geoip: native maxmind .mmdb reader, G: geoban extban, GEOIP command, whois country

echoircd.conf.example

@@ -117,3 +117,7 @@ amu_target    = both
 # hidewhois_hide_server = yes  # hide 312
 # hidewhois_hide_idle = yes    # hide 317
 # hidewhois_hide_secure = yes  # hide 671
+# --- geoip (m_geo_maxmind): native MaxMind .mmdb country lookup. Enables the
+#     G:<cc> ban extban (e.g. +b G:CN,RU), the oper GEOIP <nick|ip> command and
+#     a country line in WHOIS (opers). Point at a GeoLite2-Country.mmdb file:
+# geoip_database = /etc/echoircd/GeoLite2-Country.mmdb

src/channels.rs

@@ -921,6 +921,7 @@ impl Server {
                     Some(b'r') => crate::modules::realnameban::matches(self, uid, &b.mask[2..]),
                     Some(b'j') => crate::modules::channelban::matches(self, uid, &b.mask[2..]),
                     Some(b's') => crate::modules::serverban::matches(self, uid, &b.mask[2..]),
+                    Some(b'G') => crate::modules::geoip::geoban_match(self, uid, &b.mask[2..]),
                     _ => false,
                 }
             } else {
@@ -1132,7 +1133,7 @@ pub fn normalize_ban_mask(m: &str) -> String {
         // These extbans carry a name / spec / channel / server, not a host mask,
         // so they must not be host-normalised: g: (security group), y: (reputation
         // score), r: (realname), j: (channel), s: (server name).
-        if matches!(b[0], b'g' | b'y' | b'r' | b'j' | b's') {
+        if matches!(b[0], b'g' | b'y' | b'r' | b'j' | b's' | b'G') {
             return m.to_string();
         }
         return format!("{}:{}", &m[..1], normalize_mask(&m[2..]));

src/coremods/core_info.rs

@@ -262,6 +262,10 @@ impl Command for Whois {
             if let Some(line) = crate::modules::whoisport::line(s, tuid) {
                 s.numeric(uid, RPL_WHOISSPECIAL, &format!(":{line}"));
             }
+            // geoip: the country the user connects from — opers only
+            if let Some(line) = crate::modules::geoip::whois_line(s, tuid) {
+                s.numeric(uid, RPL_WHOISSPECIAL, &format!(":{line}"));
+            }
         }
         // opers can see through the cloak to the real host/ip
         if asker_oper && disp != realhost {

src/ircd.rs

@@ -95,6 +95,7 @@ impl Ircd {
         server.load_xlines(); // restore persisted bans (m_xline_db)
         crate::modules::metadata::load(&mut server); // restore channel metadata (m_metadata_db)
         crate::modules::reputation::load(&mut server); // restore per-IP reputation
+        crate::modules::geoip::init(&mut server); // load the GeoIP database (m_geo_maxmind)
         Ircd {
             server,
             commands: command_table(),

src/modules/geoip.rs

@@ -0,0 +1,392 @@
+//! geoip — native MaxMind DB (`.mmdb`) country lookup, with the `G:<cc>` geoban
+//! extban, the `GEOIP` command and a WHOIS country line. The `maxminddb` crate is
+//! off-limits (openssl+mio only), so the binary format is parsed by hand in pure
+//! std: the metadata section, the record-size-aware search tree, and the typed data
+//! decoder — no crate, no `unsafe`, no C FFI.
+//!
+//! Config: `geoip_database = /path/to/GeoLite2-Country.mmdb` (loaded once at boot).
+//!
+//! Behaviour reference: InspIRCd's `m_geo_maxmind` + `m_geoban` + `m_geocmd`.
+//! Original native Rust.
+
+use std::net::IpAddr;
+use std::sync::Arc;
+
+use crate::command::{CmdResult, Command};
+use crate::numeric::ERR_NOPRIVILEGES;
+use crate::server::Server;
+use crate::Uid;
+
+const MARKER: &[u8] = b"\xab\xcd\xefMaxMind.com";
+const SEPARATOR: usize = 16;
+
+/// A loaded MaxMind DB, cached in `Server.ext`.
+pub struct GeoDb(pub Arc<Mmdb>);
+
+/// A parsed `.mmdb` file: the raw bytes plus the tree geometry from its metadata.
+pub struct Mmdb {
+    data: Vec<u8>,
+    node_count: usize,
+    node_byte_size: usize, // record_size * 2 / 8
+    record_size: u32,
+    data_start: usize, // node_count*node_byte_size + 16 (first byte after the separator)
+    ip_version: u16,
+}
+
+/// Big-endian value of up to 8 bytes.
+fn be(bytes: &[u8]) -> u64 {
+    bytes.iter().fold(0u64, |acc, &b| (acc << 8) | b as u64)
+}
+
+/// A cursor over the data section for pointer-aware decoding. `base` is the offset
+/// pointers are relative to (the start of the section being decoded).
+struct Decoder<'a> {
+    data: &'a [u8],
+    base: usize,
+}
+
+impl<'a> Decoder<'a> {
+    /// Parse the control byte at `off`: returns `(type, size, payload_offset)`.
+    /// Not used for pointers (they encode size differently — see `resolve`).
+    fn control(&self, off: usize) -> Option<(u8, usize, usize)> {
+        let b = *self.data.get(off)?;
+        let mut typ = b >> 5;
+        let mut o = off + 1;
+        if typ == 0 {
+            typ = 7 + *self.data.get(o)?;
+            o += 1;
+        }
+        let mut size = (b & 0x1f) as usize;
+        if size >= 29 {
+            match size {
+                29 => {
+                    size = 29 + *self.data.get(o)? as usize;
+                    o += 1;
+                }
+                30 => {
+                    size = 285 + be(self.data.get(o..o + 2)?) as usize;
+                    o += 2;
+                }
+                _ => {
+                    size = 65821 + be(self.data.get(o..o + 3)?) as usize;
+                    o += 3;
+                }
+            }
+        }
+        Some((typ, size, o))
+    }
+
+    /// Follow pointer(s) at `off` to the concrete value offset (non-pointers pass
+    /// through). Depth-limited against malformed data.
+    fn resolve(&self, off: usize, depth: u8) -> Option<usize> {
+        if depth > 8 {
+            return None;
+        }
+        let b = *self.data.get(off)?;
+        if b >> 5 != 1 {
+            return Some(off);
+        }
+        let ps = ((b >> 3) & 0x3) as usize;
+        let v0 = (b & 0x7) as usize;
+        let ptr = match ps {
+            0 => (v0 << 8) | *self.data.get(off + 1)? as usize,
+            1 => ((v0 << 16) | be(self.data.get(off + 1..off + 3)?) as usize) + 2048,
+            2 => ((v0 << 24) | be(self.data.get(off + 1..off + 4)?) as usize) + 526336,
+            _ => be(self.data.get(off + 1..off + 5)?) as usize,
+        };
+        self.resolve(self.base + ptr, depth + 1)
+    }
+
+    /// The number of bytes the value at `off` occupies (pointers are 2–5 bytes; not
+    /// followed). Recurses into maps/arrays. Returns 0 on malformed input.
+    fn value_len(&self, off: usize) -> usize {
+        let b = match self.data.get(off) {
+            Some(&b) => b,
+            None => return 0,
+        };
+        if b >> 5 == 1 {
+            return 1 + ((b >> 3) & 0x3) as usize + 1; // 2..=5 bytes
+        }
+        let Some((typ, size, payload)) = self.control(off) else {
+            return 0;
+        };
+        let header = payload - off;
+        match typ {
+            7 => {
+                // map: `size` key/value pairs
+                let mut cur = payload;
+                for _ in 0..size {
+                    cur += self.value_len(cur); // key
+                    cur += self.value_len(cur); // value
+                }
+                cur - off
+            }
+            11 => {
+                // array: `size` elements
+                let mut cur = payload;
+                for _ in 0..size {
+                    cur += self.value_len(cur);
+                }
+                cur - off
+            }
+            14 => header,          // bool: value is in the size field, no payload
+            _ => header + size,    // string/bytes/ints/float/double
+        }
+    }
+
+    /// Read a UTF-8 string at `off` (following pointers).
+    fn string(&self, off: usize) -> Option<String> {
+        let off = self.resolve(off, 0)?;
+        let (typ, size, payload) = self.control(off)?;
+        if typ != 2 {
+            return None;
+        }
+        std::str::from_utf8(self.data.get(payload..payload + size)?)
+            .ok()
+            .map(str::to_string)
+    }
+
+    /// Read an unsigned integer at `off` (following pointers).
+    fn uint(&self, off: usize) -> Option<u64> {
+        let off = self.resolve(off, 0)?;
+        let (typ, size, payload) = self.control(off)?;
+        if !matches!(typ, 5 | 6 | 9 | 10) {
+            return None;
+        }
+        Some(be(self.data.get(payload..payload + size)?))
+    }
+
+    /// The value offset for `key` in the map at `off` (following pointers).
+    fn map_get(&self, off: usize, key: &str) -> Option<usize> {
+        let off = self.resolve(off, 0)?;
+        let (typ, size, payload) = self.control(off)?;
+        if typ != 7 {
+            return None;
+        }
+        let mut cur = payload;
+        for _ in 0..size {
+            let k = self.string(cur)?;
+            cur += self.value_len(cur);
+            if k == key {
+                return Some(cur);
+            }
+            cur += self.value_len(cur);
+        }
+        None
+    }
+}
+
+impl Mmdb {
+    /// Load and parse an `.mmdb` file.
+    pub fn open(path: &str) -> Option<Mmdb> {
+        let data = std::fs::read(path).ok()?;
+        let marker = data.windows(MARKER.len()).rposition(|w| w == MARKER)?;
+        let meta = marker + MARKER.len();
+        let d = Decoder { data: &data, base: meta };
+        let node_count = d.uint(d.map_get(meta, "node_count")?)? as usize;
+        let record_size = d.uint(d.map_get(meta, "record_size")?)? as u32;
+        let ip_version = d.uint(d.map_get(meta, "ip_version")?)? as u16;
+        if !matches!(record_size, 24 | 28 | 32) || node_count == 0 {
+            return None;
+        }
+        let node_byte_size = record_size as usize * 2 / 8;
+        let data_start = node_count * node_byte_size + SEPARATOR;
+        Some(Mmdb {
+            data,
+            node_count,
+            node_byte_size,
+            record_size,
+            data_start,
+            ip_version,
+        })
+    }
+
+    /// The left (bit=false) or right (bit=true) record of tree `node`.
+    fn record(&self, node: usize, bit: bool) -> Option<usize> {
+        let base = node * self.node_byte_size;
+        match self.record_size {
+            24 => {
+                let o = base + if bit { 3 } else { 0 };
+                Some(be(self.data.get(o..o + 3)?) as usize)
+            }
+            28 => {
+                let mid = *self.data.get(base + 3)?;
+                if bit {
+                    Some(((mid as usize & 0x0f) << 24) | be(self.data.get(base + 4..base + 7)?) as usize)
+                } else {
+                    Some(((mid as usize >> 4) << 24) | be(self.data.get(base..base + 3)?) as usize)
+                }
+            }
+            _ => {
+                let o = base + if bit { 4 } else { 0 };
+                Some(be(self.data.get(o..o + 4)?) as usize)
+            }
+        }
+    }
+
+    /// The 2-letter ISO country code for `ip`, if the database has one.
+    pub fn country(&self, ip: IpAddr) -> Option<String> {
+        // build the bit path; IPv4 in an IPv6 db is prefixed with 96 zero bits
+        let mut bits: Vec<bool> = Vec::with_capacity(128);
+        match ip {
+            IpAddr::V4(a) => {
+                if self.ip_version == 6 {
+                    bits.extend(std::iter::repeat(false).take(96));
+                }
+                for byte in a.octets() {
+                    for i in (0..8).rev() {
+                        bits.push((byte >> i) & 1 == 1);
+                    }
+                }
+            }
+            IpAddr::V6(a) => {
+                for byte in a.octets() {
+                    for i in (0..8).rev() {
+                        bits.push((byte >> i) & 1 == 1);
+                    }
+                }
+            }
+        }
+        let mut node = 0usize;
+        for bit in bits {
+            if node >= self.node_count {
+                return None;
+            }
+            let rec = self.record(node, bit)?;
+            if rec == self.node_count {
+                return None; // no data
+            }
+            if rec > self.node_count {
+                // data pointer: abs = tree_size + (rec - node_count)
+                let abs = (self.data_start - SEPARATOR) + rec - self.node_count;
+                let d = Decoder {
+                    data: &self.data,
+                    base: self.data_start,
+                };
+                let country = d.map_get(abs, "country")?;
+                return d.string(d.map_get(country, "iso_code")?);
+            }
+            node = rec;
+        }
+        None
+    }
+}
+
+/// Load the configured database into `Server.ext` at boot. Called from `Ircd::new`.
+pub fn init(s: &mut Server) {
+    let Some(path) = s.conf("geoip_database").map(str::to_string) else {
+        return;
+    };
+    match Mmdb::open(&path) {
+        Some(db) => {
+            s.ext.set(GeoDb(Arc::new(db)));
+            eprintln!("echoircd: loaded GeoIP database {path}");
+        }
+        None => eprintln!("echoircd: could not read GeoIP database {path}"),
+    }
+}
+
+/// The ISO country code of `ip` per the loaded database, uppercased.
+pub fn country_of(s: &Server, ip: IpAddr) -> Option<String> {
+    s.ext
+        .get::<GeoDb>()
+        .and_then(|db| db.0.country(ip))
+        .map(|c| c.to_ascii_uppercase())
+}
+
+/// The `G:<cc>` geoban match: does `uid`'s country equal (case-insensitively) the
+/// country code in the extban? Dispatched from `Server::ban_list_hit`.
+pub fn geoban_match(s: &Server, uid: Uid, spec: &str) -> bool {
+    let Some(ip) = s.users.get(&uid).map(|u| u.addr.ip()) else {
+        return false;
+    };
+    match country_of(s, ip) {
+        Some(cc) => spec
+            .split(',')
+            .any(|want| want.trim().eq_ignore_ascii_case(&cc)),
+        None => false,
+    }
+}
+
+/// A WHOIS line (opers only) showing the target's country.
+pub fn whois_line(s: &Server, tuid: Uid) -> Option<String> {
+    let ip = s.users.get(&tuid).map(|u| u.addr.ip())?;
+    country_of(s, ip).map(|cc| format!("is connecting from country {cc}"))
+}
+
+pub fn commands() -> Vec<Box<dyn Command>> {
+    vec![Box::new(GeoIpCmd)]
+}
+
+/// GEOIP `<nick|ip>` — oper command reporting the country of a user or raw IP.
+struct GeoIpCmd;
+impl Command for GeoIpCmd {
+    fn name(&self) -> &'static str {
+        "GEOIP"
+    }
+    fn min_params(&self) -> usize {
+        1
+    }
+    fn handle(&self, s: &mut Server, uid: Uid, params: &[String]) -> CmdResult {
+        if !s.is_oper(uid) {
+            s.numeric(
+                uid,
+                ERR_NOPRIVILEGES,
+                ":Permission Denied- You're not an IRC operator",
+            );
+            return CmdResult::Fail;
+        }
+        let target = &params[0];
+        let ip = s
+            .find_nick(target)
+            .and_then(|t| s.users.get(&t))
+            .map(|u| u.addr.ip())
+            .or_else(|| target.parse::<IpAddr>().ok());
+        let nick = s.users.get(&uid).map(|u| u.nick.clone()).unwrap_or_default();
+        let msg = match ip {
+            None => format!("GEOIP: no such nick, and {target} is not an IP"),
+            Some(ip) => match country_of(s, ip) {
+                Some(cc) => format!("GEOIP: {target} ({ip}) is in country {cc}"),
+                None => format!("GEOIP: no country found for {target} ({ip})"),
+            },
+        };
+        s.send(uid, format!(":{} NOTICE {nick} :*** {msg}", s.name));
+        CmdResult::Ok
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use std::net::Ipv4Addr;
+
+    // A real GeoLite2-Country.mmdb if one is present; otherwise the test no-ops so
+    // CI (which has no database) stays green.
+    const DB_CANDIDATES: &[&str] = &[
+        "/home/debian/irc/ircd/inspircd/run/conf/geodata/GeoLite2-Country.mmdb",
+        "/usr/share/GeoIP/GeoLite2-Country.mmdb",
+    ];
+
+    fn load() -> Option<Mmdb> {
+        DB_CANDIDATES.iter().find_map(|p| Mmdb::open(p))
+    }
+
+    #[test]
+    fn known_ips_resolve() {
+        let Some(db) = load() else {
+            return;
+        };
+        // 8.8.8.8 (Google DNS) is US in every GeoLite2 vintage.
+        assert_eq!(
+            db.country(IpAddr::V4(Ipv4Addr::new(8, 8, 8, 8))).as_deref(),
+            Some("US")
+        );
+        // A private address has no country record.
+        assert_eq!(db.country(IpAddr::V4(Ipv4Addr::new(10, 0, 0, 1))), None);
+        // IPv6 traversal (Google public DNS) also resolves to US.
+        assert_eq!(
+            db.country("2001:4860:4860::8888".parse().unwrap()).as_deref(),
+            Some("US")
+        );
+    }
+}

src/modules/mod.rs

@@ -23,6 +23,7 @@ pub mod extjwt;
 pub mod filehost;
 pub mod filter;
 pub mod flood;
+pub mod geoip;
 pub mod hashident;
 pub mod hidewhois;
 pub mod irccloudtags;
@@ -113,5 +114,6 @@ pub fn module_commands() -> Vec<Box<dyn Command>> {
         .chain(tline::commands())
         .chain(rmode::commands())
         .chain(customtitle::commands())
+        .chain(geoip::commands())
         .collect()
 }

src/server.rs

@@ -585,7 +585,7 @@ impl Server {
     /// burst and the `ISUPPORT` command (draft/extended-isupport).
     pub fn isupport_lines(&self) -> Vec<String> {
         let mut lines = vec![format!(
-            "CHANTYPES=# PREFIX=(qaohv)~&@%+ CHANMODES=beIgX,k,lfjFLHBJdK,ACDGMNOPQRSTUcimnpstuz EXTBAN=,cgjmnrsy WATCH=128 MONITOR=128 SILENCE=32 CALLERID=g WHOX CHATHISTORY=256 MSGREFTYPES=timestamp,msgid UTF8ONLY CASEMAPPING=ascii NICKLEN=30 CHANNELLEN=50 NETWORK={}",
+            "CHANTYPES=# PREFIX=(qaohv)~&@%+ CHANMODES=beIgX,k,lfjFLHBJdK,ACDGMNOPQRSTUcimnpstuz EXTBAN=,Gcgjmnrsy WATCH=128 MONITOR=128 SILENCE=32 CALLERID=g WHOX CHATHISTORY=256 MSGREFTYPES=timestamp,msgid UTF8ONLY CASEMAPPING=ascii NICKLEN=30 CHANNELLEN=50 NETWORK={}",
             self.network
         )];
         if let Some(tok) = crate::modules::network_icon::isupport(self) {