feat(tvix/store): add pathinfoservice r/5919

This adds a PathInfoService trait, and an implementation for it using
sled, and one using a HashMap.

Change-Id: I85fe4c9b6105808d7b0c095441326424ffc2d2bb
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8099
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>

4 files changed, 165 insertions, 0 deletions

diff --git a/tvix/store/src/lib.rs b/tvix/store/src/lib.rs
index 0279d13f9b..c275be5ff0 100644
--- a/tvix/store/src/lib.rs
+++ b/tvix/store/src/lib.rs
@@ -6,6 +6,7 @@ mod errors;
 pub mod blobservice;
 pub mod chunkservice;
 pub mod directoryservice;
+pub mod pathinfoservice;
 pub mod proto;
 
 pub use blobreader::BlobReader;
diff --git a/tvix/store/src/pathinfoservice/memory.rs b/tvix/store/src/pathinfoservice/memory.rs
new file mode 100644
index 0000000000..db31a2d771
--- /dev/null
+++ b/tvix/store/src/pathinfoservice/memory.rs
@@ -0,0 +1,62 @@
+use std::{
+    collections::HashMap,
+    sync::{Arc, RwLock},
+};
+
+use crate::{proto, Error};
+use nix_compat::store_path::DIGEST_SIZE;
+
+use super::PathInfoService;
+
+pub struct MemoryPathInfoService {
+    db: Arc<RwLock<HashMap<Vec<u8>, proto::PathInfo>>>,
+}
+
+impl MemoryPathInfoService {
+    pub fn new() -> Self {
+        let db = Arc::new(RwLock::new(HashMap::default()));
+
+        Self { db }
+    }
+}
+
+impl PathInfoService for MemoryPathInfoService {
+    fn get(
+        &self,
+        by_what: proto::get_path_info_request::ByWhat,
+    ) -> Result<Option<proto::PathInfo>, Error> {
+        match by_what {
+            proto::get_path_info_request::ByWhat::ByOutputHash(digest) => {
+                if digest.len() != DIGEST_SIZE {
+                    return Err(Error::InvalidRequest("invalid digest length".to_string()));
+                }
+
+                let db = self.db.read().unwrap();
+
+                match db.get(&digest) {
+                    None => Ok(None),
+                    Some(path_info) => Ok(Some(path_info.clone())),
+                }
+            }
+        }
+    }
+
+    fn put(&self, path_info: proto::PathInfo) -> Result<proto::PathInfo, Error> {
+        // Call validate on the received PathInfo message.
+        match path_info.validate() {
+            Err(e) => Err(Error::InvalidRequest(format!(
+                "failed to validate PathInfo: {}",
+                e
+            ))),
+
+            // In case the PathInfo is valid, and we were able to extract a NixPath, store it in the database.
+            // This overwrites existing PathInfo objects.
+            Ok(nix_path) => {
+                let mut db = self.db.write().unwrap();
+                db.insert(nix_path.digest.to_vec(), path_info.clone());
+
+                Ok(path_info)
+            }
+        }
+    }
+}
diff --git a/tvix/store/src/pathinfoservice/mod.rs b/tvix/store/src/pathinfoservice/mod.rs
new file mode 100644
index 0000000000..b20985dd80
--- /dev/null
+++ b/tvix/store/src/pathinfoservice/mod.rs
@@ -0,0 +1,21 @@
+mod memory;
+mod sled;
+
+use crate::{proto, Error};
+
+pub use self::memory::MemoryPathInfoService;
+pub use self::sled::SledPathInfoService;
+
+/// The base trait all PathInfo services need to implement.
+/// This is a simple get and put of [proto::Directory], returning their digest.
+pub trait PathInfoService {
+    /// Retrieve a PathInfo message.
+    fn get(
+        &self,
+        by_what: proto::get_path_info_request::ByWhat,
+    ) -> Result<Option<proto::PathInfo>, Error>;
+
+    /// Store a PathInfo message. Implementations MUST call validate and reject
+    /// invalid messages.
+    fn put(&self, path_info: proto::PathInfo) -> Result<proto::PathInfo, Error>;
+}
diff --git a/tvix/store/src/pathinfoservice/sled.rs b/tvix/store/src/pathinfoservice/sled.rs
new file mode 100644
index 0000000000..dc3c95b510
--- /dev/null
+++ b/tvix/store/src/pathinfoservice/sled.rs
@@ -0,0 +1,81 @@
+use crate::{proto, Error};
+use nix_compat::store_path::DIGEST_SIZE;
+use prost::Message;
+use std::path::PathBuf;
+use tracing::warn;
+
+use super::PathInfoService;
+
+/// SledPathInfoService stores PathInfo in a [sled](https://github.com/spacejam/sled).
+///
+/// The PathInfo messages are stored as encoded protos, and keyed by their output hash,
+/// as that's currently the only request type available.
+pub struct SledPathInfoService {
+    db: sled::Db,
+}
+
+impl SledPathInfoService {
+    pub fn new(p: PathBuf) -> Result<Self, sled::Error> {
+        let config = sled::Config::default().use_compression(true).path(p);
+        let db = config.open()?;
+
+        Ok(Self { db })
+    }
+}
+
+impl PathInfoService for SledPathInfoService {
+    fn get(
+        &self,
+        by_what: proto::get_path_info_request::ByWhat,
+    ) -> Result<Option<proto::PathInfo>, Error> {
+        match by_what {
+            proto::get_path_info_request::ByWhat::ByOutputHash(digest) => {
+                if digest.len() != DIGEST_SIZE {
+                    return Err(Error::InvalidRequest("invalid digest length".to_string()));
+                }
+
+                match self.db.get(digest) {
+                    Ok(None) => Ok(None),
+                    Ok(Some(data)) => match proto::PathInfo::decode(&*data) {
+                        Ok(path_info) => Ok(Some(path_info)),
+                        Err(e) => {
+                            warn!("failed to decode stored PathInfo: {}", e);
+                            Err(Error::StorageError(format!(
+                                "failed to decode stored PathInfo: {}",
+                                e
+                            )))
+                        }
+                    },
+                    Err(e) => {
+                        warn!("failed to retrieve PathInfo: {}", e);
+                        Err(Error::StorageError(format!(
+                            "failed to retrieve PathInfo: {}",
+                            e
+                        )))
+                    }
+                }
+            }
+        }
+    }
+
+    fn put(&self, path_info: proto::PathInfo) -> Result<proto::PathInfo, Error> {
+        // Call validate on the received PathInfo message.
+        match path_info.validate() {
+            Err(e) => Err(Error::InvalidRequest(format!(
+                "failed to validate PathInfo: {}",
+                e
+            ))),
+            // In case the PathInfo is valid, and we were able to extract a NixPath, store it in the database.
+            // This overwrites existing PathInfo objects.
+            Ok(nix_path) => match self.db.insert(nix_path.digest, path_info.encode_to_vec()) {
+                Ok(_) => Ok(path_info),
+                Err(e) => {
+                    warn!("failed to insert PathInfo: {}", e);
+                    Err(Error::StorageError(format! {
+                        "failed to insert PathInfo: {}", e
+                    }))
+                }
+            },
+        }
+    }
+}