2 releases

0.1.1 Jul 2, 2024
0.1.0 Mar 5, 2024

#493 in Database interfaces

MIT/Apache

38KB
667 lines

Read location databases in libloc format.

It features constant-time lookups of IPv6/IPv4 addresses to ASN and country data ("GeoIP"). The database also contains names of AS and countries.

The main struct of this crate is Locations. First, you'll need a database. A recent version of the official IPFire database can be obtained from https://location.ipfire.org/databases/1/location.db.xz, xz-compressed. Then you can start from Locations::open and use Locations::lookup to look up information on a particular IP address.

A short overview about the internals of the database format can be found at https://www.ipfire.org/blog/libloc-or-what-is-working-inside-it. A Kaitai struct implementation ipfire_libloc_db_v1.ksy can be found in this repository.

Examples

use libloc::Locations;

let locations = Locations::open("example-location.db")?;

let network: libloc::Network = locations.lookup("2a07:1c44:5800::1".parse().unwrap()).unwrap();
assert_eq!(network.country_code(), "DE");
assert_eq!(network.asn(), 204867);
assert_eq!(network.is_anonymous_proxy(), false);
assert_eq!(network.addrs().to_string(), "2a07:1c44:5800::/40");

let country: libloc::Country = locations.country("DE").unwrap();
assert_eq!(country.continent_code(), "EU");
assert_eq!(country.name(), "Germany");

let as_: libloc::As = locations.as_(204867).unwrap();
assert_eq!(as_.name(), "Lightning Wire Labs GmbH");

# Ok::<(), libloc::OpenError>(())

Panics

Any function from this library might panic if the database is corrupt.

Benches

This library was written for fun. It's still a lot faster than the original libloc, even though it wasn't optimized for speed.

     Running benches/locations.rs

running 4 tests
test lookup    ... bench:          87 ns/iter (+/- 15)
test lookup_v4 ... bench:          61 ns/iter (+/- 5)
test lookup_v6 ... bench:         396 ns/iter (+/- 22)
test open      ... bench:      64,012 ns/iter (+/- 4,592)

test result: ok. 0 passed; 0 failed; 0 ignored; 4 measured

     Running benches/native.rs

running 2 tests
test lookup_v6 ... bench:       1,019 ns/iter (+/- 128)
test open      ... bench:     234,420 ns/iter (+/- 19,638)

test result: ok. 0 passed; 0 failed; 0 ignored; 2 measured

It's looking up IPv6 addresses roughly 2.5x as fast, it has special handling for IPv4 addresses which makes it >10x as fast.

Dependencies

~1–1.6MB
~22K SLoC