[people/ms/libloc.git] / src / database.c

/*
	libloc - A library to determine the location of someone on the Internet

	Copyright (C) 2017 IPFire Development Team <info@ipfire.org>

	This library is free software; you can redistribute it and/or
	modify it under the terms of the GNU Lesser General Public
	License as published by the Free Software Foundation; either
	version 2.1 of the License, or (at your option) any later version.

	This library is distributed in the hope that it will be useful,
	but WITHOUT ANY WARRANTY; without even the implied warranty of
	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
	Lesser General Public License for more details.
*/

#include <arpa/inet.h>
#include <endian.h>
#include <errno.h>
#include <netinet/in.h>
#include <stddef.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/mman.h>
#include <sys/types.h>
#include <time.h>
#include <unistd.h>

#include <loc/libloc.h>
#include <loc/as.h>
#include <loc/database.h>
#include <loc/format.h>
#include <loc/network.h>
#include <loc/private.h>
#include <loc/stringpool.h>

struct loc_database {
	struct loc_ctx* ctx;
	int refcount;

	unsigned int version;
	time_t created_at;
	off_t vendor;
	off_t description;
	off_t license;

	// ASes in the database
	struct loc_database_as_v0* as_v0;
	size_t as_count;

	// Network tree
	struct loc_database_network_node_v0* network_nodes_v0;
	size_t network_nodes_count;

	// Networks
	struct loc_database_network_v0* networks_v0;
	size_t networks_count;

	struct loc_stringpool* pool;
};

static int loc_database_read_magic(struct loc_database* db, FILE* f) {
	struct loc_database_magic magic;

	// Read from file
	size_t bytes_read = fread(&magic, 1, sizeof(magic), f);

	// Check if we have been able to read enough data
	if (bytes_read < sizeof(magic)) {
		ERROR(db->ctx, "Could not read enough data to validate magic bytes\n");
		DEBUG(db->ctx, "Read %zu bytes, but needed %zu\n", bytes_read, sizeof(magic));
		return -ENOMSG;
	}

	// Compare magic bytes
	if (memcmp(LOC_DATABASE_MAGIC, magic.magic, strlen(LOC_DATABASE_MAGIC)) == 0) {
		DEBUG(db->ctx, "Magic value matches\n");

		// Parse version
		db->version = be16toh(magic.version);
		DEBUG(db->ctx, "Database version is %u\n", db->version);

		return 0;
	}

	ERROR(db->ctx, "Database format is not compatible\n");

	// Return an error
	return 1;
}

static int loc_database_read_as_section_v0(struct loc_database* db,
		FILE* f, const struct loc_database_header_v0* header) {
	off_t as_offset  = be32toh(header->as_offset);
	size_t as_length = be32toh(header->as_length);

	DEBUG(db->ctx, "Reading AS section from %jd (%zu bytes)\n", as_offset, as_length);

	if (as_length > 0) {
		db->as_v0 = mmap(NULL, as_length, PROT_READ,
			MAP_SHARED, fileno(f), as_offset);

		if (db->as_v0 == MAP_FAILED)
			return -errno;
	}

	db->as_count = as_length / sizeof(*db->as_v0);

	INFO(db->ctx, "Read %zu ASes from the database\n", db->as_count);

	return 0;
}

static int loc_database_read_network_nodes_section_v0(struct loc_database* db,
		FILE* f, const struct loc_database_header_v0* header) {
	off_t network_nodes_offset  = be32toh(header->network_tree_offset);
	size_t network_nodes_length = be32toh(header->network_tree_length);

	DEBUG(db->ctx, "Reading network nodes section from %jd (%zu bytes)\n",
		network_nodes_offset, network_nodes_length);

	if (network_nodes_length > 0) {
		db->network_nodes_v0 = mmap(NULL, network_nodes_length, PROT_READ,
			MAP_SHARED, fileno(f), network_nodes_offset);

		if (db->network_nodes_v0 == MAP_FAILED)
			return -errno;
	}

	db->network_nodes_count = network_nodes_length / sizeof(*db->network_nodes_v0);

	INFO(db->ctx, "Read %zu network nodes from the database\n", db->network_nodes_count);

	return 0;
}

static int loc_database_read_networks_section_v0(struct loc_database* db,
		FILE* f, const struct loc_database_header_v0* header) {
	off_t networks_offset  = be32toh(header->network_data_offset);
	size_t networks_length = be32toh(header->network_data_length);

	DEBUG(db->ctx, "Reading networks section from %jd (%zu bytes)\n",
		networks_offset, networks_length);

	if (networks_length > 0) {
		db->networks_v0 = mmap(NULL, networks_length, PROT_READ,
			MAP_SHARED, fileno(f), networks_offset);

		if (db->networks_v0 == MAP_FAILED)
			return -errno;
	}

	db->networks_count = networks_length / sizeof(*db->networks_v0);

	INFO(db->ctx, "Read %zu networks from the database\n", db->networks_count);

	return 0;
}

static int loc_database_read_header_v0(struct loc_database* db, FILE* f) {
	struct loc_database_header_v0 header;

	// Read from file
	size_t size = fread(&header, 1, sizeof(header), f);

	if (size < sizeof(header)) {
		ERROR(db->ctx, "Could not read enough data for header\n");
		return -ENOMSG;
	}

	// Copy over data
	db->created_at  = be64toh(header.created_at);
	db->vendor      = be32toh(header.vendor);
	db->description = be32toh(header.description);
	db->license     = be32toh(header.license);

	// Open pool
	off_t pool_offset  = be32toh(header.pool_offset);
	size_t pool_length = be32toh(header.pool_length);

	int r = loc_stringpool_open(db->ctx, &db->pool,
		f, pool_length, pool_offset);
	if (r)
		return r;

	// AS section
	r = loc_database_read_as_section_v0(db, f, &header);
	if (r)
		return r;

	// Network Nodes
	r = loc_database_read_network_nodes_section_v0(db, f, &header);
	if (r)
		return r;

	// Networks
	r = loc_database_read_networks_section_v0(db, f, &header);
	if (r)
		return r;

	return 0;
}

static int loc_database_read_header(struct loc_database* db, FILE* f) {
	switch (db->version) {
		case 0:
			return loc_database_read_header_v0(db, f);

		default:
			ERROR(db->ctx, "Incompatible database version: %u\n", db->version);
			return 1;
	}
}

static int loc_database_read(struct loc_database* db, FILE* f) {
	clock_t start = clock();

	// Read magic bytes
	int r = loc_database_read_magic(db, f);
	if (r)
		return r;

	// Read the header
	r = loc_database_read_header(db, f);
	if (r)
		return r;

	clock_t end = clock();

	INFO(db->ctx, "Opened database in %.8fs\n",
		(double)(end - start) / CLOCKS_PER_SEC);

	return 0;
}

LOC_EXPORT int loc_database_new(struct loc_ctx* ctx, struct loc_database** database, FILE* f) {
	// Fail on invalid file handle
	if (!f)
		return -EINVAL;

	struct loc_database* db = calloc(1, sizeof(*db));
	if (!db)
		return -ENOMEM;

	// Reference context
	db->ctx = loc_ref(ctx);
	db->refcount = 1;

	DEBUG(db->ctx, "Database object allocated at %p\n", db);

	int r = loc_database_read(db, f);
	if (r) {
		loc_database_unref(db);
		return r;
	}

	*database = db;

	return 0;
}

LOC_EXPORT struct loc_database* loc_database_ref(struct loc_database* db) {
	db->refcount++;

	return db;
}

static void loc_database_free(struct loc_database* db) {
	int r;

	DEBUG(db->ctx, "Releasing database %p\n", db);

	// Removing all ASes
	if (db->as_v0) {
		r = munmap(db->as_v0, db->as_count * sizeof(*db->as_v0));
		if (r)
			ERROR(db->ctx, "Could not unmap AS section: %s\n", strerror(errno));
	}

	// Remove mapped network sections
	if (db->networks_v0) {
		r = munmap(db->networks_v0, db->networks_count * sizeof(*db->networks_v0));
		if (r)
			ERROR(db->ctx, "Could not unmap networks section: %s\n", strerror(errno));
	}

	// Remove mapped network nodes section
	if (db->network_nodes_v0) {
		r = munmap(db->network_nodes_v0, db->network_nodes_count * sizeof(*db->network_nodes_v0));
		if (r)
			ERROR(db->ctx, "Could not unmap network nodes section: %s\n", strerror(errno));
	}

	loc_stringpool_unref(db->pool);

	loc_unref(db->ctx);
	free(db);
}

LOC_EXPORT struct loc_database* loc_database_unref(struct loc_database* db) {
	if (--db->refcount > 0)
		return NULL;

	loc_database_free(db);
	return NULL;
}

LOC_EXPORT time_t loc_database_created_at(struct loc_database* db) {
	return db->created_at;
}

LOC_EXPORT const char* loc_database_get_vendor(struct loc_database* db) {
	return loc_stringpool_get(db->pool, db->vendor);
}

LOC_EXPORT const char* loc_database_get_description(struct loc_database* db) {
	return loc_stringpool_get(db->pool, db->description);
}

LOC_EXPORT const char* loc_database_get_license(struct loc_database* db) {
	return loc_stringpool_get(db->pool, db->license);
}

LOC_EXPORT size_t loc_database_count_as(struct loc_database* db) {
	return db->as_count;
}

// Returns the AS at position pos
static int loc_database_fetch_as(struct loc_database* db, struct loc_as** as, off_t pos) {
	if ((size_t)pos >= db->as_count)
		return -EINVAL;

	DEBUG(db->ctx, "Fetching AS at position %jd\n", pos);

	int r;
	switch (db->version) {
		case 0:
			r = loc_as_new_from_database_v0(db->ctx, db->pool, as, db->as_v0 + pos);
			break;

		default:
			return -1;
	}

	if (r == 0) {
		DEBUG(db->ctx, "Got AS%u\n", loc_as_get_number(*as));
	}

	return r;
}

// Performs a binary search to find the AS in the list
LOC_EXPORT int loc_database_get_as(struct loc_database* db, struct loc_as** as, uint32_t number) {
	off_t lo = 0;
	off_t hi = db->as_count - 1;

	// Save start time
	clock_t start = clock();

	while (lo <= hi) {
		off_t i = (lo + hi) / 2;

		// Fetch AS in the middle between lo and hi
		int r = loc_database_fetch_as(db, as, i);
		if (r)
			return r;

		// Check if this is a match
		uint32_t as_number = loc_as_get_number(*as);
		if (as_number == number) {
			clock_t end = clock();

			// Log how fast this has been
			DEBUG(db->ctx, "Found AS%u in %.8fs\n", as_number,
				(double)(end - start) / CLOCKS_PER_SEC);

			return 0;
		}

		// If it wasn't, we release the AS and
		// adjust our search pointers
		loc_as_unref(*as);

		if (as_number < number) {
			lo = i + 1;
		} else
			hi = i - 1;
	}

	// Nothing found
	*as = NULL;

	return 1;
}

// Returns the network at position pos
static int loc_database_fetch_network(struct loc_database* db, struct loc_network** network, struct in6_addr* address, off_t pos) {
	if ((size_t)pos >= db->networks_count)
		return -EINVAL;

	DEBUG(db->ctx, "Fetching network at position %jd\n", pos);

	int r;
	switch (db->version) {
		case 0:
			r = loc_network_new_from_database_v0(db->ctx, network, address, db->networks_v0 + pos);
			break;

		default:
			return -1;
	}

	if (r == 0) {
		char* string = loc_network_str(*network);
		DEBUG(db->ctx, "Got network %s\n", string);
		free(string);
	}

	return r;
}

static int __loc_database_node_is_leaf(const struct loc_database_network_node_v0* node) {
	return (node->zero == htobe32(0xffffffff));
}

static int __loc_database_lookup_handle_leaf(struct loc_database* db, const struct in6_addr* address,
		struct loc_network** network, struct in6_addr* network_address,
		const struct loc_database_network_node_v0* node) {
	DEBUG(db->ctx, "Handling leaf node at %jd\n", node - db->network_nodes_v0);

	// Fetch the network
	int r = loc_database_fetch_network(db, network,
		network_address, be32toh(node->one));
	if (r)
		return r;

	// Check if the given IP address is inside the network
	r = loc_network_match_address(*network, address);
	if (r) {
		DEBUG(db->ctx, "Searched address is not part of the network\n");

		loc_network_unref(*network);
		*network = NULL;
		return 1;
	}

	// A network was found and the IP address matches
	return 0;
}

// Returns the highest result available
static int __loc_database_lookup_max(struct loc_database* db, const struct in6_addr* address,
		struct loc_network** network, struct in6_addr* network_address,
		const struct loc_database_network_node_v0* node, unsigned int level) {
	// If the node is a leaf node, we end here
	if (__loc_database_node_is_leaf(node))
		return __loc_database_lookup_handle_leaf(db, address, network, network_address, node);

	int r;
	off_t node_index;

	// Try to go down the ones path first
	if (node->one) {
		node_index = be32toh(node->one);
		in6_addr_set_bit(network_address, level, 1);

		// Check boundaries
		if (node_index > 0 && (size_t)node_index <= db->network_nodes_count) {
			r = __loc_database_lookup_max(db, address, network, network_address,
				db->network_nodes_v0 + node_index, level + 1);

			// Abort when match was found or error
			if (r <= 0)
				return r;
		}
	}

	// ... and if that fails, we try to go down one step on a zero
	// branch and then try the ones again...
	if (node->zero) {
		node_index = be32toh(node->zero);
		in6_addr_set_bit(network_address, level, 0);

		// Check boundaries
		if (node_index > 0 && (size_t)node_index <= db->network_nodes_count) {
			r = __loc_database_lookup_max(db, address, network, network_address,
				db->network_nodes_v0 + node_index, level + 1);

			// Abort when match was found or error
			if (r <= 0)
				return r;
		}
	}

	// End of path
	return 1;
}

// Searches for an exact match along the path
static int __loc_database_lookup(struct loc_database* db, const struct in6_addr* address,
		struct loc_network** network, struct in6_addr* network_address,
		const struct loc_database_network_node_v0* node, unsigned int level) {
	// If the node is a leaf node, we end here
	if (__loc_database_node_is_leaf(node))
		return __loc_database_lookup_handle_leaf(db, address, network, network_address, node);

	int r;
	off_t node_index;

	// Follow the path
	int bit = in6_addr_get_bit(address, level);
	in6_addr_set_bit(network_address, level, bit);

	if (bit == 0)
		node_index = be32toh(node->zero);
	else
		node_index = be32toh(node->one);

	// If we point back to root, the path ends here
	if (node_index == 0) {
		DEBUG(db->ctx, "Tree ends here\n");
		return 1;
	}

	// Check boundaries
	if ((size_t)node_index >= db->network_nodes_count)
		return -EINVAL;

	// Move on to the next node
	r = __loc_database_lookup(db, address, network, network_address,
		db->network_nodes_v0 + node_index, level + 1);

	// End here if a result was found
	if (r == 0)
		return r;

	// Raise any errors
	else if (r < 0)
		return r;

	DEBUG(db->ctx, "Could not find an exact match at %u\n", level);

	// If nothing was found, we have to search for an inexact match
	return __loc_database_lookup_max(db, address, network, network_address, node, level);
}

LOC_EXPORT int loc_database_lookup(struct loc_database* db,
		struct in6_addr* address, struct loc_network** network) {
	struct in6_addr network_address;
	memset(&network_address, 0, sizeof(network_address));

	*network = NULL;

	// Save start time
	clock_t start = clock();

	int r = __loc_database_lookup(db, address, network, &network_address,
		db->network_nodes_v0, 0);

	clock_t end = clock();

	// Log how fast this has been
	DEBUG(db->ctx, "Executed network search in %.8fs\n",
		(double)(end - start) / CLOCKS_PER_SEC);

	return r;
}

LOC_EXPORT int loc_database_lookup_from_string(struct loc_database* db,
		const char* string, struct loc_network** network) {
	struct in6_addr address;

	int r = loc_parse_address(db->ctx, string, &address);
	if (r)
		return r;

	return loc_database_lookup(db, &address, network);
}
Commit	Line	Data
2601e83e MT	1	/*
	2	libloc - A library to determine the location of someone on the Internet
	3
	4	Copyright (C) 2017 IPFire Development Team <info@ipfire.org>
	5
	6	This library is free software; you can redistribute it and/or
	7	modify it under the terms of the GNU Lesser General Public
	8	License as published by the Free Software Foundation; either
	9	version 2.1 of the License, or (at your option) any later version.
	10
	11	This library is distributed in the hope that it will be useful,
	12	but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	14	Lesser General Public License for more details.
	15	*/
	16
2a30e4de	17	#include <arpa/inet.h>
0676cd80	18	#include <endian.h>
2601e83e	19	#include <errno.h>
10778041	20	#include <netinet/in.h>
2601e83e MT	21	#include <stddef.h>
	22	#include <stdint.h>
	23	#include <stdio.h>
	24	#include <stdlib.h>
	25	#include <string.h>
c182393f	26	#include <sys/mman.h>
2601e83e	27	#include <sys/types.h>
96ea74a5	28	#include <time.h>
3f35869a	29	#include <unistd.h>
2601e83e MT	30
2601e83e MT	31	#include <loc/libloc.h>
9fc7f001 MT	32	#include <loc/as.h>
9fc7f001 MT	33	#include <loc/database.h>
a5db3e49	34	#include <loc/format.h>
10778041	35	#include <loc/network.h>
9fc7f001 MT	36	#include <loc/private.h>
9fc7f001 MT	37	#include <loc/stringpool.h>
2601e83e MT	38
	39	struct loc_database {
	40	struct loc_ctx* ctx;
	41	int refcount;
	42
	43	unsigned int version;
96ea74a5	44	time_t created_at;
2601e83e MT	45	off_t vendor;
2601e83e MT	46	off_t description;
4bf49d00	47	off_t license;
2601e83e	48
a5db3e49	49	// ASes in the database
c182393f	50	struct loc_database_as_v0* as_v0;
a5db3e49 MT	51	size_t as_count;
a5db3e49 MT	52
f66b7b09 MT	53	// Network tree
	54	struct loc_database_network_node_v0* network_nodes_v0;
	55	size_t network_nodes_count;
	56
a735a563 MT	57	// Networks
	58	struct loc_database_network_v0* networks_v0;
	59	size_t networks_count;
	60
2601e83e MT	61	struct loc_stringpool* pool;
	62	};
	63
a7431f1a	64	static int loc_database_read_magic(struct loc_database* db, FILE* f) {
2601e83e MT	65	struct loc_database_magic magic;
	66
	67	// Read from file
a7431f1a	68	size_t bytes_read = fread(&magic, 1, sizeof(magic), f);
2601e83e MT	69
	70	// Check if we have been able to read enough data
	71	if (bytes_read < sizeof(magic)) {
	72	ERROR(db->ctx, "Could not read enough data to validate magic bytes\n");
	73	DEBUG(db->ctx, "Read %zu bytes, but needed %zu\n", bytes_read, sizeof(magic));
	74	return -ENOMSG;
	75	}
	76
	77	// Compare magic bytes
	78	if (memcmp(LOC_DATABASE_MAGIC, magic.magic, strlen(LOC_DATABASE_MAGIC)) == 0) {
	79	DEBUG(db->ctx, "Magic value matches\n");
	80
	81	// Parse version
0676cd80	82	db->version = be16toh(magic.version);
2601e83e MT	83	DEBUG(db->ctx, "Database version is %u\n", db->version);
	84
	85	return 0;
	86	}
	87
	88	ERROR(db->ctx, "Database format is not compatible\n");
	89
	90	// Return an error
	91	return 1;
	92	}
	93
a5db3e49	94	static int loc_database_read_as_section_v0(struct loc_database* db,
edb4ba7c MT	95	FILE* f, const struct loc_database_header_v0* header) {
	96	off_t as_offset = be32toh(header->as_offset);
	97	size_t as_length = be32toh(header->as_length);
	98
c182393f	99	DEBUG(db->ctx, "Reading AS section from %jd (%zu bytes)\n", as_offset, as_length);
a5db3e49	100
c182393f MT	101	if (as_length > 0) {
c182393f MT	102	db->as_v0 = mmap(NULL, as_length, PROT_READ,
a7431f1a	103	MAP_SHARED, fileno(f), as_offset);
a5db3e49	104
c182393f MT	105	if (db->as_v0 == MAP_FAILED)
c182393f MT	106	return -errno;
a5db3e49 MT	107	}
a5db3e49 MT	108
c182393f MT	109	db->as_count = as_length / sizeof(*db->as_v0);
c182393f MT	110
a5db3e49 MT	111	INFO(db->ctx, "Read %zu ASes from the database\n", db->as_count);
	112
	113	return 0;
	114	}
	115
f66b7b09	116	static int loc_database_read_network_nodes_section_v0(struct loc_database* db,
edb4ba7c MT	117	FILE* f, const struct loc_database_header_v0* header) {
	118	off_t network_nodes_offset = be32toh(header->network_tree_offset);
	119	size_t network_nodes_length = be32toh(header->network_tree_length);
	120
f66b7b09 MT	121	DEBUG(db->ctx, "Reading network nodes section from %jd (%zu bytes)\n",
	122	network_nodes_offset, network_nodes_length);
	123
	124	if (network_nodes_length > 0) {
	125	db->network_nodes_v0 = mmap(NULL, network_nodes_length, PROT_READ,
	126	MAP_SHARED, fileno(f), network_nodes_offset);
	127
	128	if (db->network_nodes_v0 == MAP_FAILED)
	129	return -errno;
	130	}
	131
	132	db->network_nodes_count = network_nodes_length / sizeof(*db->network_nodes_v0);
	133
	134	INFO(db->ctx, "Read %zu network nodes from the database\n", db->network_nodes_count);
	135
	136	return 0;
	137	}
	138
a735a563 MT	139	static int loc_database_read_networks_section_v0(struct loc_database* db,
	140	FILE* f, const struct loc_database_header_v0* header) {
	141	off_t networks_offset = be32toh(header->network_data_offset);
	142	size_t networks_length = be32toh(header->network_data_length);
	143
	144	DEBUG(db->ctx, "Reading networks section from %jd (%zu bytes)\n",
	145	networks_offset, networks_length);
	146
	147	if (networks_length > 0) {
	148	db->networks_v0 = mmap(NULL, networks_length, PROT_READ,
	149	MAP_SHARED, fileno(f), networks_offset);
	150
	151	if (db->networks_v0 == MAP_FAILED)
	152	return -errno;
	153	}
	154
	155	db->networks_count = networks_length / sizeof(*db->networks_v0);
	156
	157	INFO(db->ctx, "Read %zu networks from the database\n", db->networks_count);
	158
	159	return 0;
	160	}
	161
a7431f1a	162	static int loc_database_read_header_v0(struct loc_database* db, FILE* f) {
2601e83e MT	163	struct loc_database_header_v0 header;
	164
	165	// Read from file
a7431f1a	166	size_t size = fread(&header, 1, sizeof(header), f);
2601e83e MT	167
	168	if (size < sizeof(header)) {
	169	ERROR(db->ctx, "Could not read enough data for header\n");
	170	return -ENOMSG;
	171	}
	172
	173	// Copy over data
96ea74a5	174	db->created_at = be64toh(header.created_at);
0676cd80 MT	175	db->vendor = be32toh(header.vendor);
0676cd80 MT	176	db->description = be32toh(header.description);
4bf49d00	177	db->license = be32toh(header.license);
2601e83e MT	178
2601e83e MT	179	// Open pool
0676cd80 MT	180	off_t pool_offset = be32toh(header.pool_offset);
0676cd80 MT	181	size_t pool_length = be32toh(header.pool_length);
2601e83e	182
0e974d4b	183	int r = loc_stringpool_open(db->ctx, &db->pool,
a7431f1a	184	f, pool_length, pool_offset);
2601e83e MT	185	if (r)
	186	return r;
	187
a5db3e49	188	// AS section
edb4ba7c	189	r = loc_database_read_as_section_v0(db, f, &header);
a5db3e49 MT	190	if (r)
	191	return r;
	192
f66b7b09	193	// Network Nodes
edb4ba7c	194	r = loc_database_read_network_nodes_section_v0(db, f, &header);
f66b7b09 MT	195	if (r)
	196	return r;
	197
a735a563 MT	198	// Networks
	199	r = loc_database_read_networks_section_v0(db, f, &header);
	200	if (r)
	201	return r;
	202
2601e83e MT	203	return 0;
	204	}
	205
a7431f1a	206	static int loc_database_read_header(struct loc_database* db, FILE* f) {
2601e83e MT	207	switch (db->version) {
2601e83e MT	208	case 0:
a7431f1a	209	return loc_database_read_header_v0(db, f);
2601e83e MT	210
	211	default:
	212	ERROR(db->ctx, "Incompatible database version: %u\n", db->version);
	213	return 1;
	214	}
	215	}
	216
a7431f1a	217	static int loc_database_read(struct loc_database* db, FILE* f) {
02879100 MT	218	clock_t start = clock();
	219
	220	// Read magic bytes
a7431f1a	221	int r = loc_database_read_magic(db, f);
02879100 MT	222	if (r)
	223	return r;
	224
	225	// Read the header
a7431f1a	226	r = loc_database_read_header(db, f);
02879100 MT	227	if (r)
	228	return r;
	229
	230	clock_t end = clock();
	231
	232	INFO(db->ctx, "Opened database in %.8fs\n",
	233	(double)(end - start) / CLOCKS_PER_SEC);
	234
	235	return 0;
	236	}
	237
c182393f	238	LOC_EXPORT int loc_database_new(struct loc_ctx* ctx, struct loc_database** database, FILE* f) {
a7431f1a MT	239	// Fail on invalid file handle
	240	if (!f)
	241	return -EINVAL;
	242
c182393f MT	243	struct loc_database* db = calloc(1, sizeof(*db));
	244	if (!db)
	245	return -ENOMEM;
	246
	247	// Reference context
	248	db->ctx = loc_ref(ctx);
	249	db->refcount = 1;
	250
	251	DEBUG(db->ctx, "Database object allocated at %p\n", db);
	252
a7431f1a	253	int r = loc_database_read(db, f);
02879100 MT	254	if (r) {
02879100 MT	255	loc_database_unref(db);
2601e83e	256	return r;
02879100	257	}
2601e83e	258
c182393f MT	259	*database = db;
c182393f MT	260
2601e83e	261	return 0;
2601e83e MT	262	}
2601e83e MT	263
c182393f MT	264	LOC_EXPORT struct loc_database* loc_database_ref(struct loc_database* db) {
	265	db->refcount++;
	266
	267	return db;
8f5b676a MT	268	}
8f5b676a MT	269
c182393f	270	static void loc_database_free(struct loc_database* db) {
f10ebc2d MT	271	int r;
f10ebc2d MT	272
c182393f	273	DEBUG(db->ctx, "Releasing database %p\n", db);
c34e76f1	274
c182393f MT	275	// Removing all ASes
c182393f MT	276	if (db->as_v0) {
f10ebc2d	277	r = munmap(db->as_v0, db->as_count * sizeof(*db->as_v0));
c182393f MT	278	if (r)
	279	ERROR(db->ctx, "Could not unmap AS section: %s\n", strerror(errno));
	280	}
c34e76f1	281
f10ebc2d MT	282	// Remove mapped network sections
	283	if (db->networks_v0) {
	284	r = munmap(db->networks_v0, db->networks_count * sizeof(*db->networks_v0));
	285	if (r)
	286	ERROR(db->ctx, "Could not unmap networks section: %s\n", strerror(errno));
	287	}
	288
	289	// Remove mapped network nodes section
	290	if (db->network_nodes_v0) {
	291	r = munmap(db->network_nodes_v0, db->network_nodes_count * sizeof(*db->network_nodes_v0));
	292	if (r)
	293	ERROR(db->ctx, "Could not unmap network nodes section: %s\n", strerror(errno));
	294	}
	295
c182393f	296	loc_stringpool_unref(db->pool);
c34e76f1	297
c182393f MT	298	loc_unref(db->ctx);
c182393f MT	299	free(db);
c34e76f1 MT	300	}
c34e76f1 MT	301
c182393f MT	302	LOC_EXPORT struct loc_database* loc_database_unref(struct loc_database* db) {
	303	if (--db->refcount > 0)
	304	return NULL;
78ace4ed	305
c182393f MT	306	loc_database_free(db);
	307	return NULL;
	308	}
78ace4ed	309
c182393f MT	310	LOC_EXPORT time_t loc_database_created_at(struct loc_database* db) {
	311	return db->created_at;
	312	}
78ace4ed	313
c182393f MT	314	LOC_EXPORT const char* loc_database_get_vendor(struct loc_database* db) {
	315	return loc_stringpool_get(db->pool, db->vendor);
	316	}
78ace4ed	317
c182393f MT	318	LOC_EXPORT const char* loc_database_get_description(struct loc_database* db) {
	319	return loc_stringpool_get(db->pool, db->description);
	320	}
78ace4ed	321
4bf49d00 MT	322	LOC_EXPORT const char* loc_database_get_license(struct loc_database* db) {
	323	return loc_stringpool_get(db->pool, db->license);
	324	}
	325
c182393f MT	326	LOC_EXPORT size_t loc_database_count_as(struct loc_database* db) {
c182393f MT	327	return db->as_count;
78ace4ed MT	328	}
78ace4ed MT	329
c182393f MT	330	// Returns the AS at position pos
	331	static int loc_database_fetch_as(struct loc_database* db, struct loc_as** as, off_t pos) {
	332	if ((size_t)pos >= db->as_count)
	333	return -EINVAL;
2601e83e	334
c182393f	335	DEBUG(db->ctx, "Fetching AS at position %jd\n", pos);
2601e83e MT	336
2601e83e MT	337	int r;
c182393f MT	338	switch (db->version) {
	339	case 0:
	340	r = loc_as_new_from_database_v0(db->ctx, db->pool, as, db->as_v0 + pos);
	341	break;
2601e83e	342
c182393f MT	343	default:
	344	return -1;
	345	}
2601e83e	346
c182393f MT	347	if (r == 0) {
c182393f MT	348	DEBUG(db->ctx, "Got AS%u\n", loc_as_get_number(*as));
2601e83e	349	}
2601e83e	350
c182393f MT	351	return r;
c182393f MT	352	}
c34e76f1	353
c182393f MT	354	// Performs a binary search to find the AS in the list
	355	LOC_EXPORT int loc_database_get_as(struct loc_database* db, struct loc_as** as, uint32_t number) {
	356	off_t lo = 0;
	357	off_t hi = db->as_count - 1;
c34e76f1	358
8f3e2a06 MT	359	// Save start time
	360	clock_t start = clock();
	361
c182393f MT	362	while (lo <= hi) {
c182393f MT	363	off_t i = (lo + hi) / 2;
8f5b676a	364
c182393f MT	365	// Fetch AS in the middle between lo and hi
	366	int r = loc_database_fetch_as(db, as, i);
	367	if (r)
	368	return r;
a5db3e49	369
c182393f MT	370	// Check if this is a match
c182393f MT	371	uint32_t as_number = loc_as_get_number(*as);
8f3e2a06 MT	372	if (as_number == number) {
	373	clock_t end = clock();
	374
	375	// Log how fast this has been
	376	DEBUG(db->ctx, "Found AS%u in %.8fs\n", as_number,
	377	(double)(end - start) / CLOCKS_PER_SEC);
	378
c182393f	379	return 0;
8f3e2a06	380	}
c182393f MT	381
	382	// If it wasn't, we release the AS and
	383	// adjust our search pointers
	384	loc_as_unref(*as);
	385
	386	if (as_number < number) {
	387	lo = i + 1;
	388	} else
	389	hi = i - 1;
	390	}
2601e83e	391
c182393f MT	392	// Nothing found
c182393f MT	393	*as = NULL;
2601e83e	394
8f3e2a06	395	return 1;
2601e83e	396	}
10778041 MT	397
	398	// Returns the network at position pos
	399	static int loc_database_fetch_network(struct loc_database* db, struct loc_network** network, struct in6_addr* address, off_t pos) {
	400	if ((size_t)pos >= db->networks_count)
	401	return -EINVAL;
	402
	403	DEBUG(db->ctx, "Fetching network at position %jd\n", pos);
	404
	405	int r;
	406	switch (db->version) {
	407	case 0:
	408	r = loc_network_new_from_database_v0(db->ctx, network, address, db->networks_v0 + pos);
	409	break;
	410
	411	default:
	412	return -1;
	413	}
	414
	415	if (r == 0) {
	416	char* string = loc_network_str(*network);
	417	DEBUG(db->ctx, "Got network %s\n", string);
	418	free(string);
	419	}
	420
	421	return r;
	422	}
2a30e4de	423
025ef489 MT	424	static int __loc_database_node_is_leaf(const struct loc_database_network_node_v0* node) {
	425	return (node->zero == htobe32(0xffffffff));
	426	}
	427
	428	static int __loc_database_lookup_handle_leaf(struct loc_database* db, const struct in6_addr* address,
2a30e4de MT	429	struct loc_network** network, struct in6_addr* network_address,
2a30e4de MT	430	const struct loc_database_network_node_v0* node) {
025ef489	431	DEBUG(db->ctx, "Handling leaf node at %jd\n", node - db->network_nodes_v0);
2a30e4de MT	432
	433	// Fetch the network
	434	int r = loc_database_fetch_network(db, network,
	435	network_address, be32toh(node->one));
	436	if (r)
	437	return r;
	438
	439	// Check if the given IP address is inside the network
	440	r = loc_network_match_address(*network, address);
	441	if (r) {
	442	DEBUG(db->ctx, "Searched address is not part of the network\n");
	443
	444	loc_network_unref(*network);
	445	*network = NULL;
	446	return 1;
	447	}
	448
	449	// A network was found and the IP address matches
	450	return 0;
	451	}
	452
	453	// Returns the highest result available
	454	static int __loc_database_lookup_max(struct loc_database* db, const struct in6_addr* address,
	455	struct loc_network** network, struct in6_addr* network_address,
f66f15e1	456	const struct loc_database_network_node_v0* node, unsigned int level) {
2a30e4de	457	// If the node is a leaf node, we end here
025ef489 MT	458	if (__loc_database_node_is_leaf(node))
025ef489 MT	459	return __loc_database_lookup_handle_leaf(db, address, network, network_address, node);
2a30e4de	460
025ef489	461	int r;
2a30e4de MT	462	off_t node_index;
	463
	464	// Try to go down the ones path first
	465	if (node->one) {
	466	node_index = be32toh(node->one);
	467	in6_addr_set_bit(network_address, level, 1);
	468
	469	// Check boundaries
	470	if (node_index > 0 && (size_t)node_index <= db->network_nodes_count) {
	471	r = __loc_database_lookup_max(db, address, network, network_address,
	472	db->network_nodes_v0 + node_index, level + 1);
	473
	474	// Abort when match was found or error
	475	if (r <= 0)
	476	return r;
	477	}
	478	}
	479
	480	// ... and if that fails, we try to go down one step on a zero
	481	// branch and then try the ones again...
	482	if (node->zero) {
	483	node_index = be32toh(node->zero);
	484	in6_addr_set_bit(network_address, level, 0);
	485
	486	// Check boundaries
	487	if (node_index > 0 && (size_t)node_index <= db->network_nodes_count) {
	488	r = __loc_database_lookup_max(db, address, network, network_address,
	489	db->network_nodes_v0 + node_index, level + 1);
	490
	491	// Abort when match was found or error
	492	if (r <= 0)
	493	return r;
	494	}
	495	}
	496
	497	// End of path
	498	return 1;
	499	}
	500
	501	// Searches for an exact match along the path
	502	static int __loc_database_lookup(struct loc_database* db, const struct in6_addr* address,
	503	struct loc_network** network, struct in6_addr* network_address,
f66f15e1	504	const struct loc_database_network_node_v0* node, unsigned int level) {
2a30e4de	505	// If the node is a leaf node, we end here
025ef489 MT	506	if (__loc_database_node_is_leaf(node))
025ef489 MT	507	return __loc_database_lookup_handle_leaf(db, address, network, network_address, node);
2a30e4de	508
025ef489	509	int r;
2a30e4de MT	510	off_t node_index;
	511
	512	// Follow the path
	513	int bit = in6_addr_get_bit(address, level);
	514	in6_addr_set_bit(network_address, level, bit);
	515
	516	if (bit == 0)
	517	node_index = be32toh(node->zero);
	518	else
	519	node_index = be32toh(node->one);
	520
	521	// If we point back to root, the path ends here
	522	if (node_index == 0) {
	523	DEBUG(db->ctx, "Tree ends here\n");
	524	return 1;
	525	}
	526
	527	// Check boundaries
	528	if ((size_t)node_index >= db->network_nodes_count)
	529	return -EINVAL;
	530
	531	// Move on to the next node
	532	r = __loc_database_lookup(db, address, network, network_address,
	533	db->network_nodes_v0 + node_index, level + 1);
	534
	535	// End here if a result was found
	536	if (r == 0)
	537	return r;
	538
	539	// Raise any errors
	540	else if (r < 0)
	541	return r;
	542
	543	DEBUG(db->ctx, "Could not find an exact match at %u\n", level);
	544
	545	// If nothing was found, we have to search for an inexact match
	546	return __loc_database_lookup_max(db, address, network, network_address, node, level);
	547	}
	548
	549	LOC_EXPORT int loc_database_lookup(struct loc_database* db,
	550	struct in6_addr* address, struct loc_network** network) {
	551	struct in6_addr network_address;
	552	memset(&network_address, 0, sizeof(network_address));
	553
	554	*network = NULL;
	555
	556	// Save start time
	557	clock_t start = clock();
	558
	559	int r = __loc_database_lookup(db, address, network, &network_address,
	560	db->network_nodes_v0, 0);
	561
	562	clock_t end = clock();
	563
	564	// Log how fast this has been
	565	DEBUG(db->ctx, "Executed network search in %.8fs\n",
	566	(double)(end - start) / CLOCKS_PER_SEC);
	567
	568	return r;
	569	}
	570
	571	LOC_EXPORT int loc_database_lookup_from_string(struct loc_database* db,
	572	const char* string, struct loc_network** network) {
	573	struct in6_addr address;
574
575	int r = loc_parse_address(db->ctx, string, &address);
576	if (r)
577	return r;
578
579	return loc_database_lookup(db, &address, network);
580	}