feat(src/proto): add PathInfo.validate() r/5570

This provides validation of PathInfo messages, and ensures the output
hashes are properly parsed from the root node names.

NixPath already has a more extensive test suite for various wrong
NixPaths, so it's omitted from here.

Change-Id: I5d69118df5816daabb521ddb19d178bddd1caacf
Reviewed-on: https://cl.tvl.fyi/c/depot/+/7684
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI

3 files changed, 307 insertions, 0 deletions

diff --git a/tvix/store/src/proto.rs b/tvix/store/src/proto.rs
index 75e5c34ab137..f2af69ab0dbb 100644
--- a/tvix/store/src/proto.rs
+++ b/tvix/store/src/proto.rs
@@ -5,6 +5,8 @@ use thiserror::Error;
 
 use prost::Message;
 
+use crate::nixpath::{NixPath, ParseNixPathError};
+
 tonic::include_proto!("tvix.store.v1");
 
 #[cfg(feature = "reflection")]
@@ -30,6 +32,27 @@ pub enum ValidateDirectoryError {
     InvalidDigestLen(usize),
 }
 
+/// Errors that can occur during the validation of PathInfo messages.
+#[derive(Debug, Error, PartialEq)]
+pub enum ValidatePathInfoError {
+    /// No node present
+    #[error("No node present")]
+    NoNodePresent(),
+
+    /// Invalid node name encountered.
+    #[error("{0} is an invalid node name: {1}")]
+    InvalidNodeName(String, ParseNixPathError),
+
+    /// The digest the (root) node refers to has invalid length.
+    #[error("Invalid Digest length: {0}")]
+    InvalidDigestLen(usize),
+
+    /// The number of references in the narinfo.reference_names field does not match
+    /// the number of references in the .references field.
+    #[error("Inconsistent Number of References: {0} (references) vs {0} (narinfo)")]
+    InconsistentNumberOfReferences(usize, usize),
+}
+
 /// Checks a Node name for validity as an intermediate node, and returns an
 /// error that's generated from the supplied constructor.
 ///
@@ -51,6 +74,82 @@ fn validate_digest<E>(digest: &Vec<u8>, err: fn(usize) -> E) -> Result<(), E> {
     Ok(())
 }
 
+/// Parses a root node name.
+///
+/// On success, this returns the parsed [NixPath].
+/// On error, it returns an error generated from the supplied constructor.
+fn parse_node_name_root<E>(
+    name: &str,
+    err: fn(String, ParseNixPathError) -> E,
+) -> Result<NixPath, E> {
+    match NixPath::from_string(name) {
+        Ok(np) => Ok(np),
+        Err(e) => Err(err(name.to_string(), e)),
+    }
+}
+
+impl PathInfo {
+    /// validate performs some checks on the PathInfo struct,
+    /// Returning either a [NixPath] of the root node, or a
+    /// [ValidatePathInfoError].
+    pub fn validate(&self) -> Result<NixPath, ValidatePathInfoError> {
+        // If there is a narinfo field populated, ensure the number of references there
+        // matches PathInfo.references count.
+        if let Some(narinfo) = &self.narinfo {
+            if narinfo.reference_names.len() != self.references.len() {
+                return Err(ValidatePathInfoError::InconsistentNumberOfReferences(
+                    narinfo.reference_names.len(),
+                    self.references.len(),
+                ));
+            }
+        }
+        // FUTUREWORK: parse references in reference_names. ensure they start
+        // with storeDir, and use the same digest as in self.references.
+
+        // Ensure there is a (root) node present, and it properly parses to a NixPath.
+        let root_nix_path = match &self.node {
+            None => {
+                return Err(ValidatePathInfoError::NoNodePresent());
+            }
+            Some(Node { node }) => match node {
+                None => {
+                    return Err(ValidatePathInfoError::NoNodePresent());
+                }
+                Some(node::Node::Directory(directory_node)) => {
+                    // ensure the digest has the appropriate size.
+                    validate_digest(
+                        &directory_node.digest,
+                        ValidatePathInfoError::InvalidDigestLen,
+                    )?;
+
+                    // parse the name
+                    parse_node_name_root(
+                        &directory_node.name,
+                        ValidatePathInfoError::InvalidNodeName,
+                    )?
+                }
+                Some(node::Node::File(file_node)) => {
+                    // ensure the digest has the appropriate size.
+                    validate_digest(&file_node.digest, ValidatePathInfoError::InvalidDigestLen)?;
+
+                    // parse the name
+                    parse_node_name_root(&file_node.name, ValidatePathInfoError::InvalidNodeName)?
+                }
+                Some(node::Node::Symlink(symlink_node)) => {
+                    // parse the name
+                    parse_node_name_root(
+                        &symlink_node.name,
+                        ValidatePathInfoError::InvalidNodeName,
+                    )?
+                }
+            },
+        };
+
+        // return the root nix path
+        Ok(root_nix_path)
+    }
+}
+
 /// Accepts a name, and a mutable reference to the previous name.
 /// If the passed name is larger than the previous one, the reference is updated.
 /// If it's not, an error is returned.
diff --git a/tvix/store/src/tests/mod.rs b/tvix/store/src/tests/mod.rs
index 366f96287a4c..b5fb0648e632 100644
--- a/tvix/store/src/tests/mod.rs
+++ b/tvix/store/src/tests/mod.rs
@@ -1,6 +1,8 @@
 use crate::proto::{Directory, DirectoryNode, FileNode, SymlinkNode, ValidateDirectoryError};
 use lazy_static::lazy_static;
 
+mod pathinfo;
+
 lazy_static! {
     static ref DUMMY_DIGEST: Vec<u8> = vec![
         0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
diff --git a/tvix/store/src/tests/pathinfo.rs b/tvix/store/src/tests/pathinfo.rs
new file mode 100644
index 000000000000..8f4e9d399084
--- /dev/null
+++ b/tvix/store/src/tests/pathinfo.rs
@@ -0,0 +1,206 @@
+use crate::{
+    nixpath::{NixPath, ParseNixPathError},
+    proto::{self, Node, PathInfo, ValidatePathInfoError},
+};
+use lazy_static::lazy_static;
+use test_case::test_case;
+
+lazy_static! {
+    static ref DUMMY_DIGEST: Vec<u8> = vec![
+        0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+        0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+        0x00, 0x00,
+    ];
+    static ref DUMMY_DIGEST_2: Vec<u8> = vec![
+        0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+        0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+        0x00, 0x00,
+    ];
+}
+
+const DUMMY_NAME: &str = "00000000000000000000000000000000-dummy";
+
+#[test_case(
+    None,
+    Err(ValidatePathInfoError::NoNodePresent()) ;
+    "No node"
+)]
+#[test_case(
+    Some(Node { node: None }),
+    Err(ValidatePathInfoError::NoNodePresent());
+    "No node 2"
+)]
+fn validate_no_node(t_node: Option<proto::Node>, t_result: Result<NixPath, ValidatePathInfoError>) {
+    // construct the PathInfo object
+    let p = PathInfo {
+        node: t_node,
+        ..Default::default()
+    };
+    assert_eq!(t_result, p.validate());
+}
+
+#[test_case(
+    proto::DirectoryNode {
+        name: DUMMY_NAME.to_string(),
+        digest: DUMMY_DIGEST.to_vec(),
+        size: 0,
+    },
+    Ok(NixPath::from_string(DUMMY_NAME).expect("must succeed"));
+    "ok"
+)]
+#[test_case(
+    proto::DirectoryNode {
+        name: DUMMY_NAME.to_string(),
+        digest: vec![],
+        size: 0,
+    },
+    Err(ValidatePathInfoError::InvalidDigestLen(0));
+    "invalid digest length"
+)]
+#[test_case(
+    proto::DirectoryNode {
+        name: "invalid".to_string(),
+        digest: DUMMY_DIGEST.to_vec(),
+        size: 0,
+    },
+    Err(ValidatePathInfoError::InvalidNodeName(
+        "invalid".to_string(),
+        ParseNixPathError::InvalidName("".to_string())
+    ));
+    "invalid node name"
+)]
+fn validate_directory(
+    t_directory_node: proto::DirectoryNode,
+    t_result: Result<NixPath, ValidatePathInfoError>,
+) {
+    // construct the PathInfo object
+    let p = PathInfo {
+        node: Some(Node {
+            node: Some(proto::node::Node::Directory(t_directory_node)),
+        }),
+        ..Default::default()
+    };
+    assert_eq!(t_result, p.validate());
+}
+
+#[test_case(
+    proto::FileNode {
+        name: DUMMY_NAME.to_string(),
+        digest: DUMMY_DIGEST.to_vec(),
+        size: 0,
+        executable: false,
+    },
+    Ok(NixPath::from_string(DUMMY_NAME).expect("must succeed"));
+    "ok"
+)]
+#[test_case(
+    proto::FileNode {
+        name: DUMMY_NAME.to_string(),
+        digest: vec![],
+        ..Default::default()
+    },
+    Err(ValidatePathInfoError::InvalidDigestLen(0));
+    "invalid digest length"
+)]
+#[test_case(
+    proto::FileNode {
+        name: "invalid".to_string(),
+        digest: DUMMY_DIGEST.to_vec(),
+        ..Default::default()
+    },
+    Err(ValidatePathInfoError::InvalidNodeName(
+        "invalid".to_string(),
+        ParseNixPathError::InvalidName("".to_string())
+    ));
+    "invalid node name"
+)]
+fn validate_file(t_file_node: proto::FileNode, t_result: Result<NixPath, ValidatePathInfoError>) {
+    // construct the PathInfo object
+    let p = PathInfo {
+        node: Some(Node {
+            node: Some(proto::node::Node::File(t_file_node)),
+        }),
+        ..Default::default()
+    };
+    assert_eq!(t_result, p.validate());
+}
+
+#[test_case(
+    proto::SymlinkNode {
+        name: DUMMY_NAME.to_string(),
+        ..Default::default()
+    },
+    Ok(NixPath::from_string(DUMMY_NAME).expect("must succeed"));
+    "ok"
+)]
+#[test_case(
+    proto::SymlinkNode {
+        name: "invalid".to_string(),
+        ..Default::default()
+    },
+    Err(ValidatePathInfoError::InvalidNodeName(
+        "invalid".to_string(),
+        ParseNixPathError::InvalidName("".to_string())
+    ));
+    "invalid node name"
+)]
+fn validate_symlink(
+    t_symlink_node: proto::SymlinkNode,
+    t_result: Result<NixPath, ValidatePathInfoError>,
+) {
+    // construct the PathInfo object
+    let p = PathInfo {
+        node: Some(Node {
+            node: Some(proto::node::Node::Symlink(t_symlink_node)),
+        }),
+        ..Default::default()
+    };
+    assert_eq!(t_result, p.validate());
+}
+
+#[test]
+fn validate_references() {
+    // create a PathInfo without narinfo field.
+    let path_info = PathInfo {
+        node: Some(Node {
+            node: Some(proto::node::Node::Directory(proto::DirectoryNode {
+                name: DUMMY_NAME.to_string(),
+                digest: DUMMY_DIGEST.to_vec(),
+                size: 0,
+            })),
+        }),
+        references: vec![DUMMY_DIGEST_2.to_vec()],
+        narinfo: None,
+    };
+    assert!(path_info.validate().is_ok());
+
+    // create a PathInfo with a narinfo field, but an inconsistent set of references
+    let path_info_with_narinfo_missing_refs = PathInfo {
+        narinfo: Some(proto::NarInfo {
+            nar_size: 0,
+            nar_sha256: DUMMY_DIGEST.to_vec(),
+            signatures: vec![],
+            reference_names: vec![],
+        }),
+        ..path_info.clone()
+    };
+    match path_info_with_narinfo_missing_refs
+        .validate()
+        .expect_err("must_fail")
+    {
+        ValidatePathInfoError::InconsistentNumberOfReferences(_, _) => {}
+        _ => panic!("unexpected error"),
+    };
+
+    // create a pathinfo with the correct number of references, should suceed
+    let path_info_with_narinfo = PathInfo {
+        narinfo: Some(proto::NarInfo {
+            nar_size: 0,
+            nar_sha256: DUMMY_DIGEST.to_vec(),
+            signatures: vec![],
+            reference_names: vec![format!("/nix/store/{}", DUMMY_NAME)],
+        }),
+        ..path_info.clone()
+    };
+    assert!(path_info_with_narinfo.validate().is_ok());
+}