whois_rust/

who_is.rs

use std::{
    collections::HashMap,
    fs::File,
    io::{self, Read, Write},
    net::{SocketAddr, TcpStream, ToSocketAddrs},
    path::Path,
    str::FromStr,
    sync::LazyLock,
    time::Duration,
};

use hickory_client::{
    client::{Client, SyncClient},
    op::DnsResponse,
    rr::{DNSClass, Name, RData, Record, RecordType},
    udp::UdpClientConnection,
};
use regex::Regex;
use serde_json::{Map, Value};
#[cfg(feature = "tokio")]
use tokio::io::{AsyncReadExt, AsyncWriteExt};
use validators::models::Host;

use crate::{WhoIsError, WhoIsLookupOptions, WhoIsServerValue};

const DEFAULT_WHOIS_HOST_PORT: u16 = 43;
const DEFAULT_WHOIS_HOST_QUERY: &str = "$addr\r\n";

static RE_SERVER: LazyLock<Regex> = LazyLock::new(|| {
    Regex::new(r"(ReferralServer|Registrar Whois|Whois Server|WHOIS Server|Registrar WHOIS Server):[^\S\n]*(r?whois://)?(.*)").unwrap()
});

/// Extract the referral WHOIS host from a query result. The captured value is trimmed because responses usually end lines with CRLF, and a trailing `\r` would make the host unparsable.
fn extract_referral_host(query_result: &str) -> Option<&str> {
    let host = RE_SERVER.captures(query_result)?.get(3)?.as_str().trim();

    if host.is_empty() { None } else { Some(host) }
}

/// Decode the raw bytes of a WHOIS response into a `String`. Valid UTF-8 always passes through unchanged. When the `charset` feature is disabled, invalid UTF-8 falls back to a lossy conversion (malformed bytes become `U+FFFD`).
#[cfg(not(feature = "charset"))]
fn decode_response(bytes: &[u8]) -> String {
    match std::str::from_utf8(bytes) {
        Ok(s) => s.to_string(),
        Err(_) => String::from_utf8_lossy(bytes).into_owned(),
    }
}

/// Decode the raw bytes of a WHOIS response into a `String`. Valid UTF-8 always passes through unchanged. Otherwise the encoding is detected with `chardetng` and decoded with `encoding_rs` (malformed bytes become `U+FFFD`).
#[cfg(feature = "charset")]
fn decode_response(bytes: &[u8]) -> String {
    if let Ok(s) = std::str::from_utf8(bytes) {
        return s.to_string();
    }

    let mut detector = chardetng::EncodingDetector::new();

    detector.feed(bytes, true);

    detector.guess(None, true).decode(bytes).0.into_owned()
}

/// Format an IP target as WHOIS query text. Unlike a URI authority, IPv6 addresses are **not** wrapped in `[` and `]`, which is what WHOIS servers expect.
fn format_ip_query(host: &Host) -> String {
    match host {
        Host::IPv4(ip) => ip.to_string(),
        Host::IPv6(ip) => ip.to_string(),
        Host::Domain(domain) => domain.to_string(),
    }
}

/// The `WhoIs` structure stores the list of WHOIS servers in-memory.
#[derive(Debug, Clone)]
pub struct WhoIs {
    map: HashMap<String, WhoIsServerValue>,
    ip:  WhoIsServerValue,
}

impl WhoIs {
    /// Create a `WhoIs` instance which doesn't have a WHOIS server list. You should provide the host that is used for query ip. You may want to use the host `"whois.arin.net"`.
    pub fn from_host<T: AsRef<str>>(host: T) -> Result<WhoIs, WhoIsError> {
        Ok(Self {
            map: HashMap::new(), ip: WhoIsServerValue::from_string(host)?
        })
    }

    /// Read the list of WHOIS servers (JSON data) from a file to create a `WhoIs` instance.
    #[inline]
    pub fn from_path<P: AsRef<Path>>(path: P) -> Result<WhoIs, WhoIsError> {
        let path = path.as_ref();

        let file = File::open(path)?;

        let map: Map<String, Value> = serde_json::from_reader(file)?;

        Self::from_inner(map)
    }

    #[cfg(feature = "tokio")]
    /// Read the list of WHOIS servers (JSON data) from a file to create a `WhoIs` instance. For `serde_json` doesn't support async functions, consider just using the `from_path` function.
    #[inline]
    pub async fn from_path_async<P: AsRef<Path>>(path: P) -> Result<WhoIs, WhoIsError> {
        let file = tokio::fs::read(path).await?;

        let map: Map<String, Value> = serde_json::from_slice(file.as_slice())?;

        Self::from_inner(map)
    }

    /// Read the list of WHOIS servers (JSON data) from a string to create a `WhoIs` instance.
    #[inline]
    pub fn from_string<S: AsRef<str>>(string: S) -> Result<WhoIs, WhoIsError> {
        let string = string.as_ref();

        let map: Map<String, Value> = serde_json::from_str(string)?;

        Self::from_inner(map)
    }

    fn from_inner(mut map: Map<String, Value>) -> Result<WhoIs, WhoIsError> {
        let ip = match map.remove("_") {
            Some(server) => {
                if let Value::Object(server) = server {
                    match server.get("ip") {
                        Some(server) => {
                            if server.is_null() {
                                return Err(WhoIsError::MapError(
                                    "`ip` in the `_` object in the server list is null.",
                                ));
                            }

                            WhoIsServerValue::from_value(server)?
                        },
                        None => {
                            return Err(WhoIsError::MapError(
                                "Cannot find `ip` in the `_` object in the server list.",
                            ));
                        },
                    }
                } else {
                    return Err(WhoIsError::MapError("`_` in the server list is not an object."));
                }
            },
            None => return Err(WhoIsError::MapError("Cannot find `_` in the server list.")),
        };

        let mut new_map: HashMap<String, WhoIsServerValue> = HashMap::with_capacity(map.len());

        for (k, v) in map {
            if !v.is_null() {
                let server_value = WhoIsServerValue::from_value(&v)?;
                new_map.insert(k, server_value);
            }
        }

        Ok(WhoIs {
            map: new_map,
            ip,
        })
    }
}

impl WhoIs {
    /// Try to find a WHOIS server for a TLD by querying the `_nicname._tcp.<tld>` SRV records through the given DNS server, caching any server that is found for later lookups. Returns whether a server is available for the TLD (already known or newly discovered).
    pub fn can_find_server_for_tld<T: AsRef<str>, D: AsRef<str>>(
        &mut self,
        tld: T,
        dns_server: D,
    ) -> Result<bool, WhoIsError> {
        let mut tld = tld.as_ref();
        let dns_server = dns_server.as_ref();

        let address = match dns_server.parse() {
            Ok(address) => address,
            Err(_error) => {
                return Err(WhoIsError::MapError("The DNS server address is incorrect."));
            },
        };
        let conn = UdpClientConnection::new(address).map_err(io::Error::other)?;
        let client = SyncClient::new(conn);

        loop {
            if self.map.contains_key(tld) {
                return Ok(true);
            }

            match tld.find('.') {
                Some(index) => {
                    tld = &tld[index + 1..];
                },
                None => {
                    tld = "";
                },
            }

            if tld.is_empty() {
                return Ok(false);
            }

            let name =
                Name::from_str(&format!("_nicname._tcp.{tld}.")).map_err(io::Error::other)?;
            let response: DnsResponse =
                client.query(&name, DNSClass::IN, RecordType::SRV).map_err(io::Error::other)?;
            let answers: &[Record] = response.answers();

            for record in answers {
                if let Some(RData::SRV(record)) = record.data() {
                    let target = record.target().to_string();
                    let new_server =
                        match WhoIsServerValue::from_string(target.trim_end_matches('.')) {
                            Ok(new_server) => new_server,
                            Err(_error) => continue,
                        };

                    self.map.insert(tld.to_string(), new_server);

                    return Ok(true);
                }
            }
        }
    }

    /// Try to find a WHOIS server for a TLD from the in-memory server list, walking up the labels of the given TLD/domain (e.g. `a.b.example` -> `b.example` -> `example` -> `""`) until a match is found. Returns the matched server, if any.
    pub fn get_server_by_tld(&self, mut tld: &str) -> Option<&WhoIsServerValue> {
        let mut server;

        loop {
            server = self.map.get(tld);

            if server.is_some() {
                break;
            }

            if tld.is_empty() {
                break;
            }

            match tld.find('.') {
                Some(index) => {
                    tld = &tld[index + 1..];
                },
                None => {
                    tld = "";
                },
            }
        }

        server
    }

    fn lookup_once(
        server: &WhoIsServerValue,
        text: &str,
        timeout: Option<Duration>,
    ) -> Result<(String, Vec<u8>), WhoIsError> {
        let addr = server.host.to_addr_string(DEFAULT_WHOIS_HOST_PORT);

        let mut client = if let Some(timeout) = timeout {
            let socket_addrs: Vec<SocketAddr> = addr.to_socket_addrs()?.collect();

            if socket_addrs.is_empty() {
                return Err(io::Error::new(
                    io::ErrorKind::AddrNotAvailable,
                    "the host is not resolved to any socket address",
                )
                .into());
            }

            let mut client = None;

            for socket_addr in socket_addrs.iter().take(socket_addrs.len() - 1) {
                if let Ok(c) = TcpStream::connect_timeout(socket_addr, timeout) {
                    client = Some(c);
                    break;
                }
            }

            let client = if let Some(client) = client {
                client
            } else {
                let socket_addr = &socket_addrs[socket_addrs.len() - 1];
                TcpStream::connect_timeout(socket_addr, timeout)?
            };

            client.set_read_timeout(Some(timeout))?;
            client.set_write_timeout(Some(timeout))?;
            client
        } else {
            TcpStream::connect(&addr)?
        };

        if let Some(query) = &server.query {
            client.write_all(query.replace("$addr", text).as_bytes())?;
        } else {
            client.write_all(DEFAULT_WHOIS_HOST_QUERY.replace("$addr", text).as_bytes())?;
        }

        client.flush()?;

        let mut query_result = Vec::new();

        client.read_to_end(&mut query_result)?;

        // Return the host without the port so a referral host can be compared against it.
        Ok((server.host.host.to_string(), query_result))
    }

    fn lookup_inner(
        server: &WhoIsServerValue,
        text: &str,
        timeout: Option<Duration>,
        mut follow: u16,
    ) -> Result<Vec<u8>, WhoIsError> {
        let mut query_result = Self::lookup_once(server, text, timeout)?;

        while follow > 0 {
            let body = decode_response(&query_result.1);

            if let Some(h) = extract_referral_host(&body)
                && !h.eq_ignore_ascii_case(&query_result.0)
                && let Ok(server) = WhoIsServerValue::from_string(h)
            {
                query_result = Self::lookup_once(&server, text, timeout)?;

                follow -= 1;

                continue;
            }

            break;
        }

        Ok(query_result.1)
    }

    /// Lookup a domain or an IP, returning the raw bytes of the final WHOIS response without decoding.
    pub fn lookup_raw(&self, options: WhoIsLookupOptions) -> Result<Vec<u8>, WhoIsError> {
        match &options.target.0 {
            Host::IPv4(_) | Host::IPv6(_) => {
                let server = match &options.server {
                    Some(server) => server,
                    None => &self.ip,
                };

                let text = format_ip_query(&options.target.0);

                Self::lookup_inner(server, &text, options.timeout, options.follow)
            },
            Host::Domain(domain) => {
                let server = match &options.server {
                    Some(server) => server,
                    None => match self.get_server_by_tld(domain.as_str()) {
                        Some(server) => server,
                        None => {
                            return Err(WhoIsError::MapError(
                                "No whois server is known for this kind of object.",
                            ));
                        },
                    },
                };

                // The domain is already ASCII (Punycode) encoded; decode it back to Unicode for servers that require it.
                let text = server.encode_domain(domain);

                Self::lookup_inner(server, text.as_ref(), options.timeout, options.follow)
            },
        }
    }

    /// Lookup a domain or an IP.
    #[inline]
    pub fn lookup(&self, options: WhoIsLookupOptions) -> Result<String, WhoIsError> {
        Ok(decode_response(&self.lookup_raw(options)?))
    }
}

#[cfg(feature = "tokio")]
impl WhoIs {
    async fn lookup_once_async(
        server: &WhoIsServerValue,
        text: &str,
        timeout: Option<Duration>,
    ) -> Result<(String, Vec<u8>), WhoIsError> {
        let addr = server.host.to_addr_string(DEFAULT_WHOIS_HOST_PORT);

        if let Some(timeout) = timeout {
            let socket_addrs: Vec<SocketAddr> = addr.to_socket_addrs()?.collect();

            if socket_addrs.is_empty() {
                return Err(io::Error::new(
                    io::ErrorKind::AddrNotAvailable,
                    "the host is not resolved to any socket address",
                )
                .into());
            }

            let mut client = None;

            for socket_addr in socket_addrs.iter().take(socket_addrs.len() - 1) {
                if let Ok(c) =
                    tokio::time::timeout(timeout, tokio::net::TcpStream::connect(&socket_addr))
                        .await?
                {
                    client = Some(c);
                    break;
                }
            }

            let mut client = if let Some(client) = client {
                client
            } else {
                let socket_addr = &socket_addrs[socket_addrs.len() - 1];
                tokio::time::timeout(timeout, tokio::net::TcpStream::connect(socket_addr)).await??
            };

            if let Some(query) = &server.query {
                tokio::time::timeout(
                    timeout,
                    client.write_all(query.replace("$addr", text).as_bytes()),
                )
                .await??;
            } else {
                tokio::time::timeout(
                    timeout,
                    client.write_all(DEFAULT_WHOIS_HOST_QUERY.replace("$addr", text).as_bytes()),
                )
                .await??;
            }

            tokio::time::timeout(timeout, client.flush()).await??;

            let mut query_result = Vec::new();

            tokio::time::timeout(timeout, client.read_to_end(&mut query_result)).await??;

            // Return the host without the port so a referral host can be compared against it.
            Ok((server.host.host.to_string(), query_result))
        } else {
            let mut client = tokio::net::TcpStream::connect(&addr).await?;

            if let Some(query) = &server.query {
                client.write_all(query.replace("$addr", text).as_bytes()).await?;
            } else {
                client
                    .write_all(DEFAULT_WHOIS_HOST_QUERY.replace("$addr", text).as_bytes())
                    .await?;
            }

            client.flush().await?;

            let mut query_result = Vec::new();

            client.read_to_end(&mut query_result).await?;

            // Return the host without the port so a referral host can be compared against it.
            Ok((server.host.host.to_string(), query_result))
        }
    }

    async fn lookup_inner_async<'a>(
        server: &'a WhoIsServerValue,
        text: &'a str,
        timeout: Option<Duration>,
        mut follow: u16,
    ) -> Result<Vec<u8>, WhoIsError> {
        let mut query_result = Self::lookup_once_async(server, text, timeout).await?;

        while follow > 0 {
            let body = decode_response(&query_result.1);

            if let Some(h) = extract_referral_host(&body)
                && !h.eq_ignore_ascii_case(&query_result.0)
                && let Ok(server) = WhoIsServerValue::from_string(h)
            {
                query_result = Self::lookup_once_async(&server, text, timeout).await?;

                follow -= 1;

                continue;
            }

            break;
        }

        Ok(query_result.1)
    }

    /// Lookup a domain or an IP, returning the raw bytes of the final WHOIS response without decoding.
    pub async fn lookup_raw_async(
        &self,
        options: WhoIsLookupOptions,
    ) -> Result<Vec<u8>, WhoIsError> {
        match &options.target.0 {
            Host::IPv4(_) | Host::IPv6(_) => {
                let server = match &options.server {
                    Some(server) => server,
                    None => &self.ip,
                };

                let text = format_ip_query(&options.target.0);

                Self::lookup_inner_async(server, &text, options.timeout, options.follow).await
            },
            Host::Domain(domain) => {
                let server = match &options.server {
                    Some(server) => server,
                    None => match self.get_server_by_tld(domain.as_str()) {
                        Some(server) => server,
                        None => {
                            return Err(WhoIsError::MapError(
                                "No whois server is known for this kind of object.",
                            ));
                        },
                    },
                };

                // The domain is already ASCII (Punycode) encoded; decode it back to Unicode for servers that require it.
                let text = server.encode_domain(domain);

                Self::lookup_inner_async(server, text.as_ref(), options.timeout, options.follow)
                    .await
            },
        }
    }

    /// Lookup a domain or an IP.
    #[inline]
    pub async fn lookup_async(&self, options: WhoIsLookupOptions) -> Result<String, WhoIsError> {
        Ok(decode_response(&self.lookup_raw_async(options).await?))
    }
}

#[cfg(test)]
mod tests {
    use super::{decode_response, extract_referral_host};

    #[test]
    fn extract_referral_host_trims_trailing_cr() {
        let body = "Domain: example.com\r\nReferralServer: whois://whois.arin.net\r\n";

        assert_eq!(Some("whois.arin.net"), extract_referral_host(body));
    }

    #[test]
    fn decode_response_passes_through_utf8() {
        assert_eq!("café", decode_response("café".as_bytes()));
    }

    // "café" with `é` as the single byte 0xE9 (Windows-1252 / ISO-8859-1).
    #[cfg(feature = "charset")]
    #[test]
    fn decode_response_detects_windows_1252() {
        assert_eq!("café", decode_response(&[b'c', b'a', b'f', 0xE9]));
    }

    #[cfg(not(feature = "charset"))]
    #[test]
    fn decode_response_is_lossy_without_charset() {
        assert_eq!("caf\u{FFFD}", decode_response(&[b'c', b'a', b'f', 0xE9]));
    }
}