refactor(tvix): move castore into tvix-castore crate r/6629

This splits the pure content-addressed layers from tvix-store into a
`castore` crate, and only leaves PathInfo related things, as well as the
CLI entrypoint in the tvix-store crate.

Notable changes:
 - `fixtures` and `utils` had to be moved out of the `test` cfg, so they
   can be imported from tvix-store.
 - Some ad-hoc fixtures in the test were moved to proper fixtures in the
   same step.
 - The protos are now created by a (more static) recipe in the protos/
   directory.

The (now two) golang targets are commented out, as it's not possible to
update them properly in the same CL. This will be done by a followup CL
once this is merged (and whitby deployed)

Bug: https://b.tvl.fyi/issues/301

Change-Id: I8d675d4bf1fb697eb7d479747c1b1e3635718107
Reviewed-on: https://cl.tvl.fyi/c/depot/+/9370
Reviewed-by: tazjin <tazjin@tvl.su>
Reviewed-by: flokli <flokli@flokli.de>
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Reviewed-by: Connor Brewster <cbrewster@hey.com>

4 files changed, 68 insertions, 63 deletions

diff --git a/tvix/store/src/fs/inode_tracker.rs b/tvix/store/src/fs/inode_tracker.rs
index ad1ef859a2..daf6b4ee79 100644
--- a/tvix/store/src/fs/inode_tracker.rs
+++ b/tvix/store/src/fs/inode_tracker.rs
@@ -1,8 +1,8 @@
 use std::{collections::HashMap, sync::Arc};
 
-use crate::{proto, B3Digest};
-
 use super::inodes::{DirectoryInodeData, InodeData};
+use tvix_castore::proto as castorepb;
+use tvix_castore::B3Digest;
 
 /// InodeTracker keeps track of inodes, stores data being these inodes and deals
 /// with inode allocation.
@@ -139,21 +139,21 @@ impl InodeTracker {
     // Consume a list of children with zeroed inodes, and allocate (or fetch existing) inodes.
     fn allocate_inodes_for_children(
         &mut self,
-        children: Vec<(u64, proto::node::Node)>,
-    ) -> Vec<(u64, proto::node::Node)> {
+        children: Vec<(u64, castorepb::node::Node)>,
+    ) -> Vec<(u64, castorepb::node::Node)> {
         // allocate new inodes for all children
-        let mut children_new: Vec<(u64, proto::node::Node)> = Vec::new();
+        let mut children_new: Vec<(u64, castorepb::node::Node)> = Vec::new();
 
         for (child_ino, ref child_node) in children {
             debug_assert_eq!(0, child_ino, "expected child inode to be 0");
             let child_ino = match child_node {
-                proto::node::Node::Directory(directory_node) => {
+                castorepb::node::Node::Directory(directory_node) => {
                     // Try putting the sparse data in. If we already have a
                     // populated version, it'll not update it.
                     self.put(directory_node.into())
                 }
-                proto::node::Node::File(file_node) => self.put(file_node.into()),
-                proto::node::Node::Symlink(symlink_node) => self.put(symlink_node.into()),
+                castorepb::node::Node::File(file_node) => self.put(file_node.into()),
+                castorepb::node::Node::Symlink(symlink_node) => self.put(symlink_node.into()),
             };
 
             children_new.push((child_ino, child_node.clone()))
@@ -198,8 +198,8 @@ impl InodeTracker {
 #[cfg(test)]
 mod tests {
     use crate::fs::inodes::DirectoryInodeData;
-    use crate::proto;
     use crate::tests::fixtures;
+    use tvix_castore::proto as castorepb;
 
     use super::InodeData;
     use super::InodeTracker;
@@ -304,7 +304,7 @@ mod tests {
                 let (child_ino, child_node) = children.first().unwrap();
                 assert_ne!(dir_ino, *child_ino);
                 assert_eq!(
-                    &proto::node::Node::File(
+                    &castorepb::node::Node::File(
                         fixtures::DIRECTORY_WITH_KEEP.files.first().unwrap().clone()
                     ),
                     child_node
@@ -362,7 +362,9 @@ mod tests {
                     let (child_ino, child_node) = &children[0];
                     assert!(!seen_inodes.contains(child_ino));
                     assert_eq!(
-                        &proto::node::Node::File(fixtures::DIRECTORY_COMPLICATED.files[0].clone()),
+                        &castorepb::node::Node::File(
+                            fixtures::DIRECTORY_COMPLICATED.files[0].clone()
+                        ),
                         child_node
                     );
                     seen_inodes.push(*child_ino);
@@ -373,7 +375,7 @@ mod tests {
                     let (child_ino, child_node) = &children[1];
                     assert!(!seen_inodes.contains(child_ino));
                     assert_eq!(
-                        &proto::node::Node::Symlink(
+                        &castorepb::node::Node::Symlink(
                             fixtures::DIRECTORY_COMPLICATED.symlinks[0].clone()
                         ),
                         child_node
@@ -386,7 +388,7 @@ mod tests {
                     let (child_ino, child_node) = &children[2];
                     assert!(!seen_inodes.contains(child_ino));
                     assert_eq!(
-                        &proto::node::Node::Directory(
+                        &castorepb::node::Node::Directory(
                             fixtures::DIRECTORY_COMPLICATED.directories[0].clone()
                         ),
                         child_node
@@ -439,7 +441,7 @@ mod tests {
                 let (child_node_inode, child_node) = children.first().unwrap();
                 assert_ne!(dir_complicated_ino, *child_node_inode);
                 assert_eq!(
-                    &proto::node::Node::File(
+                    &castorepb::node::Node::File(
                         fixtures::DIRECTORY_WITH_KEEP.files.first().unwrap().clone()
                     ),
                     child_node
diff --git a/tvix/store/src/fs/inodes.rs b/tvix/store/src/fs/inodes.rs
index e8959ce362..928f510590 100644
--- a/tvix/store/src/fs/inodes.rs
+++ b/tvix/store/src/fs/inodes.rs
@@ -1,6 +1,7 @@
 //! This module contains all the data structures used to track information
 //! about inodes, which present tvix-store nodes in a filesystem.
-use crate::{proto, B3Digest};
+use tvix_castore::proto as castorepb;
+use tvix_castore::B3Digest;
 
 #[derive(Clone, Debug)]
 pub enum InodeData {
@@ -10,33 +11,33 @@ pub enum InodeData {
 }
 
 /// This encodes the two different states of [InodeData::Directory].
-/// Either the data still is sparse (we only saw a [proto::DirectoryNode], but
-/// didn't fetch the [proto::Directory] struct yet,
-/// or we processed a lookup and did fetch the data.
+/// Either the data still is sparse (we only saw a [castorepb::DirectoryNode],
+/// but didn't fetch the [castorepb::Directory] struct yet, or we processed a
+/// lookup and did fetch the data.
 #[derive(Clone, Debug)]
 pub enum DirectoryInodeData {
-    Sparse(B3Digest, u32),                              // digest, size
-    Populated(B3Digest, Vec<(u64, proto::node::Node)>), // [(child_inode, node)]
+    Sparse(B3Digest, u32),                                  // digest, size
+    Populated(B3Digest, Vec<(u64, castorepb::node::Node)>), // [(child_inode, node)]
 }
 
-impl From<&proto::node::Node> for InodeData {
-    fn from(value: &proto::node::Node) -> Self {
+impl From<&castorepb::node::Node> for InodeData {
+    fn from(value: &castorepb::node::Node) -> Self {
         match value {
-            proto::node::Node::Directory(directory_node) => directory_node.into(),
-            proto::node::Node::File(file_node) => file_node.into(),
-            proto::node::Node::Symlink(symlink_node) => symlink_node.into(),
+            castorepb::node::Node::Directory(directory_node) => directory_node.into(),
+            castorepb::node::Node::File(file_node) => file_node.into(),
+            castorepb::node::Node::Symlink(symlink_node) => symlink_node.into(),
         }
     }
 }
 
-impl From<&proto::SymlinkNode> for InodeData {
-    fn from(value: &proto::SymlinkNode) -> Self {
+impl From<&castorepb::SymlinkNode> for InodeData {
+    fn from(value: &castorepb::SymlinkNode) -> Self {
         InodeData::Symlink(value.target.clone())
     }
 }
 
-impl From<&proto::FileNode> for InodeData {
-    fn from(value: &proto::FileNode) -> Self {
+impl From<&castorepb::FileNode> for InodeData {
+    fn from(value: &castorepb::FileNode) -> Self {
         InodeData::Regular(
             value.digest.clone().try_into().unwrap(),
             value.size,
@@ -46,8 +47,8 @@ impl From<&proto::FileNode> for InodeData {
 }
 
 /// Converts a DirectoryNode to a sparsely populated InodeData::Directory.
-impl From<&proto::DirectoryNode> for InodeData {
-    fn from(value: &proto::DirectoryNode) -> Self {
+impl From<&castorepb::DirectoryNode> for InodeData {
+    fn from(value: &castorepb::DirectoryNode) -> Self {
         InodeData::Directory(DirectoryInodeData::Sparse(
             value.digest.clone().try_into().unwrap(),
             value.size,
@@ -57,11 +58,12 @@ impl From<&proto::DirectoryNode> for InodeData {
 
 /// converts a proto::Directory to a InodeData::Directory(DirectoryInodeData::Populated(..)).
 /// The inodes for each child are 0, because it's up to the InodeTracker to allocate them.
-impl From<proto::Directory> for InodeData {
-    fn from(value: proto::Directory) -> Self {
+impl From<castorepb::Directory> for InodeData {
+    fn from(value: castorepb::Directory) -> Self {
         let digest = value.digest();
 
-        let children: Vec<(u64, proto::node::Node)> = value.nodes().map(|node| (0, node)).collect();
+        let children: Vec<(u64, castorepb::node::Node)> =
+            value.nodes().map(|node| (0, node)).collect();
 
         InodeData::Directory(DirectoryInodeData::Populated(digest, children))
     }
diff --git a/tvix/store/src/fs/mod.rs b/tvix/store/src/fs/mod.rs
index 02d3bb3221..59b8f0d085 100644
--- a/tvix/store/src/fs/mod.rs
+++ b/tvix/store/src/fs/mod.rs
@@ -8,13 +8,8 @@ pub mod fuse;
 #[cfg(test)]
 mod tests;
 
-use crate::{
-    blobservice::{BlobReader, BlobService},
-    directoryservice::DirectoryService,
-    pathinfoservice::PathInfoService,
-    proto::{node::Node, NamedNode},
-    B3Digest, Error,
-};
+use crate::pathinfoservice::PathInfoService;
+
 use fuse_backend_rs::api::filesystem::{Context, FileSystem, FsOptions, ROOT_ID};
 use futures::StreamExt;
 use nix_compat::store_path::StorePath;
@@ -32,6 +27,12 @@ use tokio::{
     sync::mpsc,
 };
 use tracing::{debug, info_span, warn};
+use tvix_castore::{
+    blobservice::{BlobReader, BlobService},
+    directoryservice::DirectoryService,
+    proto::{node::Node, NamedNode},
+    B3Digest, Error,
+};
 
 use self::{
     file_attr::{gen_file_attr, ROOT_FILE_ATTR},
diff --git a/tvix/store/src/fs/tests.rs b/tvix/store/src/fs/tests.rs
index 6837f8aa29..2adea0ceb3 100644
--- a/tvix/store/src/fs/tests.rs
+++ b/tvix/store/src/fs/tests.rs
@@ -5,17 +5,17 @@ use std::path::Path;
 use std::sync::Arc;
 use tokio::{fs, io};
 use tokio_stream::wrappers::ReadDirStream;
+use tvix_castore::blobservice::BlobService;
+use tvix_castore::directoryservice::DirectoryService;
 
 use tempfile::TempDir;
 
-use crate::blobservice::BlobService;
-use crate::directoryservice::DirectoryService;
 use crate::fs::{fuse::FuseDaemon, TvixStoreFs};
 use crate::pathinfoservice::PathInfoService;
-use crate::proto;
-use crate::proto::{DirectoryNode, FileNode, PathInfo};
+use crate::proto::PathInfo;
 use crate::tests::fixtures;
 use crate::tests::utils::{gen_blob_service, gen_directory_service, gen_pathinfo_service};
+use tvix_castore::proto as castorepb;
 
 const BLOB_A_NAME: &str = "00000000000000000000000000000000-test";
 const BLOB_B_NAME: &str = "55555555555555555555555555555555-test";
@@ -67,8 +67,8 @@ async fn populate_blob_a(
 
     // Create a PathInfo for it
     let path_info = PathInfo {
-        node: Some(proto::Node {
-            node: Some(proto::node::Node::File(FileNode {
+        node: Some(castorepb::Node {
+            node: Some(castorepb::node::Node::File(castorepb::FileNode {
                 name: BLOB_A_NAME.into(),
                 digest: fixtures::BLOB_A_DIGEST.clone().into(),
                 size: fixtures::BLOB_A.len() as u32,
@@ -97,8 +97,8 @@ async fn populate_blob_b(
 
     // Create a PathInfo for it
     let path_info = PathInfo {
-        node: Some(proto::Node {
-            node: Some(proto::node::Node::File(FileNode {
+        node: Some(castorepb::Node {
+            node: Some(castorepb::node::Node::File(castorepb::FileNode {
                 name: BLOB_B_NAME.into(),
                 digest: fixtures::BLOB_B_DIGEST.clone().into(),
                 size: fixtures::BLOB_B.len() as u32,
@@ -131,8 +131,8 @@ async fn populate_helloworld_blob(
 
     // Create a PathInfo for it
     let path_info = PathInfo {
-        node: Some(proto::Node {
-            node: Some(proto::node::Node::File(FileNode {
+        node: Some(castorepb::Node {
+            node: Some(castorepb::node::Node::File(castorepb::FileNode {
                 name: HELLOWORLD_BLOB_NAME.into(),
                 digest: fixtures::HELLOWORLD_BLOB_DIGEST.clone().into(),
                 size: fixtures::HELLOWORLD_BLOB_CONTENTS.len() as u32,
@@ -154,8 +154,8 @@ async fn populate_symlink(
 ) {
     // Create a PathInfo for it
     let path_info = PathInfo {
-        node: Some(proto::Node {
-            node: Some(proto::node::Node::Symlink(proto::SymlinkNode {
+        node: Some(castorepb::Node {
+            node: Some(castorepb::node::Node::Symlink(castorepb::SymlinkNode {
                 name: SYMLINK_NAME.into(),
                 target: BLOB_A_NAME.into(),
             })),
@@ -177,8 +177,8 @@ async fn populate_symlink2(
 ) {
     // Create a PathInfo for it
     let path_info = PathInfo {
-        node: Some(proto::Node {
-            node: Some(proto::node::Node::Symlink(proto::SymlinkNode {
+        node: Some(castorepb::Node {
+            node: Some(castorepb::node::Node::Symlink(castorepb::SymlinkNode {
                 name: SYMLINK_NAME2.into(),
                 target: "/nix/store/somewhereelse".into(),
             })),
@@ -211,8 +211,8 @@ async fn populate_directory_with_keep(
 
     // upload pathinfo
     let path_info = PathInfo {
-        node: Some(proto::Node {
-            node: Some(proto::node::Node::Directory(DirectoryNode {
+        node: Some(castorepb::Node {
+            node: Some(castorepb::node::Node::Directory(castorepb::DirectoryNode {
                 name: DIRECTORY_WITH_KEEP_NAME.into(),
                 digest: fixtures::DIRECTORY_WITH_KEEP.digest().into(),
                 size: fixtures::DIRECTORY_WITH_KEEP.size(),
@@ -235,8 +235,8 @@ async fn populate_pathinfo_without_directory(
 ) {
     // upload pathinfo
     let path_info = PathInfo {
-        node: Some(proto::Node {
-            node: Some(proto::node::Node::Directory(DirectoryNode {
+        node: Some(castorepb::Node {
+            node: Some(castorepb::node::Node::Directory(castorepb::DirectoryNode {
                 name: DIRECTORY_WITH_KEEP_NAME.into(),
                 digest: fixtures::DIRECTORY_WITH_KEEP.digest().into(),
                 size: fixtures::DIRECTORY_WITH_KEEP.size(),
@@ -258,8 +258,8 @@ async fn populate_blob_a_without_blob(
 ) {
     // Create a PathInfo for blob A
     let path_info = PathInfo {
-        node: Some(proto::Node {
-            node: Some(proto::node::Node::File(FileNode {
+        node: Some(castorepb::Node {
+            node: Some(castorepb::node::Node::File(castorepb::FileNode {
                 name: BLOB_A_NAME.into(),
                 digest: fixtures::BLOB_A_DIGEST.clone().into(),
                 size: fixtures::BLOB_A.len() as u32,
@@ -300,8 +300,8 @@ async fn populate_directory_complicated(
 
     // upload pathinfo
     let path_info = PathInfo {
-        node: Some(proto::Node {
-            node: Some(proto::node::Node::Directory(DirectoryNode {
+        node: Some(castorepb::Node {
+            node: Some(castorepb::node::Node::Directory(castorepb::DirectoryNode {
                 name: DIRECTORY_COMPLICATED_NAME.into(),
                 digest: fixtures::DIRECTORY_COMPLICATED.digest().into(),
                 size: fixtures::DIRECTORY_COMPLICATED.size(),