refactor(tvix/castore/blobsvc): remove BlobService::from_url r/7020

Make blobservice::from_addr use the more specific constructors.

Change-Id: Id9637e279d6910ce6d92ff0086a984be5c65a8c8
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10028
Tested-by: BuildkiteCI
Reviewed-by: Connor Brewster <cbrewster@hey.com>

5 files changed, 98 insertions, 241 deletions

diff --git a/tvix/castore/src/blobservice/from_addr.rs b/tvix/castore/src/blobservice/from_addr.rs
index 2e0a30697d..916c00442c 100644
--- a/tvix/castore/src/blobservice/from_addr.rs
+++ b/tvix/castore/src/blobservice/from_addr.rs
@@ -1,6 +1,8 @@
 use std::sync::Arc;
 use url::Url;
 
+use crate::{proto::blob_service_client::BlobServiceClient, Error};
+
 use super::{BlobService, GRPCBlobService, MemoryBlobService, SledBlobService};
 
 /// Constructs a new instance of a [BlobService] from an URI.
@@ -16,11 +18,43 @@ pub fn from_addr(uri: &str) -> Result<Arc<dyn BlobService>, crate::Error> {
         .map_err(|e| crate::Error::StorageError(format!("unable to parse url: {}", e)))?;
 
     Ok(if url.scheme() == "memory" {
-        Arc::new(MemoryBlobService::from_url(&url)?)
+        // memory doesn't support host or path in the URL.
+        if url.has_host() || !url.path().is_empty() {
+            return Err(Error::StorageError("invalid url".to_string()));
+        }
+        Arc::new(MemoryBlobService::default())
     } else if url.scheme() == "sled" {
-        Arc::new(SledBlobService::from_url(&url)?)
+        // sled doesn't support host, and a path can be provided (otherwise
+        // it'll live in memory only).
+        if url.has_host() {
+            return Err(Error::StorageError("no host allowed".to_string()));
+        }
+
+        if url.path() == "/" {
+            return Err(Error::StorageError(
+                "cowardly refusing to open / with sled".to_string(),
+            ));
+        }
+
+        // TODO: expose compression and other parameters as URL parameters?
+
+        if url.path().is_empty() {
+            return Ok(Arc::new(
+                SledBlobService::new_temporary().map_err(|e| Error::StorageError(e.to_string()))?,
+            ));
+        }
+        return Ok(Arc::new(
+            SledBlobService::new(url.path().into())
+                .map_err(|e| Error::StorageError(e.to_string()))?,
+        ));
     } else if url.scheme().starts_with("grpc+") {
-        Arc::new(GRPCBlobService::from_url(&url)?)
+        // schemes starting with grpc+ go to the GRPCPathInfoService.
+        //   That's normally grpc+unix for unix sockets, and grpc+http(s) for the HTTP counterparts.
+        // - In the case of unix sockets, there must be a path, but may not be a host.
+        // - In the case of non-unix sockets, there must be a host, but no path.
+        // Constructing the channel is handled by tvix_castore::channel::from_url.
+        let client = BlobServiceClient::new(crate::channel::from_url(&url)?);
+        Arc::new(GRPCBlobService::from_client(client))
     } else {
         Err(crate::Error::StorageError(format!(
             "unknown scheme: {}",
@@ -28,3 +62,59 @@ pub fn from_addr(uri: &str) -> Result<Arc<dyn BlobService>, crate::Error> {
         )))?
     })
 }
+
+#[cfg(test)]
+mod tests {
+    use super::from_addr;
+    use lazy_static::lazy_static;
+    use tempfile::TempDir;
+    use test_case::test_case;
+
+    lazy_static! {
+        static ref TMPDIR_SLED_1: TempDir = TempDir::new().unwrap();
+        static ref TMPDIR_SLED_2: TempDir = TempDir::new().unwrap();
+    }
+
+    /// This uses an unsupported scheme.
+    #[test_case("http://foo.example/test", false; "unsupported scheme")]
+    /// This configures sled in temporary mode.
+    #[test_case("sled://", true; "sled valid temporary")]
+    /// This configures sled with /, which should fail.
+    #[test_case("sled:///", false; "sled invalid root")]
+    /// This configures sled with a host, not path, which should fail.
+    #[test_case("sled://foo.example", false; "sled invalid host")]
+    /// This configures sled with a valid path path, which should succeed.
+    #[test_case(&format!("sled://{}", &TMPDIR_SLED_1.path().to_str().unwrap()), true; "sled valid path")]
+    /// This configures sled with a host, and a valid path path, which should fail.
+    #[test_case(&format!("sled://foo.example{}", &TMPDIR_SLED_2.path().to_str().unwrap()), false; "sled invalid host with valid path")]
+    /// This correctly sets the scheme, and doesn't set a path.
+    #[test_case("memory://", true; "memory valid")]
+    /// This sets a memory url host to `foo`
+    #[test_case("memory://foo", false; "memory invalid host")]
+    /// This sets a memory url path to "/", which is invalid.
+    #[test_case("memory:///", false; "memory invalid root path")]
+    /// This sets a memory url path to "/foo", which is invalid.
+    #[test_case("memory:///foo", false; "memory invalid root path foo")]
+    fn test_from_addr(uri_str: &str, is_ok: bool) {
+        assert_eq!(from_addr(uri_str).is_ok(), is_ok)
+    }
+
+    // the gRPC tests below don't fail, because we connect lazily.
+
+    /// Correct scheme to connect to a unix socket.
+    #[test_case("grpc+unix:///path/to/somewhere", true; "grpc valid unix socket")]
+    /// Correct scheme for unix socket, but setting a host too, which is invalid.
+    #[test_case("grpc+unix://host.example/path/to/somewhere", false; "grpc invalid unix socket and host")]
+    /// Correct scheme to connect to localhost, with port 12345
+    #[test_case("grpc+http://[::1]:12345", true; "grpc valid IPv6 localhost port 12345")]
+    /// Correct scheme to connect to localhost over http, without specifying a port.
+    #[test_case("grpc+http://localhost", true; "grpc valid http host without port")]
+    /// Correct scheme to connect to localhost over http, without specifying a port.
+    #[test_case("grpc+https://localhost", true; "grpc valid https host without port")]
+    /// Correct scheme to connect to localhost over http, but with additional path, which is invalid.
+    #[test_case("grpc+http://localhost/some-path", false; "grpc valid invalid host and path")]
+    #[tokio::test]
+    async fn test_from_addr_tokio(uri_str: &str, is_ok: bool) {
+        assert_eq!(from_addr(uri_str).is_ok(), is_ok)
+    }
+}
diff --git a/tvix/castore/src/blobservice/grpc.rs b/tvix/castore/src/blobservice/grpc.rs
index 8d3bdfa8fa..85863c2919 100644
--- a/tvix/castore/src/blobservice/grpc.rs
+++ b/tvix/castore/src/blobservice/grpc.rs
@@ -38,18 +38,6 @@ impl GRPCBlobService {
 
 #[async_trait]
 impl BlobService for GRPCBlobService {
-    /// Constructs a [GRPCBlobService] from the passed [url::Url]:
-    /// - scheme has to match `grpc+*://`.
-    ///   That's normally grpc+unix for unix sockets, and grpc+http(s) for the HTTP counterparts.
-    /// - In the case of unix sockets, there must be a path, but may not be a host.
-    /// - In the case of non-unix sockets, there must be a host, but no path.
-    fn from_url(url: &url::Url) -> Result<Self, crate::Error> {
-        let channel = crate::channel::from_url(url)?;
-        Ok(Self::from_client(
-            proto::blob_service_client::BlobServiceClient::new(channel),
-        ))
-    }
-
     #[instrument(skip(self, digest), fields(blob.digest=%digest))]
     async fn has(&self, digest: &B3Digest) -> Result<bool, crate::Error> {
         let mut grpc_client = self.grpc_client.clone();
@@ -251,66 +239,12 @@ mod tests {
 
     use crate::blobservice::MemoryBlobService;
     use crate::fixtures;
+    use crate::proto::blob_service_client::BlobServiceClient;
     use crate::proto::GRPCBlobServiceWrapper;
 
     use super::BlobService;
     use super::GRPCBlobService;
 
-    /// This uses the wrong scheme
-    #[test]
-    fn test_invalid_scheme() {
-        let url = url::Url::parse("http://foo.example/test").expect("must parse");
-
-        assert!(GRPCBlobService::from_url(&url).is_err());
-    }
-
-    /// This uses the correct scheme for a unix socket.
-    /// The fact that /path/to/somewhere doesn't exist yet is no problem, because we connect lazily.
-    #[tokio::test]
-    async fn test_valid_unix_path() {
-        let url = url::Url::parse("grpc+unix:///path/to/somewhere").expect("must parse");
-
-        assert!(GRPCBlobService::from_url(&url).is_ok());
-    }
-
-    /// This uses the correct scheme for a unix socket,
-    /// but sets a host, which is unsupported.
-    #[tokio::test]
-    async fn test_invalid_unix_path_with_domain() {
-        let url =
-            url::Url::parse("grpc+unix://host.example/path/to/somewhere").expect("must parse");
-
-        assert!(GRPCBlobService::from_url(&url).is_err());
-    }
-
-    /// This uses the correct scheme for a HTTP server.
-    /// The fact that nothing is listening there is no problem, because we connect lazily.
-    #[tokio::test]
-    async fn test_valid_http() {
-        let url = url::Url::parse("grpc+http://localhost").expect("must parse");
-
-        assert!(GRPCBlobService::from_url(&url).is_ok());
-    }
-
-    /// This uses the correct scheme for a HTTPS server.
-    /// The fact that nothing is listening there is no problem, because we connect lazily.
-    #[tokio::test]
-    async fn test_valid_https() {
-        let url = url::Url::parse("grpc+https://localhost").expect("must parse");
-
-        assert!(GRPCBlobService::from_url(&url).is_ok());
-    }
-
-    /// This uses the correct scheme, but also specifies
-    /// an additional path, which is not supported for gRPC.
-    /// The fact that nothing is listening there is no problem, because we connect lazily.
-    #[tokio::test]
-    async fn test_invalid_http_with_path() {
-        let url = url::Url::parse("grpc+https://localhost/some-path").expect("must parse");
-
-        assert!(GRPCBlobService::from_url(&url).is_err());
-    }
-
     /// This ensures connecting via gRPC works as expected.
     #[tokio::test]
     async fn test_valid_unix_path_ping_pong() {
@@ -353,7 +287,10 @@ mod tests {
         let grpc_client = {
             let url = url::Url::parse(&format!("grpc+unix://{}", socket_path.display()))
                 .expect("must parse");
-            GRPCBlobService::from_url(&url).expect("must succeed")
+            let client =
+                BlobServiceClient::new(crate::channel::from_url(&url).expect("must succeed"));
+
+            GRPCBlobService::from_client(client)
         };
 
         let has = grpc_client
diff --git a/tvix/castore/src/blobservice/memory.rs b/tvix/castore/src/blobservice/memory.rs
index 383127344a..8f4e50aa37 100644
--- a/tvix/castore/src/blobservice/memory.rs
+++ b/tvix/castore/src/blobservice/memory.rs
@@ -17,22 +17,6 @@ pub struct MemoryBlobService {
 
 #[async_trait]
 impl BlobService for MemoryBlobService {
-    /// Constructs a [MemoryBlobService] from the passed [url::Url]:
-    /// - scheme has to be `memory://`
-    /// - there may not be a host.
-    /// - there may not be a path.
-    fn from_url(url: &url::Url) -> Result<Self, Error> {
-        if url.scheme() != "memory" {
-            return Err(crate::Error::StorageError("invalid scheme".to_string()));
-        }
-
-        if url.has_host() || !url.path().is_empty() {
-            return Err(crate::Error::StorageError("invalid url".to_string()));
-        }
-
-        Ok(Self::default())
-    }
-
     #[instrument(skip(self, digest), fields(blob.digest=%digest))]
     async fn has(&self, digest: &B3Digest) -> Result<bool, Error> {
         let db = self.db.read().unwrap();
@@ -148,49 +132,3 @@ impl BlobWriter for MemoryBlobWriter {
         }
     }
 }
-
-#[cfg(test)]
-mod tests {
-    use super::BlobService;
-    use super::MemoryBlobService;
-
-    /// This uses a wrong scheme.
-    #[test]
-    fn test_invalid_scheme() {
-        let url = url::Url::parse("http://foo.example/test").expect("must parse");
-
-        assert!(MemoryBlobService::from_url(&url).is_err());
-    }
-
-    /// This correctly sets the scheme, and doesn't set a path.
-    #[test]
-    fn test_valid_scheme() {
-        let url = url::Url::parse("memory://").expect("must parse");
-
-        assert!(MemoryBlobService::from_url(&url).is_ok());
-    }
-
-    /// This sets the host to `foo`
-    #[test]
-    fn test_invalid_host() {
-        let url = url::Url::parse("memory://foo").expect("must parse");
-
-        assert!(MemoryBlobService::from_url(&url).is_err());
-    }
-
-    /// This has the path "/", which is invalid.
-    #[test]
-    fn test_invalid_has_path() {
-        let url = url::Url::parse("memory:///").expect("must parse");
-
-        assert!(MemoryBlobService::from_url(&url).is_err());
-    }
-
-    /// This has the path "/foo", which is invalid.
-    #[test]
-    fn test_invalid_path2() {
-        let url = url::Url::parse("memory:///foo").expect("must parse");
-
-        assert!(MemoryBlobService::from_url(&url).is_err());
-    }
-}
diff --git a/tvix/castore/src/blobservice/mod.rs b/tvix/castore/src/blobservice/mod.rs
index 5ecf25ac13..3267116e40 100644
--- a/tvix/castore/src/blobservice/mod.rs
+++ b/tvix/castore/src/blobservice/mod.rs
@@ -24,12 +24,6 @@ pub use self::sled::SledBlobService;
 /// close funtion, to finalize a blob and get its digest.
 #[async_trait]
 pub trait BlobService: Send + Sync {
-    /// Create a new instance by passing in a connection URL.
-    /// TODO: check if we want to make this async, instead of lazily connecting
-    fn from_url(url: &url::Url) -> Result<Self, Error>
-    where
-        Self: Sized;
-
     /// Check if the service has the blob, by its content hash.
     async fn has(&self, digest: &B3Digest) -> Result<bool, Error>;
 
diff --git a/tvix/castore/src/blobservice/sled.rs b/tvix/castore/src/blobservice/sled.rs
index 101ae78698..12391ed72a 100644
--- a/tvix/castore/src/blobservice/sled.rs
+++ b/tvix/castore/src/blobservice/sled.rs
@@ -31,34 +31,6 @@ impl SledBlobService {
 
 #[async_trait]
 impl BlobService for SledBlobService {
-    /// Constructs a [SledBlobService] from the passed [url::Url]:
-    /// - scheme has to be `sled://`
-    /// - there may not be a host.
-    /// - a path to the sled needs to be provided (which may not be `/`).
-    fn from_url(url: &url::Url) -> Result<Self, Error> {
-        if url.scheme() != "sled" {
-            return Err(crate::Error::StorageError("invalid scheme".to_string()));
-        }
-
-        if url.has_host() {
-            return Err(crate::Error::StorageError(format!(
-                "invalid host: {}",
-                url.host().unwrap()
-            )));
-        }
-
-        // TODO: expose compression and other parameters as URL parameters, drop new and new_temporary?
-        if url.path().is_empty() {
-            Self::new_temporary().map_err(|e| Error::StorageError(e.to_string()))
-        } else if url.path() == "/" {
-            Err(crate::Error::StorageError(
-                "cowardly refusing to open / with sled".to_string(),
-            ))
-        } else {
-            Self::new(url.path().into()).map_err(|e| Error::StorageError(e.to_string()))
-        }
-    }
-
     #[instrument(skip(self), fields(blob.digest=%digest))]
     async fn has(&self, digest: &B3Digest) -> Result<bool, Error> {
         match self.db.contains_key(digest.as_slice()) {
@@ -173,77 +145,3 @@ impl BlobWriter for SledBlobWriter {
         }
     }
 }
-
-#[cfg(test)]
-mod tests {
-    use tempfile::TempDir;
-
-    use super::BlobService;
-    use super::SledBlobService;
-
-    /// This uses a wrong scheme.
-    #[test]
-    fn test_invalid_scheme() {
-        let url = url::Url::parse("http://foo.example/test").expect("must parse");
-
-        assert!(SledBlobService::from_url(&url).is_err());
-    }
-
-    /// This uses the correct scheme, and doesn't specify a path (temporary sled).
-    #[test]
-    fn test_valid_scheme_temporary() {
-        let url = url::Url::parse("sled://").expect("must parse");
-
-        assert!(SledBlobService::from_url(&url).is_ok());
-    }
-
-    /// This sets the path to a location that doesn't exist, which should fail (as sled doesn't mkdir -p)
-    #[test]
-    fn test_nonexistent_path() {
-        let tmpdir = TempDir::new().unwrap();
-
-        let mut url = url::Url::parse("sled://foo.example").expect("must parse");
-        url.set_path(tmpdir.path().join("foo").join("bar").to_str().unwrap());
-
-        assert!(SledBlobService::from_url(&url).is_err());
-    }
-
-    /// This uses the correct scheme, and specifies / as path (which should fail
-    // for obvious reasons)
-    #[test]
-    fn test_invalid_path_root() {
-        let url = url::Url::parse("sled:///").expect("must parse");
-
-        assert!(SledBlobService::from_url(&url).is_err());
-    }
-
-    /// This uses the correct scheme, and sets a tempdir as location.
-    #[test]
-    fn test_valid_scheme_path() {
-        let tmpdir = TempDir::new().unwrap();
-
-        let mut url = url::Url::parse("sled://").expect("must parse");
-        url.set_path(tmpdir.path().to_str().unwrap());
-
-        assert!(SledBlobService::from_url(&url).is_ok());
-    }
-
-    /// This sets a host, rather than a path, which should fail.
-    #[test]
-    fn test_invalid_host() {
-        let url = url::Url::parse("sled://foo.example").expect("must parse");
-
-        assert!(SledBlobService::from_url(&url).is_err());
-    }
-
-    /// This sets a host AND a valid path, which should fail
-    #[test]
-    fn test_invalid_host_and_path() {
-        let tmpdir = TempDir::new().unwrap();
-
-        let mut url = url::Url::parse("sled://foo.example").expect("must parse");
-        url.set_path(tmpdir.path().to_str().unwrap());
-
-        assert!(SledBlobService::from_url(&url).is_err());
-    }
-}