openmoat_proxy/
request.rs

1//! The request head a client sends the proxy: `CONNECT host:port` or one
2//! absolute-form plain-HTTP request (`GET http://host/path HTTP/1.1`).
3//!
4//! Hosts come out in the form openmoat-core's net rules match: lowercase, without
5//! a trailing dot or IPv6 brackets, IP literals in their canonical spelling.
6
7use std::net::{IpAddr, Ipv4Addr, Ipv6Addr};
8
9/// Most headers one request head may carry.
10pub const MAX_HEADERS: usize = 64;
11
12/// Headers that describe one connection, not the request (RFC 9110 §7.6.1),
13/// plus the proxy credentials no upstream should see. `Transfer-Encoding` is
14/// kept: the body is relayed byte for byte, so its framing must arrive intact.
15const HOP_BY_HOP: &[&str] = &[
16    "connection",
17    "keep-alive",
18    "proxy-connection",
19    "proxy-authenticate",
20    "proxy-authorization",
21    "te",
22    "trailer",
23    "upgrade",
24];
25
26/// What the client asked the proxy to reach.
27#[derive(Debug, Clone, PartialEq, Eq)]
28pub enum Target {
29    /// `CONNECT host:port`: a tunnel, checked by the TLS SNI it then carries.
30    Connect {
31        /// Normalised host.
32        host: String,
33        /// Port; never 0.
34        port: u16,
35    },
36    /// One plain-HTTP request, rewritten for the upstream.
37    Http {
38        /// Normalised host from the absolute-form target.
39        host: String,
40        /// Port; 80 when the target names none.
41        port: u16,
42        /// The head to send upstream: origin-form request line, hop-by-hop
43        /// headers removed, `Host` present and `Connection: close` added.
44        head: Vec<u8>,
45    },
46}
47
48/// A parsed request head.
49#[derive(Debug, Clone, PartialEq, Eq)]
50pub struct Request {
51    /// The method, as sent (`CONNECT`, `GET`, …).
52    pub method: String,
53    /// Where the request goes.
54    pub target: Target,
55    /// Bytes of the input the head occupied; what follows is body or tunnel data.
56    pub head_len: usize,
57}
58
59impl Request {
60    /// The target host.
61    #[must_use]
62    pub fn host(&self) -> &str {
63        match &self.target {
64            Target::Connect { host, .. } | Target::Http { host, .. } => host,
65        }
66    }
67
68    /// The target port.
69    #[must_use]
70    pub fn port(&self) -> u16 {
71        match &self.target {
72            Target::Connect { port, .. } | Target::Http { port, .. } => *port,
73        }
74    }
75}
76
77/// Why a request head is refused.
78#[derive(Debug, Clone, PartialEq, Eq, thiserror::Error)]
79#[non_exhaustive]
80pub enum RequestError {
81    /// Not HTTP/1.x, or a syntax error `httparse` reports.
82    #[error("malformed request: {0}")]
83    Malformed(String),
84    /// A well-formed request the proxy does not serve.
85    #[error("{0}")]
86    Unsupported(&'static str),
87}
88
89/// Parse one request head from the start of `buf`. `Ok(None)` means the head is
90/// not complete yet; the caller bounds how much it reads.
91pub fn parse(buf: &[u8]) -> Result<Option<Request>, RequestError> {
92    let mut headers = [httparse::EMPTY_HEADER; MAX_HEADERS];
93    let mut req = httparse::Request::new(&mut headers);
94    let head_len = match req.parse(buf) {
95        Ok(httparse::Status::Complete(n)) => n,
96        Ok(httparse::Status::Partial) => return Ok(None),
97        Err(e) => return Err(RequestError::Malformed(e.to_string())),
98    };
99    let (Some(method), Some(raw_target), Some(version)) = (req.method, req.path, req.version)
100    else {
101        return Err(RequestError::Malformed(
102            "incomplete request line".to_owned(),
103        ));
104    };
105    let target = if method == "CONNECT" {
106        let (host, port) = authority(raw_target, None)?;
107        Target::Connect { host, port }
108    } else {
109        http_target(method, raw_target, version, req.headers)?
110    };
111    Ok(Some(Request {
112        method: method.to_owned(),
113        target,
114        head_len,
115    }))
116}
117
118fn http_target(
119    method: &str,
120    raw_target: &str,
121    version: u8,
122    headers: &[httparse::Header<'_>],
123) -> Result<Target, RequestError> {
124    let Some((scheme, rest)) = raw_target.split_once("://") else {
125        return Err(RequestError::Unsupported(
126            "only absolute-form requests (http://host/path) are proxied",
127        ));
128    };
129    if !scheme.eq_ignore_ascii_case("http") {
130        return Err(RequestError::Unsupported(
131            "only http:// is forwarded; use CONNECT for https",
132        ));
133    }
134    let end = rest.find(['/', '?', '#']).unwrap_or(rest.len());
135    let (raw_authority, path) = rest.split_at(end);
136    let path = path.split('#').next().unwrap_or_default();
137    let (host, port) = authority(raw_authority, Some(80))?;
138    check_framing(version, headers)?;
139
140    let mut connection_tokens: Vec<String> = Vec::new();
141    for h in headers
142        .iter()
143        .filter(|h| h.name.eq_ignore_ascii_case("connection"))
144    {
145        let value = String::from_utf8_lossy(h.value);
146        connection_tokens.extend(value.split(',').map(|t| t.trim().to_ascii_lowercase()));
147    }
148    let slash = if path.starts_with('/') { "" } else { "/" };
149    let mut head = format!("{method} {slash}{path} HTTP/1.{version}\r\n").into_bytes();
150    let mut has_host = false;
151    for h in headers {
152        let name = h.name.to_ascii_lowercase();
153        if HOP_BY_HOP.contains(&name.as_str()) || connection_tokens.contains(&name) {
154            continue;
155        }
156        if name == "host" {
157            if has_host || host_header(h.value) != Some((host.clone(), port)) {
158                return Err(RequestError::Unsupported(
159                    "Host header does not match the request target",
160                ));
161            }
162            has_host = true;
163        }
164        head.extend_from_slice(h.name.as_bytes());
165        head.extend_from_slice(b": ");
166        head.extend_from_slice(h.value);
167        head.extend_from_slice(b"\r\n");
168    }
169    if !has_host {
170        head.extend_from_slice(format!("Host: {raw_authority}\r\n").as_bytes());
171    }
172    head.extend_from_slice(b"Connection: close\r\n\r\n");
173    Ok(Target::Http { host, port, head })
174}
175
176/// Refuse the framing ambiguities request smuggling relies on: both
177/// `Content-Length` and `Transfer-Encoding`, repeated or list-form length,
178/// or a `Transfer-Encoding` whose tokens are anything but a single `chunked`.
179/// HTTP/1.0 has no transfer codings, so a 1.0 request carrying
180/// `Transfer-Encoding` is malformed (RFC 9112 §6.1): a server may read its
181/// body by another rule.
182fn check_framing(version: u8, headers: &[httparse::Header<'_>]) -> Result<(), RequestError> {
183    const AMBIGUOUS: RequestError = RequestError::Unsupported(
184        "ambiguous body framing (Content-Length with Transfer-Encoding, repeated, \
185         list-form length, or a transfer coding other than chunked)",
186    );
187    let mut lengths: Vec<&[u8]> = Vec::new();
188    let mut encodings: Vec<&[u8]> = Vec::new();
189    for h in headers {
190        if h.name.eq_ignore_ascii_case("content-length") {
191            lengths.push(h.value);
192        } else if h.name.eq_ignore_ascii_case("transfer-encoding") {
193            encodings.push(h.value);
194        }
195    }
196    if version == 0 && !encodings.is_empty() {
197        return Err(RequestError::Malformed(
198            "Transfer-Encoding in an HTTP/1.0 request".to_owned(),
199        ));
200    }
201    if lengths.len() > 1 || (!lengths.is_empty() && !encodings.is_empty()) {
202        return Err(AMBIGUOUS);
203    }
204    if encodings.len() > 1 {
205        return Err(AMBIGUOUS);
206    }
207    if let Some(bytes) = encodings.first() {
208        let value = std::str::from_utf8(bytes).map_err(|_| AMBIGUOUS)?;
209        let tokens: Vec<String> = value
210            .split(',')
211            .map(|t| t.trim().to_ascii_lowercase())
212            .collect();
213        if tokens.as_slice() != ["chunked"] {
214            return Err(AMBIGUOUS);
215        }
216    }
217    if let Some(bytes) = lengths.first() {
218        let value = std::str::from_utf8(bytes).map_err(|_| AMBIGUOUS)?;
219        let trimmed = value.trim();
220        // RFC 9112 §8.6: a `Content-Length` is one or more decimal digits.
221        // The explicit all-ASCII-digit check rejects `+10`, `-10`, embedded
222        // whitespace (`10 10`), hex (`0x0a`) and the list form (`10, 10`);
223        // `u64::from_str` guards against overflow after that.
224        let digits_only = !trimmed.is_empty() && trimmed.bytes().all(|b| b.is_ascii_digit());
225        if !digits_only || trimmed.parse::<u64>().is_err() {
226            return Err(AMBIGUOUS);
227        }
228    }
229    Ok(())
230}
231
232fn host_header(value: &[u8]) -> Option<(String, u16)> {
233    authority(std::str::from_utf8(value).ok()?.trim(), Some(80)).ok()
234}
235
236/// `host[:port]` or `[v6][:port]`, with no userinfo. `default_port` is `None`
237/// when a port is required (CONNECT).
238fn authority(raw: &str, default_port: Option<u16>) -> Result<(String, u16), RequestError> {
239    const BAD: RequestError = RequestError::Unsupported("invalid host or port in request target");
240    if raw.contains('@') {
241        return Err(RequestError::Unsupported(
242            "credentials in the request target are refused",
243        ));
244    }
245    let (host, port) = if let Some(rest) = raw.strip_prefix('[') {
246        let (v6, after) = rest.split_once(']').ok_or(BAD)?;
247        let ip: Ipv6Addr = v6.parse().map_err(|_| BAD)?;
248        let port = match after {
249            "" => None,
250            p => Some(p.strip_prefix(':').ok_or(BAD)?),
251        };
252        (ip.to_string(), port)
253    } else {
254        let (host, port) = match raw.split_once(':') {
255            Some((h, p)) => (h, Some(p)),
256            None => (raw, None),
257        };
258        (normalise_name(host).ok_or(BAD)?, port)
259    };
260    let port = match (port, default_port) {
261        (Some(p), _) => p.parse::<u16>().ok().filter(|p| *p != 0).ok_or(BAD)?,
262        (None, Some(d)) => d,
263        (None, None) => return Err(BAD),
264    };
265    Ok((host, port))
266}
267
268/// A DNS name or IPv4 literal in canonical form; `None` when it is neither.
269pub(crate) fn normalise_name(raw: &str) -> Option<String> {
270    let name = raw.strip_suffix('.').unwrap_or(raw).to_ascii_lowercase();
271    if let Ok(ip) = name.parse::<Ipv4Addr>() {
272        return Some(IpAddr::V4(ip).to_string());
273    }
274    let valid = !name.is_empty()
275        && name.len() <= 253
276        && name.split('.').all(|label| {
277            !label.is_empty()
278                && label.len() <= 63
279                && label
280                    .bytes()
281                    .all(|b| b.is_ascii_alphanumeric() || b == b'-' || b == b'_')
282        });
283    // No top-level domain is a number. A name whose last label is digits or
284    // `0x…` is an address in another spelling (`1572394766`, `0x5d.0xb8.0xd7.0x0e`,
285    // `0177.0.0.1`) that the system resolver turns into an IP, past rules that
286    // name the address the usual way.
287    let last = name.rsplit('.').next().unwrap_or_default();
288    let numeric = last.bytes().all(|b| b.is_ascii_digit()) || last.starts_with("0x");
289    (valid && !numeric).then_some(name)
290}
291
292#[cfg(test)]
293mod tests;
openmoat_proxy/request.rs

openmoat_proxy/
request.rs
