merge upstream

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "activitypub_federation"
-version = "0.5.0-beta.3"
+version = "0.5.0-beta.4"
 edition = "2021"
 description = "High-level Activitypub framework"
 keywords = ["activitypub", "activitystreams", "federation", "fediverse"]
@@ -9,38 +9,38 @@ repository = "https://github.com/LemmyNet/activitypub-federation-rust"
 documentation = "https://docs.rs/activitypub_federation/"
 
 [dependencies]
-chrono = { version = "0.4.26", features = ["clock"], default-features = false }
-serde = { version = "1.0.164", features = ["derive"] }
-async-trait = "0.1.68"
-url = { version = "2.4.0", features = ["serde"] }
-serde_json = { version = "1.0.96", features = ["preserve_order"] }
-anyhow = "1.0.71"
-reqwest = { version = "0.11.18", features = ["json", "stream"] }
-reqwest-middleware = "0.2.2"
-tracing = "0.1.37"
-base64 = "0.21.2"
-openssl = "0.10.54"
+chrono = { version = "0.4.31", features = ["clock"], default-features = false }
+serde = { version = "1.0.189", features = ["derive"] }
+async-trait = "0.1.74"
+url = { version = "2.4.1", features = ["serde"] }
+serde_json = { version = "1.0.107", features = ["preserve_order"] }
+anyhow = "1.0.75"
+reqwest = { version = "0.11.22", features = ["json", "stream"] }
+reqwest-middleware = "0.2.3"
+tracing = "0.1.40"
+base64 = "0.21.5"
+openssl = "0.10.57"
 once_cell = "1.18.0"
 http = "0.2.9"
-sha2 = "0.10.6"
-thiserror = "1.0.40"
+sha2 = "0.10.8"
+thiserror = "1.0.50"
 derive_builder = "0.12.0"
 itertools = "0.11.0"
-dyn-clone = "1.0.11"
+dyn-clone = "1.0.14"
 enum_delegate = "0.2.0"
-httpdate = "1.0.2"
+httpdate = "1.0.3"
 http-signature-normalization-reqwest = { version = "0.10.0", default-features = false, features = [
   "default-spawner",
   "sha-2",
   "middleware",
 ] }
 http-signature-normalization = "0.7.0"
-bytes = "1.4.0"
+bytes = "1.5.0"
 futures-core = { version = "0.3.28", default-features = false }
-pin-project-lite = "0.2.9"
+pin-project-lite = "0.2.13"
 activitystreams-kinds = "0.3.0"
-regex = { version = "1.8.4", default-features = false, features = ["std", "unicode-case"] }
-tokio = { version = "1.21.2", features = [
+regex = { version = "1.10.2", default-features = false, features = ["std", "unicode-case"] }
+tokio = { version = "1.33.0", features = [
   "sync",
   "rt",
   "rt-multi-thread",
@@ -48,7 +48,7 @@ tokio = { version = "1.21.2", features = [
 ] }
 
 # Actix-web
-actix-web = { version = "4.3.1", default-features = false, optional = true }
+actix-web = { version = "4.4.0", default-features = false, optional = true }
 
 # Axum
 axum = { git = "https://github.com/tokio-rs/axum.git", features = [

src/fetch/collection_id.rs

@@ -40,9 +40,10 @@ where
     where
         <Kind as Collection>::Error: From<Error>,
     {
-        let json = fetch_object_http(&self.0, data).await?;
-        Kind::verify(&json, &self.0, data).await?;
-        Kind::from_json(json, owner, data).await
+        let res = fetch_object_http(&self.0, data).await?;
+        let redirect_url = &res.url;
+        Kind::verify(&res.object, redirect_url, data).await?;
+        Kind::from_json(res.object, owner, data).await
     }
 }
 

src/fetch/mod.rs

@@ -23,6 +23,14 @@ pub mod object_id;
 /// Resolves identifiers of the form `name@example.com`
 pub mod webfinger;
 
+/// Response from fetching a remote object
+pub struct FetchObjectResponse<Kind> {
+    /// The resolved object
+    pub object: Kind,
+    /// Contains the final URL (different from request URL in case of redirect)
+    pub url: Url,
+}
+
 /// Fetch a remote object over HTTP and convert to `Kind`.
 ///
 /// [crate::fetch::object_id::ObjectId::dereference] wraps this function to add caching and
@@ -38,7 +46,7 @@ pub mod webfinger;
 pub async fn fetch_object_http<T: Clone, Kind: DeserializeOwned>(
     url: &Url,
     data: &Data<T>,
-) -> Result<Kind, Error> {
+) -> Result<FetchObjectResponse<Kind>, Error> {
     fetch_object_http_with_accept(url, data, FEDERATION_CONTENT_TYPE).await
 }
 
@@ -48,7 +56,7 @@ async fn fetch_object_http_with_accept<T: Clone, Kind: DeserializeOwned>(
     url: &Url,
     data: &Data<T>,
     content_type: &str,
-) -> Result<Kind, Error> {
+) -> Result<FetchObjectResponse<Kind>, Error> {
     let config = &data.config;
     // dont fetch local objects this way
     debug_assert!(url.domain() != Some(&config.domain));
@@ -84,5 +92,9 @@ async fn fetch_object_http_with_accept<T: Clone, Kind: DeserializeOwned>(
         return Err(Error::ObjectDeleted);
     }
 
-    res.json_limited().await
+    let url = res.url().clone();
+    Ok(FetchObjectResponse {
+        object: res.json_limited().await?,
+        url,
+    })
 }

src/fetch/object_id.rs

@@ -156,10 +156,11 @@ where
             return Err(anyhow!("Fetched remote object {} which was deleted", self).into());
         }
 
-        let res2 = res?;
+        let res = res?;
+        let redirect_url = &res.url;
 
-        Kind::verify(&res2, self.inner(), data).await?;
-        Kind::from_json(res2, data).await
+        Kind::verify(&res.object, redirect_url, data).await?;
+        Kind::from_json(res.object, data).await
     }
 }
 

src/fetch/webfinger.rs

@@ -38,7 +38,8 @@ where
 
     let res: Webfinger =
         fetch_object_http_with_accept(&Url::parse(&fetch_url)?, data, "application/jrd+json")
-            .await?;
+            .await?
+            .object;
 
     debug_assert_eq!(res.subject, format!("acct:{identifier}"));
     let links: Vec<Url> = res
@@ -91,14 +92,10 @@ pub fn extract_webfinger_name<T>(query: &str, data: &Data<T>) -> Result<String,
 where
     T: Clone,
 {
+    // Regex to extract usernames from webfinger query. Supports different alphabets using `\p{L}`.
     // TODO: would be nice if we could implement this without regex and remove the dependency
-    // Regex taken from Mastodon -
-    // https://github.com/mastodon/mastodon/blob/2b113764117c9ab98875141bcf1758ba8be58173/app/models/account.rb#L65
-    let regex = Regex::new(&format!(
-        "^acct:((?i)[a-z0-9_]+([a-z0-9_\\.-]+[a-z0-9_]+)?)@{}$",
-        data.domain()
-    ))
-    .map_err(Error::other)?;
+    let regex =
+        Regex::new(&format!(r"^acct:([\p{{L}}0-9_]+)@{}$", data.domain())).map_err(Error::other)?;
     Ok(regex
         .captures(query)
         .and_then(|c| c.get(1))
@@ -238,4 +235,31 @@ mod tests {
         webfinger_resolve_actor::<DbConnection, DbUser>("graf@poa.st", &data).await?;
         Ok(())
     }
+
+    #[tokio::test]
+    async fn test_webfinger_extract_name() -> Result<(), Error> {
+        use crate::traits::tests::DbConnection;
+        let data = Data {
+            config: FederationConfig::builder()
+                .domain("example.com")
+                .app_data(DbConnection)
+                .build()
+                .await
+                .unwrap(),
+            request_counter: Default::default(),
+        };
+        assert_eq!(
+            Ok("test123".to_string()),
+            extract_webfinger_name("acct:test123@example.com", &data)
+        );
+        assert_eq!(
+            Ok("Владимир".to_string()),
+            extract_webfinger_name("acct:Владимир@example.com", &data)
+        );
+        assert_eq!(
+            Ok("تجريب".to_string()),
+            extract_webfinger_name("acct:تجريب@example.com", &data)
+        );
+        Ok(())
+    }
 }