll->pl = NULL;
	313	pack_list_insert(&new_pll->pl, list);
	314	new_pll->next = ret;
	315	ret = new_pll;
	316	list = list->next;
	317	}
	318	return ret;
c283ab21 LS	319	}
c283ab21 LS	320
6d016c9c LS	321	while (list->next) {
	322	subset = get_permutations(list->next, n - 1);
	323	while (subset) {
bdd478d6	324	new_pll = xmalloc(sizeof(*new_pll));
6d016c9c LS	325	new_pll->pl = subset->pl;
	326	pack_list_insert(&new_pll->pl, list);
	327	new_pll->next = ret;
	328	ret = new_pll;
	329	subset = subset->next;
751a71e2	330	}
6d016c9c	331	list = list->next;
c283ab21	332	}
6d016c9c	333	return ret;
c283ab21 LS	334	}
c283ab21 LS	335
bd22c904	336	static int is_superset(struct pack_list pl, struct llist list)
c283ab21	337	{
1c3039e8	338	struct llist *diff;
c283ab21 LS	339
	340	diff = llist_copy(list);
	341
	342	while (pl) {
d1ab1577	343	llist_sorted_difference_inplace(diff, pl->all_objects);
c283ab21 LS	344	if (diff->size == 0) { /* we're done */
	345	llist_free(diff);
	346	return 1;
	347	}
	348	pl = pl->next;
	349	}
	350	llist_free(diff);
	351	return 0;
	352	}
	353
bd22c904	354	static size_t sizeof_union(struct packed_git p1, struct packed_git p2)
c283ab21 LS	355	{
c283ab21 LS	356	size_t ret = 0;
8c681e07	357	unsigned long p1_off = 0, p2_off = 0, p1_step, p2_step;
42873078	358	const unsigned char p1_base, p2_base;
c283ab21	359
42873078 NP	360	p1_base = p1->index_data;
42873078 NP	361	p2_base = p2->index_data;
8c681e07 NP	362	p1_base += 256 * 4 + ((p1->index_version < 2) ? 4 : 8);
	363	p2_base += 256 * 4 + ((p2->index_version < 2) ? 4 : 8);
	364	p1_step = (p1->index_version < 2) ? 24 : 20;
	365	p2_step = (p2->index_version < 2) ? 24 : 20;
c283ab21	366
8c681e07 NP	367	while (p1_off < p1->num_objects * p1_step &&
8c681e07 NP	368	p2_off < p2->num_objects * p2_step)
c283ab21	369	{
a89fccd2	370	int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
c283ab21 LS	371	/* cmp ~ p1 - p2 */
	372	if (cmp == 0) {
	373	ret++;
8c681e07 NP	374	p1_off += p1_step;
8c681e07 NP	375	p2_off += p2_step;
c283ab21 LS	376	continue;
	377	}
	378	if (cmp < 0) { /* p1 has the object, p2 doesn't */
8c681e07	379	p1_off += p1_step;
c283ab21	380	} else { /* p2 has the object, p1 doesn't */
8c681e07	381	p2_off += p2_step;
c283ab21 LS	382	}
	383	}
	384	return ret;
	385	}
	386
	387	/* another O(n^2) function ... */
bd22c904	388	static size_t get_pack_redundancy(struct pack_list *pl)
c283ab21 LS	389	{
c283ab21 LS	390	struct pack_list *subset;
bd22c904	391	size_t ret = 0;
1c3039e8 LS	392
	393	if (pl == NULL)
	394	return 0;
	395
c283ab21	396	while ((subset = pl->next)) {
eeefa7c9	397	while (subset) {
c283ab21 LS	398	ret += sizeof_union(pl->pack, subset->pack);
	399	subset = subset->next;
	400	}
	401	pl = pl->next;
	402	}
	403	return ret;
	404	}
	405
c4001d92	406	static inline off_t pack_set_bytecount(struct pack_list *pl)
c283ab21	407	{
c4001d92	408	off_t ret = 0;
c283ab21 LS	409	while (pl) {
	410	ret += pl->pack->pack_size;
	411	ret += pl->pack->index_size;
	412	pl = pl->next;
	413	}
	414	return ret;
	415	}
	416
bd22c904	417	static void minimize(struct pack_list **min)
c283ab21 LS	418	{
	419	struct pack_list pl, unique = NULL,
	420	non_unique = NULL, min_perm = NULL;
	421	struct pll perm, perm_all, perm_ok = NULL, new_perm;
1c3039e8	422	struct llist *missing;
c4001d92	423	off_t min_perm_size = 0, perm_size;
6d016c9c	424	int n;
c283ab21	425
1c3039e8	426	pl = local_packs;
c283ab21	427	while (pl) {
eeefa7c9	428	if (pl->unique_objects->size)
c283ab21 LS	429	pack_list_insert(&unique, pl);
	430	else
	431	pack_list_insert(&non_unique, pl);
	432	pl = pl->next;
	433	}
	434	/* find out which objects are missing from the set of unique packs */
	435	missing = llist_copy(all_objects);
	436	pl = unique;
	437	while (pl) {
6d016c9c	438	llist_sorted_difference_inplace(missing, pl->all_objects);
c283ab21 LS	439	pl = pl->next;
	440	}
	441
1c3039e8	442	/* return if there are no objects missing from the unique set */
c283ab21 LS	443	if (missing->size == 0) {
c283ab21 LS	444	*min = unique;
077a34e0	445	free(missing);
c283ab21 LS	446	return;
	447	}
	448
	449	/* find the permutations which contain all missing objects */
6d016c9c LS	450	for (n = 1; n <= pack_list_size(non_unique) && !perm_ok; n++) {
	451	perm_all = perm = get_permutations(non_unique, n);
	452	while (perm) {
	453	if (is_superset(perm->pl, missing)) {
	454	new_perm = xmalloc(sizeof(struct pll));
	455	memcpy(new_perm, perm, sizeof(struct pll));
	456	new_perm->next = perm_ok;
	457	perm_ok = new_perm;
	458	}
	459	perm = perm->next;
c283ab21	460	}
6d016c9c LS	461	if (perm_ok)
	462	break;
	463	pll_free(perm_all);
c283ab21	464	}
c283ab21	465	if (perm_ok == NULL)
d7530708	466	die("Internal error: No complete sets found!");
c283ab21 LS	467
	468	/* find the permutation with the smallest size */
	469	perm = perm_ok;
	470	while (perm) {
	471	perm_size = pack_set_bytecount(perm->pl);
c4001d92	472	if (!min_perm_size \|\| min_perm_size > perm_size) {
c283ab21 LS	473	min_perm_size = perm_size;
	474	min_perm = perm->pl;
	475	}
	476	perm = perm->next;
	477	}
	478	*min = min_perm;
	479	/* add the unique packs to the list */
	480	pl = unique;
eeefa7c9	481	while (pl) {
c283ab21 LS	482	pack_list_insert(min, pl);
	483	pl = pl->next;
	484	}
	485	}
	486
bd22c904	487	static void load_all_objects(void)
c283ab21	488	{
1c3039e8	489	struct pack_list *pl = local_packs;
c283ab21	490	struct llist_item hint, l;
c283ab21 LS	491
	492	llist_init(&all_objects);
	493
	494	while (pl) {
c283ab21 LS	495	hint = NULL;
	496	l = pl->all_objects->front;
	497	while (l) {
	498	hint = llist_insert_sorted_unique(all_objects,
	499	l->sha1, hint);
	500	l = l->next;
	501	}
	502	pl = pl->next;
	503	}
1c3039e8 LS	504	/* remove objects present in remote packs */
	505	pl = altodb_packs;
	506	while (pl) {
	507	llist_sorted_difference_inplace(all_objects, pl->all_objects);
	508	pl = pl->next;
	509	}
c283ab21 LS	510	}
	511
	512	/* this scales like O(n^2) */
bd22c904	513	static void cmp_local_packs(void)
c283ab21	514	{
1c3039e8	515	struct pack_list subset, pl = local_packs;
c283ab21	516
1c3039e8	517	while ((subset = pl)) {
eeefa7c9	518	while ((subset = subset->next))
1c3039e8 LS	519	cmp_two_packs(pl, subset);
	520	pl = pl->next;
	521	}
06a45c8c	522	}
1c3039e8	523
bd22c904	524	static void scan_alt_odb_packs(void)
06a45c8c LS	525	{
	526	struct pack_list local, alt;
	527
	528	alt = altodb_packs;
	529	while (alt) {
	530	local = local_packs;
	531	while (local) {
	532	llist_sorted_difference_inplace(local->unique_objects,
	533	alt->all_objects);
	534	local = local->next;
1c3039e8	535	}
6d016c9c	536	llist_sorted_difference_inplace(all_objects, alt->all_objects);
06a45c8c	537	alt = alt->next;
c283ab21 LS	538	}
	539	}
	540
bd22c904	541	static struct pack_list * add_pack(struct packed_git *p)
c283ab21 LS	542	{
c283ab21 LS	543	struct pack_list l;
8c681e07	544	unsigned long off = 0, step;
42873078	545	const unsigned char *base;
c283ab21	546
06a45c8c LS	547	if (!p->pack_local && !(alt_odb \|\| verbose))
	548	return NULL;
	549
c283ab21 LS	550	l.pack = p;
	551	llist_init(&l.all_objects);
	552
eaa86770	553	if (open_pack_index(p))
d079837e SP	554	return NULL;
d079837e SP	555
42873078	556	base = p->index_data;
8c681e07 NP	557	base += 256 * 4 + ((p->index_version < 2) ? 4 : 8);
	558	step = (p->index_version < 2) ? 24 : 20;
	559	while (off < p->num_objects * step) {
c283ab21	560	llist_insert_back(l.all_objects, base + off);
8c681e07	561	off += step;
c283ab21 LS	562	}
	563	/* this list will be pruned in cmp_two_packs later */
	564	l.unique_objects = llist_copy(l.all_objects);
1c3039e8 LS	565	if (p->pack_local)
	566	return pack_list_insert(&local_packs, &l);
	567	else
06a45c8c	568	return pack_list_insert(&altodb_packs, &l);
c283ab21 LS	569	}
c283ab21 LS	570
377d0276	571	static struct pack_list * add_pack_file(const char *filename)
c283ab21 LS	572	{
	573	struct packed_git *p = packed_git;
	574
	575	if (strlen(filename) < 40)
d7530708	576	die("Bad pack filename: %s", filename);
c283ab21 LS	577
	578	while (p) {
	579	if (strstr(p->pack_name, filename))
c283ab21 LS	580	return add_pack(p);
	581	p = p->next;
	582	}
d7530708	583	die("Filename %s not found in packed_git", filename);
c283ab21 LS	584	}
c283ab21 LS	585
bd22c904	586	static void load_all(void)
c283ab21 LS	587	{
	588	struct packed_git *p = packed_git;
	589
	590	while (p) {
1c3039e8	591	add_pack(p);
c283ab21 LS	592	p = p->next;
	593	}
	594	}
	595
377d0276	596	int cmd_pack_redundant(int argc, const char *argv, const char prefix)
c283ab21 LS	597	{
	598	int i;
	599	struct pack_list min, red, *pl;
bb931cf9	600	struct llist *ignore;
b99a394c JH	601	unsigned char *sha1;
b99a394c JH	602	char buf[42]; /* 40 byte sha1 + \n + \0 */
c283ab21	603
99caeed0 JN	604	if (argc == 2 && !strcmp(argv[1], "-h"))
	605	usage(pack_redundant_usage);
	606
c283ab21 LS	607	for (i = 1; i < argc; i++) {
c283ab21 LS	608	const char *arg = argv[i];
eeefa7c9	609	if (!strcmp(arg, "--")) {
9bc0f32c	610	i++;
c283ab21	611	break;
9bc0f32c	612	}
eeefa7c9	613	if (!strcmp(arg, "--all")) {
1c3039e8	614	load_all_packs = 1;
c283ab21 LS	615	continue;
c283ab21 LS	616	}
eeefa7c9	617	if (!strcmp(arg, "--verbose")) {
c283ab21 LS	618	verbose = 1;
	619	continue;
	620	}
eeefa7c9	621	if (!strcmp(arg, "--alt-odb")) {
1c3039e8 LS	622	alt_odb = 1;
	623	continue;
	624	}
eeefa7c9	625	if (*arg == '-')
9bc0f32c	626	usage(pack_redundant_usage);
c283ab21 LS	627	else
	628	break;
	629	}
	630
	631	prepare_packed_git();
	632
1c3039e8	633	if (load_all_packs)
c283ab21 LS	634	load_all();
	635	else
	636	while (*(argv + i) != NULL)
	637	add_pack_file(*(argv + i++));
	638
1c3039e8	639	if (local_packs == NULL)
d7530708	640	die("Zero packs found!");
c283ab21	641
c283ab21 LS	642	load_all_objects();
c283ab21 LS	643
06a45c8c LS	644	cmp_local_packs();
	645	if (alt_odb)
	646	scan_alt_odb_packs();
	647
bb931cf9 LS	648	/* ignore objects given on stdin */
	649	llist_init(&ignore);
	650	if (!isatty(0)) {
	651	while (fgets(buf, sizeof(buf), stdin)) {
	652	sha1 = xmalloc(20);
	653	if (get_sha1_hex(buf, sha1))
	654	die("Bad sha1 on stdin: %s", buf);
	655	llist_insert_sorted_unique(ignore, sha1, NULL);
	656	}
	657	}
	658	llist_sorted_difference_inplace(all_objects, ignore);
	659	pl = local_packs;
	660	while (pl) {
	661	llist_sorted_difference_inplace(pl->unique_objects, ignore);
	662	pl = pl->next;
	663	}
	664
c283ab21	665	minimize(&min);
06a45c8c	666
c283ab21	667	if (verbose) {
abacbe41 KR	668	fprintf(stderr, "There are %lu packs available in alt-odbs.\n",
abacbe41 KR	669	(unsigned long)pack_list_size(altodb_packs));
c283ab21 LS	670	fprintf(stderr, "The smallest (bytewise) set of packs is:\n");
	671	pl = min;
	672	while (pl) {
	673	fprintf(stderr, "\t%s\n", pl->pack->pack_name);
	674	pl = pl->next;
	675	}
abacbe41 KR	676	fprintf(stderr, "containing %lu duplicate objects "
	677	"with a total size of %lukb.\n",
	678	(unsigned long)get_pack_redundancy(min),
	679	(unsigned long)pack_set_bytecount(min)/1024);
	680	fprintf(stderr, "A total of %lu unique objects were considered.\n",
	681	(unsigned long)all_objects->size);
c283ab21 LS	682	fprintf(stderr, "Redundant packs (with indexes):\n");
c283ab21 LS	683	}
1c3039e8	684	pl = red = pack_list_difference(local_packs, min);
c283ab21 LS	685	while (pl) {
c283ab21 LS	686	printf("%s\n%s\n",
1c3039e8 LS	687	sha1_pack_index_name(pl->pack->sha1),
1c3039e8 LS	688	pl->pack->pack_name);
c283ab21 LS	689	pl = pl->next;
c283ab21 LS	690	}
bb931cf9	691	if (verbose)
b99a394c JH	692	fprintf(stderr, "%luMB of redundant packs in total.\n",
b99a394c JH	693	(unsigned long)pack_set_bytecount(red)/(1024*1024));
c283ab21 LS	694
	695	return 0;
	696	}