this repo has no description
lewis.moe
1use reqwest::{Client, ClientBuilder, Url};
2use std::net::{IpAddr, SocketAddr, ToSocketAddrs};
3use std::sync::OnceLock;
4use std::time::Duration;
5use tracing::warn;
6
7pub const DEFAULT_HEADERS_TIMEOUT: Duration = Duration::from_secs(10);
8pub const DEFAULT_BODY_TIMEOUT: Duration = Duration::from_secs(30);
9pub const DEFAULT_CONNECT_TIMEOUT: Duration = Duration::from_secs(5);
10pub const MAX_RESPONSE_SIZE: u64 = 10 * 1024 * 1024;
11
12static PROXY_CLIENT: OnceLock<Client> = OnceLock::new();
13
14pub fn proxy_client() -> &'static Client {
15 PROXY_CLIENT.get_or_init(|| {
16 ClientBuilder::new()
17 .timeout(DEFAULT_BODY_TIMEOUT)
18 .connect_timeout(DEFAULT_CONNECT_TIMEOUT)
19 .pool_max_idle_per_host(10)
20 .pool_idle_timeout(Duration::from_secs(90))
21 .redirect(reqwest::redirect::Policy::none())
22 .build()
23 .expect("Failed to build HTTP client - this indicates a TLS or system configuration issue")
24 })
25}
26
27pub fn is_ssrf_safe(url: &str) -> Result<(), SsrfError> {
28 let parsed = Url::parse(url).map_err(|_| SsrfError::InvalidUrl)?;
29
30 let scheme = parsed.scheme();
31 if scheme != "https" {
32 let allow_http = std::env::var("ALLOW_HTTP_PROXY").is_ok()
33 || url.starts_with("http://127.0.0.1")
34 || url.starts_with("http://localhost");
35
36 if !allow_http {
37 return Err(SsrfError::InsecureProtocol(scheme.to_string()));
38 }
39 }
40
41 let host = parsed.host_str().ok_or(SsrfError::NoHost)?;
42
43 if host == "localhost" {
44 return Ok(());
45 }
46
47 if let Ok(ip) = host.parse::<IpAddr>() {
48 if ip.is_loopback() {
49 return Ok(());
50 }
51 if !is_unicast_ip(&ip) {
52 return Err(SsrfError::NonUnicastIp(ip.to_string()));
53 }
54 return Ok(());
55 }
56
57 let port = parsed.port().unwrap_or(if scheme == "https" { 443 } else { 80 });
58 let socket_addrs: Vec<SocketAddr> = match (host, port).to_socket_addrs() {
59 Ok(addrs) => addrs.collect(),
60 Err(_) => return Err(SsrfError::DnsResolutionFailed(host.to_string())),
61 };
62
63 for addr in &socket_addrs {
64 if !is_unicast_ip(&addr.ip()) {
65 warn!(
66 "DNS resolution for {} returned non-unicast IP: {}",
67 host,
68 addr.ip()
69 );
70 return Err(SsrfError::NonUnicastIp(addr.ip().to_string()));
71 }
72 }
73
74 Ok(())
75}
76
77fn is_unicast_ip(ip: &IpAddr) -> bool {
78 match ip {
79 IpAddr::V4(v4) => {
80 !v4.is_loopback()
81 && !v4.is_broadcast()
82 && !v4.is_multicast()
83 && !v4.is_unspecified()
84 && !v4.is_link_local()
85 && !is_private_v4(v4)
86 }
87 IpAddr::V6(v6) => !v6.is_loopback() && !v6.is_multicast() && !v6.is_unspecified(),
88 }
89}
90
91fn is_private_v4(ip: &std::net::Ipv4Addr) -> bool {
92 let octets = ip.octets();
93 octets[0] == 10
94 || (octets[0] == 172 && (16..=31).contains(&octets[1]))
95 || (octets[0] == 192 && octets[1] == 168)
96 || (octets[0] == 169 && octets[1] == 254)
97}
98
99#[derive(Debug, Clone)]
100pub enum SsrfError {
101 InvalidUrl,
102 InsecureProtocol(String),
103 NoHost,
104 NonUnicastIp(String),
105 DnsResolutionFailed(String),
106}
107
108impl std::fmt::Display for SsrfError {
109 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
110 match self {
111 SsrfError::InvalidUrl => write!(f, "Invalid URL"),
112 SsrfError::InsecureProtocol(p) => write!(f, "Insecure protocol: {}", p),
113 SsrfError::NoHost => write!(f, "No host in URL"),
114 SsrfError::NonUnicastIp(ip) => write!(f, "Non-unicast IP address: {}", ip),
115 SsrfError::DnsResolutionFailed(host) => write!(f, "DNS resolution failed for: {}", host),
116 }
117 }
118}
119
120impl std::error::Error for SsrfError {}
121
122pub const HEADERS_TO_FORWARD: &[&str] = &[
123 "accept-language",
124 "atproto-accept-labelers",
125 "x-bsky-topics",
126];
127
128pub const RESPONSE_HEADERS_TO_FORWARD: &[&str] = &[
129 "atproto-repo-rev",
130 "atproto-content-labelers",
131 "retry-after",
132 "content-type",
133];
134
135pub fn validate_at_uri(uri: &str) -> Result<AtUriParts, &'static str> {
136 if !uri.starts_with("at://") {
137 return Err("URI must start with at://");
138 }
139
140 let path = uri.trim_start_matches("at://");
141 let parts: Vec<&str> = path.split('/').collect();
142
143 if parts.is_empty() {
144 return Err("URI missing DID");
145 }
146
147 let did = parts[0];
148 if !did.starts_with("did:") {
149 return Err("Invalid DID in URI");
150 }
151
152 if parts.len() > 1 {
153 let collection = parts[1];
154 if collection.is_empty() || !collection.contains('.') {
155 return Err("Invalid collection NSID");
156 }
157 }
158
159 Ok(AtUriParts {
160 did: did.to_string(),
161 collection: parts.get(1).map(|s| s.to_string()),
162 rkey: parts.get(2).map(|s| s.to_string()),
163 })
164}
165
166#[derive(Debug, Clone)]
167pub struct AtUriParts {
168 pub did: String,
169 pub collection: Option<String>,
170 pub rkey: Option<String>,
171}
172
173pub fn validate_limit(limit: Option<u32>, default: u32, max: u32) -> u32 {
174 match limit {
175 Some(l) if l == 0 => default,
176 Some(l) if l > max => max,
177 Some(l) => l,
178 None => default,
179 }
180}
181
182pub fn validate_did(did: &str) -> Result<(), &'static str> {
183 if !did.starts_with("did:") {
184 return Err("Invalid DID format");
185 }
186
187 let parts: Vec<&str> = did.split(':').collect();
188 if parts.len() < 3 {
189 return Err("DID must have at least method and identifier");
190 }
191
192 let method = parts[1];
193 if method != "plc" && method != "web" {
194 return Err("Unsupported DID method");
195 }
196
197 Ok(())
198}
199
200#[cfg(test)]
201mod tests {
202 use super::*;
203
204 #[test]
205 fn test_ssrf_safe_https() {
206 assert!(is_ssrf_safe("https://api.bsky.app/xrpc/test").is_ok());
207 }
208
209 #[test]
210 fn test_ssrf_blocks_http_by_default() {
211 let result = is_ssrf_safe("http://external.example.com/xrpc/test");
212 assert!(matches!(result, Err(SsrfError::InsecureProtocol(_)) | Err(SsrfError::DnsResolutionFailed(_))));
213 }
214
215 #[test]
216 fn test_ssrf_allows_localhost_http() {
217 assert!(is_ssrf_safe("http://127.0.0.1:8080/test").is_ok());
218 assert!(is_ssrf_safe("http://localhost:8080/test").is_ok());
219 }
220
221 #[test]
222 fn test_validate_at_uri() {
223 let result = validate_at_uri("at://did:plc:test/app.bsky.feed.post/abc123");
224 assert!(result.is_ok());
225 let parts = result.unwrap();
226 assert_eq!(parts.did, "did:plc:test");
227 assert_eq!(parts.collection, Some("app.bsky.feed.post".to_string()));
228 assert_eq!(parts.rkey, Some("abc123".to_string()));
229 }
230
231 #[test]
232 fn test_validate_at_uri_invalid() {
233 assert!(validate_at_uri("https://example.com").is_err());
234 assert!(validate_at_uri("at://notadid/collection/rkey").is_err());
235 }
236
237 #[test]
238 fn test_validate_limit() {
239 assert_eq!(validate_limit(None, 50, 100), 50);
240 assert_eq!(validate_limit(Some(0), 50, 100), 50);
241 assert_eq!(validate_limit(Some(200), 50, 100), 100);
242 assert_eq!(validate_limit(Some(75), 50, 100), 75);
243 }
244
245 #[test]
246 fn test_validate_did() {
247 assert!(validate_did("did:plc:abc123").is_ok());
248 assert!(validate_did("did:web:example.com").is_ok());
249 assert!(validate_did("notadid").is_err());
250 assert!(validate_did("did:unknown:test").is_err());
251 }
252}