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

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

#include "cache.h"

#define BLKSIZE 512

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

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;
		new = xmalloc(sizeof(struct llist_item) * 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)
{
	int p1_off, p2_off;
	const unsigned char *p1_base, *p2_base;
	struct llist_item *p1_hint = NULL, *p2_hint = NULL;

	p1_off = p2_off = 256 * 4 + 4;
	p1_base = p1->pack->index_data;
	p2_base = p2->pack->index_data;

	while (p1_off <= p1->pack->index_size - 3 * 20 &&
	       p2_off <= p2->pack->index_size - 3 * 20)
	{
		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+=24;
			p2_off+=24;
			continue;
		}
		if (cmp < 0) { /* p1 has the object, p2 doesn't */
			p1_off+=24;
		} else { /* p2 has the object, p1 doesn't */
			p2_off+=24;
		}
	}
}

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, *pll;

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

	if (n == 1) {
		while (list) {
			new_pll = xmalloc(sizeof(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(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;
	int p1_off, p2_off;
	const unsigned char *p1_base, *p2_base;

	p1_off = p2_off = 256 * 4 + 4;
	p1_base = p1->index_data;
	p2_base = p2->index_data;

	while (p1_off <= p1->index_size - 3 * 20 &&
	       p2_off <= p2->index_size - 3 * 20)
	{
		int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
		/* cmp ~ p1 - p2 */
		if (cmp == 0) {
			ret++;
			p1_off+=24;
			p2_off+=24;
			continue;
		}
		if (cmp < 0) { /* p1 has the object, p2 doesn't */
			p1_off+=24;
		} else { /* p2 has the object, p1 doesn't */
			p2_off+=24;
		}
	}
	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;
		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!\n");

	/* 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;
	size_t off;
	const unsigned char *base;

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

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

	off = 256 * 4 + 4;
	base = p->index_data;
	while (off <= p->index_size - 3 * 20) {
		llist_insert_back(l.all_objects, base + off);
		off += 24;
	}
	/* 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(char *filename)
{
	struct packed_git *p = packed_git;

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

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

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

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

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

	setup_git_directory();

	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!\n");

	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
	9	#include "cache.h"
	10
6d016c9c LS	11	#define BLKSIZE 512
6d016c9c LS	12
9bc0f32c	13	static const char pack_redundant_usage[] =
1c3039e8	14	"git-pack-redundant [ --verbose ] [ --alt-odb ] < --all \| <.pack filename> ...>";
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 {
	55	int i = 1;
	56	new = xmalloc(sizeof(struct llist_item) * BLKSIZE);
	57	for(;i < BLKSIZE; i++) {
	58	llist_item_put(&new[i]);
	59	}
	60	}
60435f68 AR	61	return new;
	62	}
	63
60435f68	64	static void llist_free(struct llist *list)
c283ab21 LS	65	{
	66	while((list->back = list->front)) {
	67	list->front = list->front->next;
60435f68	68	llist_item_put(list->back);
c283ab21 LS	69	}
	70	free(list);
	71	}
	72
bd22c904	73	static inline void llist_init(struct llist **list)
c283ab21 LS	74	{
	75	*list = xmalloc(sizeof(struct llist));
	76	(list)->front = (list)->back = NULL;
	77	(*list)->size = 0;
	78	}
	79
bd22c904	80	static struct llist * llist_copy(struct llist *list)
c283ab21 LS	81	{
	82	struct llist *ret;
	83	struct llist_item new, old, *prev;
	84
	85	llist_init(&ret);
	86
	87	if ((ret->size = list->size) == 0)
	88	return ret;
	89
60435f68	90	new = ret->front = llist_item_get();
c283ab21 LS	91	new->sha1 = list->front->sha1;
	92
	93	old = list->front->next;
	94	while (old) {
	95	prev = new;
60435f68	96	new = llist_item_get();
c283ab21 LS	97	prev->next = new;
	98	new->sha1 = old->sha1;
	99	old = old->next;
	100	}
	101	new->next = NULL;
	102	ret->back = new;
	103
	104	return ret;
	105	}
	106
42873078 NP	107	static inline struct llist_item llist_insert(struct llist list,
	108	struct llist_item *after,
	109	const unsigned char *sha1)
c283ab21	110	{
60435f68	111	struct llist_item *new = llist_item_get();
c283ab21 LS	112	new->sha1 = sha1;
	113	new->next = NULL;
	114
	115	if (after != NULL) {
	116	new->next = after->next;
	117	after->next = new;
	118	if (after == list->back)
	119	list->back = new;
	120	} else {/* insert in front */
	121	if (list->size == 0)
	122	list->back = new;
	123	else
	124	new->next = list->front;
	125	list->front = new;
	126	}
	127	list->size++;
	128	return new;
	129	}
	130
42873078 NP	131	static inline struct llist_item llist_insert_back(struct llist list,
42873078 NP	132	const unsigned char *sha1)
c283ab21 LS	133	{
	134	return llist_insert(list, list->back, sha1);
	135	}
	136
42873078 NP	137	static inline struct llist_item llist_insert_sorted_unique(struct llist list,
42873078 NP	138	const unsigned char sha1, struct llist_item hint)
c283ab21 LS	139	{
	140	struct llist_item prev = NULL, l;
	141
	142	l = (hint == NULL) ? list->front : hint;
	143	while (l) {
a89fccd2	144	int cmp = hashcmp(l->sha1, sha1);
c283ab21 LS	145	if (cmp > 0) { /* we insert before this entry */
	146	return llist_insert(list, prev, sha1);
	147	}
	148	if(!cmp) { /* already exists */
	149	return l;
	150	}
	151	prev = l;
	152	l = l->next;
	153	}
	154	/* insert at the end */
	155	return llist_insert_back(list, sha1);
	156	}
	157
c283ab21	158	/* returns a pointer to an item in front of sha1 */
b99a394c	159	static inline struct llist_item * llist_sorted_remove(struct llist list, const unsigned char sha1, struct llist_item *hint)
c283ab21 LS	160	{
	161	struct llist_item prev, l;
	162
	163	redo_from_start:
	164	l = (hint == NULL) ? list->front : hint;
	165	prev = NULL;
	166	while (l) {
a89fccd2	167	int cmp = hashcmp(l->sha1, sha1);
c283ab21 LS	168	if (cmp > 0) /* not in list, since sorted */
	169	return prev;
	170	if(!cmp) { /* found */
	171	if (prev == NULL) {
	172	if (hint != NULL && hint != list->front) {
	173	/* we don't know the previous element */
	174	hint = NULL;
	175	goto redo_from_start;
	176	}
	177	list->front = l->next;
	178	} else
	179	prev->next = l->next;
	180	if (l == list->back)
	181	list->back = prev;
60435f68	182	llist_item_put(l);
c283ab21 LS	183	list->size--;
	184	return prev;
	185	}
	186	prev = l;
	187	l = l->next;
	188	}
	189	return prev;
	190	}
	191
1a41e743	192	/* computes A\B */
bd22c904	193	static void llist_sorted_difference_inplace(struct llist *A,
1a41e743 LS	194	struct llist *B)
	195	{
	196	struct llist_item hint, b;
	197
	198	hint = NULL;
	199	b = B->front;
	200
	201	while (b) {
	202	hint = llist_sorted_remove(A, b->sha1, hint);
	203	b = b->next;
	204	}
	205	}
	206
bd22c904	207	static inline struct pack_list * pack_list_insert(struct pack_list **pl,
c283ab21 LS	208	struct pack_list *entry)
	209	{
	210	struct pack_list *p = xmalloc(sizeof(struct pack_list));
	211	memcpy(p, entry, sizeof(struct pack_list));
	212	p->next = *pl;
	213	*pl = p;
	214	return p;
	215	}
	216
bd22c904	217	static inline size_t pack_list_size(struct pack_list *pl)
1c3039e8 LS	218	{
	219	size_t ret = 0;
	220	while(pl) {
	221	ret++;
	222	pl = pl->next;
	223	}
	224	return ret;
	225	}
	226
d1ab1577 AR	227	static struct pack_list * pack_list_difference(const struct pack_list *A,
d1ab1577 AR	228	const struct pack_list *B)
c283ab21	229	{
d1ab1577 AR	230	struct pack_list *ret;
d1ab1577 AR	231	const struct pack_list *pl;
c283ab21 LS	232
	233	if (A == NULL)
	234	return NULL;
	235
	236	pl = B;
	237	while (pl != NULL) {
	238	if (A->pack == pl->pack)
	239	return pack_list_difference(A->next, B);
	240	pl = pl->next;
	241	}
	242	ret = xmalloc(sizeof(struct pack_list));
	243	memcpy(ret, A, sizeof(struct pack_list));
	244	ret->next = pack_list_difference(A->next, B);
	245	return ret;
	246	}
	247
bd22c904	248	static void cmp_two_packs(struct pack_list p1, struct pack_list p2)
c283ab21 LS	249	{
c283ab21 LS	250	int p1_off, p2_off;
42873078	251	const unsigned char p1_base, p2_base;
c283ab21	252	struct llist_item p1_hint = NULL, p2_hint = NULL;
1d7f171c	253
c283ab21	254	p1_off = p2_off = 256 * 4 + 4;
42873078 NP	255	p1_base = p1->pack->index_data;
42873078 NP	256	p2_base = p2->pack->index_data;
c283ab21 LS	257
	258	while (p1_off <= p1->pack->index_size - 3 * 20 &&
	259	p2_off <= p2->pack->index_size - 3 * 20)
	260	{
a89fccd2	261	int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
c283ab21 LS	262	/* cmp ~ p1 - p2 */
	263	if (cmp == 0) {
	264	p1_hint = llist_sorted_remove(p1->unique_objects,
	265	p1_base + p1_off, p1_hint);
	266	p2_hint = llist_sorted_remove(p2->unique_objects,
	267	p1_base + p1_off, p2_hint);
	268	p1_off+=24;
	269	p2_off+=24;
	270	continue;
	271	}
	272	if (cmp < 0) { /* p1 has the object, p2 doesn't */
	273	p1_off+=24;
	274	} else { /* p2 has the object, p1 doesn't */
	275	p2_off+=24;
	276	}
	277	}
	278	}
	279
962554c6	280	static void pll_free(struct pll *l)
751a71e2	281	{
6d016c9c LS	282	struct pll *old;
	283	struct pack_list *opl;
	284
	285	while (l) {
	286	old = l;
	287	while (l->pl) {
	288	opl = l->pl;
	289	l->pl = opl->next;
	290	free(opl);
751a71e2	291	}
6d016c9c LS	292	l = l->next;
6d016c9c LS	293	free(old);
751a71e2	294	}
751a71e2 LS	295	}
751a71e2 LS	296
c283ab21 LS	297	/* all the permutations have to be free()d at the same time,
	298	* since they refer to each other
	299	*/
6d016c9c	300	static struct pll * get_permutations(struct pack_list *list, int n)
c283ab21	301	{
6d016c9c LS	302	struct pll subset, ret = NULL, new_pll = NULL, pll;
	303
	304	if (list == NULL \|\| pack_list_size(list) < n \|\| n == 0)
c283ab21 LS	305	return NULL;
c283ab21 LS	306
6d016c9c LS	307	if (n == 1) {
	308	while (list) {
	309	new_pll = xmalloc(sizeof(pll));
	310	new_pll->pl = NULL;
	311	pack_list_insert(&new_pll->pl, list);
	312	new_pll->next = ret;
	313	ret = new_pll;
	314	list = list->next;
	315	}
	316	return ret;
c283ab21 LS	317	}
c283ab21 LS	318
6d016c9c LS	319	while (list->next) {
	320	subset = get_permutations(list->next, n - 1);
	321	while (subset) {
	322	new_pll = xmalloc(sizeof(pll));
	323	new_pll->pl = subset->pl;
	324	pack_list_insert(&new_pll->pl, list);
	325	new_pll->next = ret;
	326	ret = new_pll;
	327	subset = subset->next;
751a71e2	328	}
6d016c9c	329	list = list->next;
c283ab21	330	}
6d016c9c	331	return ret;
c283ab21 LS	332	}
c283ab21 LS	333
bd22c904	334	static int is_superset(struct pack_list pl, struct llist list)
c283ab21	335	{
1c3039e8	336	struct llist *diff;
c283ab21 LS	337
	338	diff = llist_copy(list);
	339
	340	while (pl) {
d1ab1577	341	llist_sorted_difference_inplace(diff, pl->all_objects);
c283ab21 LS	342	if (diff->size == 0) { /* we're done */
	343	llist_free(diff);
	344	return 1;
	345	}
	346	pl = pl->next;
	347	}
	348	llist_free(diff);
	349	return 0;
	350	}
	351
bd22c904	352	static size_t sizeof_union(struct packed_git p1, struct packed_git p2)
c283ab21 LS	353	{
	354	size_t ret = 0;
	355	int p1_off, p2_off;
42873078	356	const unsigned char p1_base, p2_base;
c283ab21 LS	357
c283ab21 LS	358	p1_off = p2_off = 256 * 4 + 4;
42873078 NP	359	p1_base = p1->index_data;
42873078 NP	360	p2_base = p2->index_data;
c283ab21 LS	361
	362	while (p1_off <= p1->index_size - 3 * 20 &&
	363	p2_off <= p2->index_size - 3 * 20)
	364	{
a89fccd2	365	int cmp = hashcmp(p1_base + p1_off, p2_base + p2_off);
c283ab21 LS	366	/* cmp ~ p1 - p2 */
	367	if (cmp == 0) {
	368	ret++;
	369	p1_off+=24;
	370	p2_off+=24;
	371	continue;
	372	}
	373	if (cmp < 0) { /* p1 has the object, p2 doesn't */
	374	p1_off+=24;
	375	} else { /* p2 has the object, p1 doesn't */
	376	p2_off+=24;
	377	}
	378	}
	379	return ret;
	380	}
	381
	382	/* another O(n^2) function ... */
bd22c904	383	static size_t get_pack_redundancy(struct pack_list *pl)
c283ab21 LS	384	{
c283ab21 LS	385	struct pack_list *subset;
bd22c904	386	size_t ret = 0;
1c3039e8 LS	387
	388	if (pl == NULL)
	389	return 0;
	390
c283ab21 LS	391	while ((subset = pl->next)) {
	392	while(subset) {
	393	ret += sizeof_union(pl->pack, subset->pack);
	394	subset = subset->next;
	395	}
	396	pl = pl->next;
	397	}
	398	return ret;
	399	}
	400
c4001d92	401	static inline off_t pack_set_bytecount(struct pack_list *pl)
c283ab21	402	{
c4001d92	403	off_t ret = 0;
c283ab21 LS	404	while (pl) {
	405	ret += pl->pack->pack_size;
	406	ret += pl->pack->index_size;
	407	pl = pl->next;
	408	}
	409	return ret;
	410	}
	411
bd22c904	412	static void minimize(struct pack_list **min)
c283ab21 LS	413	{
	414	struct pack_list pl, unique = NULL,
	415	non_unique = NULL, min_perm = NULL;
	416	struct pll perm, perm_all, perm_ok = NULL, new_perm;
1c3039e8	417	struct llist *missing;
c4001d92	418	off_t min_perm_size = 0, perm_size;
6d016c9c	419	int n;
c283ab21	420
1c3039e8	421	pl = local_packs;
c283ab21 LS	422	while (pl) {
	423	if(pl->unique_objects->size)
	424	pack_list_insert(&unique, pl);
	425	else
	426	pack_list_insert(&non_unique, pl);
	427	pl = pl->next;
	428	}
	429	/* find out which objects are missing from the set of unique packs */
	430	missing = llist_copy(all_objects);
	431	pl = unique;
	432	while (pl) {
6d016c9c	433	llist_sorted_difference_inplace(missing, pl->all_objects);
c283ab21 LS	434	pl = pl->next;
	435	}
	436
1c3039e8	437	/* return if there are no objects missing from the unique set */
c283ab21 LS	438	if (missing->size == 0) {
	439	*min = unique;
	440	return;
	441	}
	442
	443	/* find the permutations which contain all missing objects */
6d016c9c LS	444	for (n = 1; n <= pack_list_size(non_unique) && !perm_ok; n++) {
	445	perm_all = perm = get_permutations(non_unique, n);
	446	while (perm) {
	447	if (is_superset(perm->pl, missing)) {
	448	new_perm = xmalloc(sizeof(struct pll));
	449	memcpy(new_perm, perm, sizeof(struct pll));
	450	new_perm->next = perm_ok;
	451	perm_ok = new_perm;
	452	}
	453	perm = perm->next;
c283ab21	454	}
6d016c9c LS	455	if (perm_ok)
	456	break;
	457	pll_free(perm_all);
c283ab21	458	}
c283ab21 LS	459	if (perm_ok == NULL)
	460	die("Internal error: No complete sets found!\n");
	461
	462	/* find the permutation with the smallest size */
	463	perm = perm_ok;
	464	while (perm) {
	465	perm_size = pack_set_bytecount(perm->pl);
c4001d92	466	if (!min_perm_size \|\| min_perm_size > perm_size) {
c283ab21 LS	467	min_perm_size = perm_size;
	468	min_perm = perm->pl;
	469	}
	470	perm = perm->next;
	471	}
	472	*min = min_perm;
	473	/* add the unique packs to the list */
	474	pl = unique;
	475	while(pl) {
	476	pack_list_insert(min, pl);
	477	pl = pl->next;
	478	}
	479	}
	480
bd22c904	481	static void load_all_objects(void)
c283ab21	482	{
1c3039e8	483	struct pack_list *pl = local_packs;
c283ab21	484	struct llist_item hint, l;
c283ab21 LS	485
	486	llist_init(&all_objects);
	487
	488	while (pl) {
c283ab21 LS	489	hint = NULL;
	490	l = pl->all_objects->front;
	491	while (l) {
	492	hint = llist_insert_sorted_unique(all_objects,
	493	l->sha1, hint);
	494	l = l->next;
	495	}
	496	pl = pl->next;
	497	}
1c3039e8 LS	498	/* remove objects present in remote packs */
	499	pl = altodb_packs;
	500	while (pl) {
	501	llist_sorted_difference_inplace(all_objects, pl->all_objects);
	502	pl = pl->next;
	503	}
c283ab21 LS	504	}
	505
	506	/* this scales like O(n^2) */
bd22c904	507	static void cmp_local_packs(void)
c283ab21	508	{
1c3039e8	509	struct pack_list subset, pl = local_packs;
c283ab21	510
1c3039e8	511	while ((subset = pl)) {
c283ab21	512	while((subset = subset->next))
1c3039e8 LS	513	cmp_two_packs(pl, subset);
	514	pl = pl->next;
	515	}
06a45c8c	516	}
1c3039e8	517
bd22c904	518	static void scan_alt_odb_packs(void)
06a45c8c LS	519	{
	520	struct pack_list local, alt;
	521
	522	alt = altodb_packs;
	523	while (alt) {
	524	local = local_packs;
	525	while (local) {
	526	llist_sorted_difference_inplace(local->unique_objects,
	527	alt->all_objects);
	528	local = local->next;
1c3039e8	529	}
6d016c9c	530	llist_sorted_difference_inplace(all_objects, alt->all_objects);
06a45c8c	531	alt = alt->next;
c283ab21 LS	532	}
	533	}
	534
bd22c904	535	static struct pack_list * add_pack(struct packed_git *p)
c283ab21 LS	536	{
	537	struct pack_list l;
	538	size_t off;
42873078	539	const unsigned char *base;
c283ab21	540
06a45c8c LS	541	if (!p->pack_local && !(alt_odb \|\| verbose))
	542	return NULL;
	543
c283ab21 LS	544	l.pack = p;
	545	llist_init(&l.all_objects);
	546
	547	off = 256 * 4 + 4;
42873078	548	base = p->index_data;
c283ab21 LS	549	while (off <= p->index_size - 3 * 20) {
c283ab21 LS	550	llist_insert_back(l.all_objects, base + off);
06a45c8c	551	off += 24;
c283ab21 LS	552	}
	553	/* this list will be pruned in cmp_two_packs later */
	554	l.unique_objects = llist_copy(l.all_objects);
1c3039e8 LS	555	if (p->pack_local)
	556	return pack_list_insert(&local_packs, &l);
	557	else
06a45c8c	558	return pack_list_insert(&altodb_packs, &l);
c283ab21 LS	559	}
c283ab21 LS	560
bd22c904	561	static struct pack_list * add_pack_file(char *filename)
c283ab21 LS	562	{
	563	struct packed_git *p = packed_git;
	564
	565	if (strlen(filename) < 40)
	566	die("Bad pack filename: %s\n", filename);
	567
	568	while (p) {
	569	if (strstr(p->pack_name, filename))
c283ab21 LS	570	return add_pack(p);
	571	p = p->next;
	572	}
	573	die("Filename %s not found in packed_git\n", filename);
	574	}
	575
bd22c904	576	static void load_all(void)
c283ab21 LS	577	{
	578	struct packed_git *p = packed_git;
	579
	580	while (p) {
1c3039e8	581	add_pack(p);
c283ab21 LS	582	p = p->next;
	583	}
	584	}
	585
	586	int main(int argc, char **argv)
	587	{
	588	int i;
	589	struct pack_list min, red, *pl;
bb931cf9	590	struct llist *ignore;
b99a394c JH	591	unsigned char *sha1;
b99a394c JH	592	char buf[42]; /* 40 byte sha1 + \n + \0 */
c283ab21	593
53228a5f JH	594	setup_git_directory();
53228a5f JH	595
c283ab21 LS	596	for (i = 1; i < argc; i++) {
c283ab21 LS	597	const char *arg = argv[i];
9bc0f32c LS	598	if(!strcmp(arg, "--")) {
9bc0f32c LS	599	i++;
c283ab21	600	break;
9bc0f32c	601	}
1c3039e8 LS	602	if(!strcmp(arg, "--all")) {
1c3039e8 LS	603	load_all_packs = 1;
c283ab21 LS	604	continue;
c283ab21 LS	605	}
1c3039e8	606	if(!strcmp(arg, "--verbose")) {
c283ab21 LS	607	verbose = 1;
	608	continue;
	609	}
1c3039e8 LS	610	if(!strcmp(arg, "--alt-odb")) {
	611	alt_odb = 1;
	612	continue;
	613	}
c283ab21	614	if(*arg == '-')
9bc0f32c	615	usage(pack_redundant_usage);
c283ab21 LS	616	else
	617	break;
	618	}
	619
	620	prepare_packed_git();
	621
1c3039e8	622	if (load_all_packs)
c283ab21 LS	623	load_all();
	624	else
	625	while (*(argv + i) != NULL)
	626	add_pack_file(*(argv + i++));
	627
1c3039e8	628	if (local_packs == NULL)
c283ab21 LS	629	die("Zero packs found!\n");
c283ab21 LS	630
c283ab21 LS	631	load_all_objects();
c283ab21 LS	632
06a45c8c LS	633	cmp_local_packs();
	634	if (alt_odb)
	635	scan_alt_odb_packs();
	636
bb931cf9 LS	637	/* ignore objects given on stdin */
	638	llist_init(&ignore);
	639	if (!isatty(0)) {
	640	while (fgets(buf, sizeof(buf), stdin)) {
	641	sha1 = xmalloc(20);
	642	if (get_sha1_hex(buf, sha1))
	643	die("Bad sha1 on stdin: %s", buf);
	644	llist_insert_sorted_unique(ignore, sha1, NULL);
	645	}
	646	}
	647	llist_sorted_difference_inplace(all_objects, ignore);
	648	pl = local_packs;
	649	while (pl) {
	650	llist_sorted_difference_inplace(pl->unique_objects, ignore);
	651	pl = pl->next;
	652	}
	653
c283ab21	654	minimize(&min);
06a45c8c	655
c283ab21	656	if (verbose) {
abacbe41 KR	657	fprintf(stderr, "There are %lu packs available in alt-odbs.\n",
abacbe41 KR	658	(unsigned long)pack_list_size(altodb_packs));
c283ab21 LS	659	fprintf(stderr, "The smallest (bytewise) set of packs is:\n");
	660	pl = min;
	661	while (pl) {
	662	fprintf(stderr, "\t%s\n", pl->pack->pack_name);
	663	pl = pl->next;
	664	}
abacbe41 KR	665	fprintf(stderr, "containing %lu duplicate objects "
	666	"with a total size of %lukb.\n",
	667	(unsigned long)get_pack_redundancy(min),
	668	(unsigned long)pack_set_bytecount(min)/1024);
	669	fprintf(stderr, "A total of %lu unique objects were considered.\n",
	670	(unsigned long)all_objects->size);
c283ab21 LS	671	fprintf(stderr, "Redundant packs (with indexes):\n");
c283ab21 LS	672	}
1c3039e8	673	pl = red = pack_list_difference(local_packs, min);
c283ab21 LS	674	while (pl) {
c283ab21 LS	675	printf("%s\n%s\n",
1c3039e8 LS	676	sha1_pack_index_name(pl->pack->sha1),
1c3039e8 LS	677	pl->pack->pack_name);
c283ab21 LS	678	pl = pl->next;
c283ab21 LS	679	}
bb931cf9	680	if (verbose)
b99a394c JH	681	fprintf(stderr, "%luMB of redundant packs in total.\n",
b99a394c JH	682	(unsigned long)pack_set_bytecount(red)/(1024*1024));
c283ab21 LS	683
	684	return 0;
	685	}