//! A crate for performing GeoIP lookups using the Tor GeoIP database.

// @@ begin lint list maintained by maint/add_warning @@

#![allow(renamed_and_removed_lints)] // @@REMOVE_WHEN(ci_arti_stable)

#![allow(unknown_lints)] // @@REMOVE_WHEN(ci_arti_nightly)

#![warn(missing_docs)]

#![warn(noop_method_call)]

#![warn(unreachable_pub)]

#![warn(clippy::all)]

#![deny(clippy::await_holding_lock)]

#![deny(clippy::cargo_common_metadata)]

#![deny(clippy::cast_lossless)]

#![deny(clippy::checked_conversions)]

#![warn(clippy::cognitive_complexity)]

#![deny(clippy::debug_assert_with_mut_call)]

#![deny(clippy::exhaustive_enums)]

#![deny(clippy::exhaustive_structs)]

#![deny(clippy::expl_impl_clone_on_copy)]

#![deny(clippy::fallible_impl_from)]

#![deny(clippy::implicit_clone)]

#![deny(clippy::large_stack_arrays)]

#![warn(clippy::manual_ok_or)]

#![deny(clippy::missing_docs_in_private_items)]

#![warn(clippy::needless_borrow)]

#![warn(clippy::needless_pass_by_value)]

#![warn(clippy::option_option)]

#![deny(clippy::print_stderr)]

#![deny(clippy::print_stdout)]

#![warn(clippy::rc_buffer)]

#![deny(clippy::ref_option_ref)]

#![warn(clippy::semicolon_if_nothing_returned)]

#![warn(clippy::trait_duplication_in_bounds)]

#![deny(clippy::unchecked_duration_subtraction)]

#![deny(clippy::unnecessary_wraps)]

#![warn(clippy::unseparated_literal_suffix)]

#![deny(clippy::unwrap_used)]

#![deny(clippy::mod_module_files)]

#![allow(clippy::let_unit_value)] // This can reasonably be done for explicitness

#![allow(clippy::uninlined_format_args)]

#![allow(clippy::significant_drop_in_scrutinee)] // arti/-/merge_requests/588/#note_2812945

#![allow(clippy::result_large_err)] // temporary workaround for arti#587

#![allow(clippy::needless_raw_string_hashes)] // complained-about code is fine, often best

#![allow(clippy::needless_lifetimes)] // See arti#1765

//! <!-- @@ end lint list maintained by maint/add_warning @@ -->

// TODO #1645 (either remove this, or decide to have it everywhere)

#![cfg_attr(not(all(feature = "full")), allow(unused))]

pub use crate::err::Error;

use rangemap::RangeInclusiveMap;

use std::fmt::{Debug, Display, Formatter};

use std::net::{IpAddr, Ipv6Addr};

use std::num::{NonZeroU32, NonZeroU8, TryFromIntError};

use std::str::FromStr;

use std::sync::{Arc, OnceLock};

mod err;

/// An embedded copy of the latest geoip v4 database at the time of compilation.

///

/// FIXME(eta): This does use a few megabytes of binary size, which is less than ideal.

///             It would be better to parse it at compile time or something.

#[cfg(feature = "embedded-db")]

static EMBEDDED_DB_V4: &str = include_str!("../data/geoip");

/// An embedded copy of the latest geoip v6 database at the time of compilation.

#[cfg(feature = "embedded-db")]

static EMBEDDED_DB_V6: &str = include_str!("../data/geoip6");

/// A parsed copy of the embedded database.

#[cfg(feature = "embedded-db")]

static EMBEDDED_DB_PARSED: OnceLock<Arc<GeoipDb>> = OnceLock::new();

/// A two-letter country code.

///

/// Specifically, this type represents a purported "ISO 3166-1 alpha-2" country

/// code, such as "IT" for Italy or "UY" for Uruguay.

///

/// It does not include the sentinel value `??` that we use to represent

/// "country unknown"; if you need that, use [`OptionCc`]. Other than that, we

/// do not check whether the country code represents a real country: we only

/// ensure that it is a pair of printing ASCII characters.

///

/// Note that the geoip databases included with Arti will only include real

/// countries; we do not include the pseudo-countries `A1` through `An` for

/// "anonymous proxies", since doing so would mean putting nearly all Tor relays

/// into one of those countries.

#[derive(Copy, Clone, Eq, PartialEq)]

pub struct CountryCode {

    /// The underlying value (two printable ASCII characters, stored uppercase).

    ///

    /// The special value `??` is excluded, since it is not a country; use

    /// `OptionCc` instead if you need to represent that.

    ///

    /// We store these as `NonZeroU8` so that an `Option<CountryCode>` only has to

    /// take 2 bytes. This helps with alignment and storage.

    inner: [NonZeroU8; 2],

}

impl CountryCode {

    /// Make a new `CountryCode`.

        /// Try to convert an array of 2 bytes into an array of 2 nonzero bytes.

        #[inline]

        })

    /// Get the actual country code.

    ///

    /// This just calls `.as_ref()`.

}

impl Display for CountryCode {

}

impl Debug for CountryCode {

}

impl AsRef<str> for CountryCode {

        /// Convert a reference to an array of 2 nonzero bytes to a reference to

        /// an array of 2 bytes.

        #[inline]

        // This shouldn't ever panic, since we shouldn't feed non-utf8 country

        // codes in.

        //

        // In theory we could use from_utf8_unchecked, but that's probably not

        // needed.

}

impl FromStr for CountryCode {

    type Err = Error;

}

/// Wrapper for an `Option<`[`CountryCode`]`>` that encodes `None` as `??`.

///

/// Used so that we can implement foreign traits.

#[derive(

    Copy, Clone, Debug, Eq, PartialEq, derive_more::Into, derive_more::From, derive_more::AsRef,

)]

#[allow(clippy::exhaustive_structs)]

pub struct OptionCc(pub Option<CountryCode>);

impl FromStr for OptionCc {

    type Err = Error;

        }

}

impl Display for OptionCc {

        }

}

/// A country code / ASN definition.

///

/// Type lifted from `geoip-db-tool` in the C-tor source.

#[derive(Copy, Clone, Eq, PartialEq, Debug)]

struct NetDefn {

    /// The country code.

    ///

    /// We translate the value "??" into None.

    cc: Option<CountryCode>,

    /// The ASN, if we have one. We translate the value "0" into None.

    asn: Option<NonZeroU32>,

}

impl NetDefn {

    /// Make a new `NetDefn`.

    /// Return the country code.

    /// Return the ASN, if there is one.

}

/// A database of IP addresses to country codes.

#[derive(Clone, Eq, PartialEq, Debug)]

pub struct GeoipDb {

    /// The IPv4 subset of the database, with v4 addresses stored as 32-bit integers.

    map_v4: RangeInclusiveMap<u32, NetDefn>,

    /// The IPv6 subset of the database, with v6 addresses stored as 128-bit integers.

    map_v6: RangeInclusiveMap<u128, NetDefn>,

}

impl GeoipDb {

    /// Make a new `GeoipDb` using a compiled-in copy of the GeoIP database.

    ///

    /// The returned instance of the database is shared with `Arc` across all invocations of this

    /// function in the same program.

    #[cfg(feature = "embedded-db")]

    /// Make a new `GeoipDb` using provided copies of the v4 and v6 database, in Tor legacy format.

        }

        // This is slightly copypasta, but probably less readable to merge into one thing.

        }

    /// Get the `NetDefn` for an IP address.

        }

    /// Get a 2-letter country code for the given IP address, if this data is available.

    /// Determine a 2-letter country code for a host with multiple IP addresses.

    ///

    /// This looks up all of the IP addresses with `lookup_country_code`. If the lookups

    /// return different countries, `None` is returned. IP addresses that fail to resolve

    /// into a country are ignored if some of the other addresses do resolve successfully.

                // If we already have a return value and it's different, then return None;

                // a server can't be in two different countries.

        }

    /// Return the ASN the IP address is in, if this data is available.

}

/// A (representation of a) host on the network which may have a known country code.

pub trait HasCountryCode {

    /// Return the country code in which this server is most likely located.

    ///

    /// This is usually implemented by simple GeoIP lookup on the addresses provided by `HasAddrs`.

    /// It follows that the server might not actually be in the returned country, but this is a

    /// halfway decent estimate for what other servers might guess the server's location to be

    /// (and thus useful for e.g. getting around simple geo-blocks, or having webpages return

    /// the correct localised versions).

    ///

    /// Returning `None` signifies that no country code information is available. (Conflicting

    /// GeoIP lookup results might also cause `None` to be returned.)

    fn country_code(&self) -> Option<CountryCode>;

}

#[cfg(test)]

mod test {

    // @@ begin test lint list maintained by maint/add_warning @@

    #![allow(clippy::bool_assert_comparison)]

    #![allow(clippy::clone_on_copy)]

    #![allow(clippy::dbg_macro)]

    #![allow(clippy::mixed_attributes_style)]

    #![allow(clippy::print_stderr)]

    #![allow(clippy::print_stdout)]

    #![allow(clippy::single_char_pattern)]

    #![allow(clippy::unwrap_used)]

    #![allow(clippy::unchecked_duration_subtraction)]

    #![allow(clippy::useless_vec)]

    #![allow(clippy::needless_pass_by_value)]

    //! <!-- @@ end test lint list maintained by maint/add_warning @@ -->

    use super::*;

    use std::net::Ipv4Addr;

    // NOTE(eta): this test takes a whole 1.6 seconds in *non-release* mode

    #[test]

    #[cfg(feature = "embedded-db")]

    fn embedded_db() {

        let db = GeoipDb::new_embedded();

        assert_eq!(

            db.lookup_country_code(Ipv4Addr::new(8, 8, 8, 8).into())

                .map(|x| x.as_ref()),

            Some("US")

        );

        assert_eq!(

            db.lookup_country_code("2001:4860:4860::8888".parse().unwrap())

                .map(|x| x.as_ref()),

            Some("US")

        );

    }

    #[test]

    fn basic_lookups() {

        let src_v4 = r#"

        16909056,16909311,GB

        "#;

        let src_v6 = r#"

        fe80::,fe81::,US

        dead:beef::,dead:ffff::,??

        "#;

        let db = GeoipDb::new_from_legacy_format(src_v4, src_v6).unwrap();

        assert_eq!(

            db.lookup_country_code(Ipv4Addr::new(1, 2, 3, 4).into())

                .map(|x| x.as_ref()),

            Some("GB")

        );

        assert_eq!(

            db.lookup_country_code(Ipv4Addr::new(1, 1, 1, 1).into()),

            None

        );

        assert_eq!(

            db.lookup_country_code("fe80::dead:beef".parse().unwrap())

                .map(|x| x.as_ref()),

            Some("US")

        );

        assert_eq!(

            db.lookup_country_code("fe81::dead:beef".parse().unwrap()),

            None

        );

        assert_eq!(

            db.lookup_country_code("dead:beef::1".parse().unwrap()),

            None

        );

    }

    #[test]

    fn cc_parse() -> Result<(), Error> {

        // real countries.

        assert_eq!(CountryCode::from_str("us")?, CountryCode::from_str("US")?);

        assert_eq!(CountryCode::from_str("UY")?, CountryCode::from_str("UY")?);

        // not real as of this writing, but still representable.

        assert_eq!(CountryCode::from_str("A7")?, CountryCode::from_str("a7")?);

        assert_eq!(CountryCode::from_str("xz")?, CountryCode::from_str("xz")?);

        // Can't convert to two bytes.

        assert!(matches!(

            CountryCode::from_str("z"),

            Err(Error::BadCountryCode(_))

        ));

        assert!(matches!(

            CountryCode::from_str("🐻‍❄️"),

            Err(Error::BadCountryCode(_))

        ));

        assert!(matches!(

            CountryCode::from_str("Sheboygan"),

            Err(Error::BadCountryCode(_))

        ));

        // Can convert to two bytes, but still not printable ascii

        assert!(matches!(

            CountryCode::from_str("\r\n"),

            Err(Error::BadCountryCode(_))

        ));

        assert!(matches!(

            CountryCode::from_str("\0\0"),

            Err(Error::BadCountryCode(_))

        ));

        assert!(matches!(

            CountryCode::from_str("¡"),

            Err(Error::BadCountryCode(_))

        ));

        // Not a country.

        assert!(matches!(

            CountryCode::from_str("??"),

            Err(Error::NowhereNotSupported)

        ));

        Ok(())

    }

    #[test]

    fn opt_cc_parse() -> Result<(), Error> {

        assert_eq!(

            CountryCode::from_str("br")?,

            OptionCc::from_str("BR")?.0.unwrap()

        );

        assert!(OptionCc::from_str("??")?.0.is_none());

        Ok(())

    }

}