lohr: validate webhook signature

Previously lohr was unusable in a production setting, anyone could forge
a malicious webhook and either:

- mirror a private repo of yours to another remote they own
- wipe a repo of yours by forcing mirroring from an empty mirror

This is no longer the case!

Cargo.toml

@@ -12,8 +12,12 @@ repository = "https://github.com/alarsyo/lohr"
 
 [dependencies]
 anyhow = "1.0.40"
+hex = "0.4.3"
+hmac = "0.10.1"
 log = "0.4.14"
 rocket = "0.4.7"
 rocket_contrib = { version = "0.4.7", features = [ "json" ] }
 serde = { version = "1.0.125", features = [ "derive" ] }
+serde_json = "1.0.64"
 serde_yaml = "0.8.17"
+sha2 = "0.9.3"

README.org

@@ -29,20 +29,28 @@ Setting up =lohr= should be quite simple:
 
 1. Create a =Rocket.toml= file and [[https://rocket.rs/v0.4/guide/configuration/][add your configuration]].
 
-2. Run =lohr=:
+2. Export a secret variable:
+
+   #+begin_src sh
+   $ export LOHR_SECRET=42 # please don't use this secret
+   #+end_src
+
+3. Run =lohr=:
 
    #+begin_src sh
    $ cargo run # or `cargo run --release` for production usage
    #+end_src
 
-3. Configure your favorite git server to send a webhook to =lohr='s address on
+4. Configure your favorite git server to send a webhook to =lohr='s address on
    every push event.
 
    I used [[https://docs.gitea.io/en-us/webhooks/][Gitea's webhooks format]], but I *think* they're similar to GitHub and
    GitLab's webhooks, so these should work too! (If they don't, *please* file an
    issue!)
 
-4. Add a =.lohr= file containing the remotes you want to mirror this repo to:
+   Don't forget to set the webhook secret to the one you chose above.
+
+5. Add a =.lohr= file containing the remotes you want to mirror this repo to:
 
    #+begin_example
    [email protected]:you/your_repo

src/main.rs

@@ -10,7 +10,6 @@ use std::sync::{
 use std::thread;
 
 use rocket::{http::Status, post, routes, State};
-use rocket_contrib::json::Json;
 
 use log::error;
 
@@ -23,10 +22,14 @@ use job::Job;
 mod settings;
 use settings::GlobalSettings;
 
+mod signature;
+use signature::SignedJson;
+
 struct JobSender(Mutex<Sender<Job>>);
+struct Secret(String);
 
 #[post("/", data = "<payload>")]
-fn gitea_webhook(payload: Json<GiteaWebHook>, sender: State<JobSender>) -> Status {
+fn gitea_webhook(payload: SignedJson<GiteaWebHook>, sender: State<JobSender>) -> Status {
     // TODO: validate Gitea signature
 
     {
@@ -66,6 +69,9 @@ fn main() -> anyhow::Result<()> {
     let homedir: PathBuf = homedir.into();
     let homedir = homedir.canonicalize().expect("LOHR_HOME isn't valid!");
 
+    let secret = env::var("LOHR_SECRET")
+        .expect("please provide a secret, otherwise anyone can send you a malicious webhook");
+
     let config = parse_config(homedir.clone())?;
 
     thread::spawn(move || {
@@ -75,6 +81,7 @@ fn main() -> anyhow::Result<()> {
     rocket::ignite()
         .mount("/", routes![gitea_webhook])
         .manage(JobSender(Mutex::new(sender)))
+        .manage(Secret(secret))
         .launch();
 
     Ok(())

src/signature.rs

@@ -0,0 +1,122 @@
+use std::{
+    io::{Read, Write},
+    ops::{Deref, DerefMut},
+};
+
+use rocket::{
+    data::{FromData, Outcome},
+    http::ContentType,
+    State,
+};
+use rocket::{
+    data::{Transform, Transformed},
+    http::Status,
+};
+use rocket::{Data, Request};
+
+use anyhow::anyhow;
+use serde::Deserialize;
+
+use crate::Secret;
+
+const X_GITEA_SIGNATURE: &str = "X-Gitea-Signature";
+
+fn validate_signature(secret: &str, signature: &str, data: &str) -> bool {
+    use hmac::{Hmac, Mac, NewMac};
+    use sha2::Sha256;
+
+    type HmacSha256 = Hmac<Sha256>;
+
+    let mut mac = HmacSha256::new_varkey(secret.as_bytes()).expect("this should never fail");
+
+    mac.update(data.as_bytes());
+
+    match hex::decode(signature) {
+        Ok(bytes) => mac.verify(&bytes).is_ok(),
+        Err(_) => false,
+    }
+}
+
+pub struct SignedJson<T>(pub T);
+
+impl<T> Deref for SignedJson<T> {
+    type Target = T;
+
+    fn deref(&self) -> &T {
+        &self.0
+    }
+}
+
+impl<T> DerefMut for SignedJson<T> {
+    fn deref_mut(&mut self) -> &mut T {
+        &mut self.0
+    }
+}
+
+const LIMIT: u64 = 1 << 20;
+
+// This is a one to one implementation of request_contrib::Json's FromData, but with HMAC
+// validation.
+//
+// Tracking issue for chaining Data guards to avoid this:
+// https://github.com/SergioBenitez/Rocket/issues/775
+impl<'a, T> FromData<'a> for SignedJson<T>
+where
+    T: Deserialize<'a>,
+{
+    type Error = anyhow::Error;
+    type Owned = String;
+    type Borrowed = str;
+
+    fn transform(
+        request: &Request,
+        data: Data,
+    ) -> rocket::data::Transform<Outcome<Self::Owned, Self::Error>> {
+        let size_limit = request.limits().get("json").unwrap_or(LIMIT);
+        let mut s = String::with_capacity(512);
+        match data.open().take(size_limit).read_to_string(&mut s) {
+            Ok(_) => Transform::Borrowed(Outcome::Success(s)),
+            Err(e) => Transform::Borrowed(Outcome::Failure((
+                Status::BadRequest,
+                anyhow!("couldn't read json: {}", e),
+            ))),
+        }
+    }
+
+    fn from_data(request: &Request, o: Transformed<'a, Self>) -> Outcome<Self, Self::Error> {
+        let json_ct = ContentType::new("application", "json");
+        if request.content_type() != Some(&json_ct) {
+            return Outcome::Failure((Status::BadRequest, anyhow!("wrong content type")));
+        }
+
+        let signatures = request.headers().get(X_GITEA_SIGNATURE).collect::<Vec<_>>();
+        if signatures.len() != 1 {
+            return Outcome::Failure((
+                Status::BadRequest,
+                anyhow!("request header needs exactly one signature"),
+            ));
+        }
+
+        let signature = signatures[0];
+
+        let content = o.borrowed()?;
+
+        let secret = request.guard::<State<Secret>>().unwrap();
+
+        if !validate_signature(&secret.0, &signature, content) {
+            return Outcome::Failure((Status::BadRequest, anyhow!("couldn't verify signature")));
+        }
+
+        let content = match serde_json::from_str(content) {
+            Ok(content) => content,
+            Err(e) => {
+                return Outcome::Failure((
+                    Status::BadRequest,
+                    anyhow!("couldn't parse json: {}", e),
+                ))
+            }
+        };
+
+        Outcome::Success(SignedJson(content))
+    }
+}