Add a cached fs abstraction

crates/common/Cargo.toml

@@ -10,5 +10,7 @@ rust-version.workspace = true
 
 [dependencies]
 anyhow = { workspace = true }
+moka = { workspace = true, features = ["sync"] }
+once_cell = { workspace = true }
 semver = { workspace = true }
 tokio = { workspace = true, default-features = false, features = ["time"] }

crates/common/src/cached_fs/mod.rs

@@ -0,0 +1,49 @@
+//! This module implements a cached file system allowing for results to be stored in-memory rather
+//! rather being queried from the file system again.
+
+use std::fs;
+use std::io::{Error, Result};
+use std::path::{Path, PathBuf};
+
+use moka::sync::Cache;
+use once_cell::sync::Lazy;
+
+pub fn read(path: impl AsRef<Path>) -> Result<Vec<u8>> {
+    static READ_CACHE: Lazy<Cache<PathBuf, Vec<u8>>> = Lazy::new(|| Cache::new(10_000));
+
+    let path = path.as_ref().canonicalize()?;
+    match READ_CACHE.get(path.as_path()) {
+        Some(content) => Ok(content),
+        None => {
+            let content = fs::read(path.as_path())?;
+            READ_CACHE.insert(path, content.clone());
+            Ok(content)
+        }
+    }
+}
+
+pub fn read_to_string(path: impl AsRef<Path>) -> Result<String> {
+    let content = read(path)?;
+    String::from_utf8(content).map_err(|_| {
+        Error::new(
+            std::io::ErrorKind::InvalidData,
+            "The contents of the file are not valid UTF8",
+        )
+    })
+}
+
+pub fn read_dir(path: impl AsRef<Path>) -> Result<Box<dyn Iterator<Item = Result<PathBuf>>>> {
+    static READ_DIR_CACHE: Lazy<Cache<PathBuf, Vec<PathBuf>>> = Lazy::new(|| Cache::new(10_000));
+
+    let path = path.as_ref().canonicalize()?;
+    match READ_DIR_CACHE.get(path.as_path()) {
+        Some(entries) => Ok(Box::new(entries.into_iter().map(Ok)) as Box<_>),
+        None => {
+            let entries = fs::read_dir(path.as_path())?
+                .flat_map(|maybe_entry| maybe_entry.map(|entry| entry.path()))
+                .collect();
+            READ_DIR_CACHE.insert(path.clone(), entries);
+            Ok(read_dir(path).unwrap())
+        }
+    }
+}

crates/common/src/iterators/files_with_extension_iterator.rs

@@ -19,6 +19,11 @@ pub struct FilesWithExtensionIterator {
     /// this vector then they will be returned when the [`Iterator::next`] method is called. If not
     /// then we visit one of the next directories to visit.
     files_matching_allowed_extensions: Vec<PathBuf>,
+
+    /// This option controls if the the cached file system should be used or not. This could be
+    /// better for certain cases where the entries in the directories do not change and therefore
+    /// caching can be used.
+    use_cached_fs: bool,
 }
 
 impl FilesWithExtensionIterator {
@@ -27,6 +32,7 @@ impl FilesWithExtensionIterator {
             allowed_extensions: Default::default(),
             directories_to_search: vec![root_directory.as_ref().to_path_buf()],
             files_matching_allowed_extensions: Default::default(),
+            use_cached_fs: Default::default(),
         }
     }
 
@@ -37,6 +43,11 @@ impl FilesWithExtensionIterator {
         self.allowed_extensions.insert(allowed_extension.into());
         self
     }
+
+    pub fn with_use_cached_fs(mut self, use_cached_fs: bool) -> Self {
+        self.use_cached_fs = use_cached_fs;
+        self
+    }
 }
 
 impl Iterator for FilesWithExtensionIterator {
@@ -49,16 +60,19 @@ impl Iterator for FilesWithExtensionIterator {
 
         let directory_to_search = self.directories_to_search.pop()?;
 
-        // Read all of the entries in the directory. If we failed to read this dir's entires then we
-        // elect to just ignore it and look in the next directory, we do that by calling the next
-        // method again on the iterator, which is an intentional decision that we made here instead
-        // of panicking.
-        let Ok(dir_entries) = std::fs::read_dir(directory_to_search) else {
-            return self.next();
+        let iterator = if self.use_cached_fs {
+            let Ok(dir_entries) = crate::cached_fs::read_dir(directory_to_search.as_path()) else {
+                return self.next();
+            };
+            Box::new(dir_entries) as Box<dyn Iterator<Item = std::io::Result<PathBuf>>>
+        } else {
+            let Ok(dir_entries) = std::fs::read_dir(directory_to_search) else {
+                return self.next();
+            };
+            Box::new(dir_entries.map(|maybe_entry| maybe_entry.map(|entry| entry.path()))) as Box<_>
         };
 
-        for entry in dir_entries.flatten() {
-            let entry_path = entry.path();
+        for entry_path in iterator.flatten() {
             if entry_path.is_dir() {
                 self.directories_to_search.push(entry_path)
             } else if entry_path.is_file()

crates/common/src/lib.rs

@@ -1,6 +1,7 @@
 //! This crate provides common concepts, functionality, types, macros, and more that other crates in
 //! the workspace can benefit from.
 
+pub mod cached_fs;
 pub mod fs;
 pub mod futures;
 pub mod iterators;

crates/compiler/src/lib.rs

@@ -5,7 +5,6 @@
 
 use std::{
     collections::HashMap,
-    fs::read_to_string,
     hash::Hash,
     path::{Path, PathBuf},
 };
@@ -16,6 +15,7 @@ use semver::Version;
 use serde::{Deserialize, Serialize};
 
 use revive_common::EVMVersion;
+use revive_dt_common::cached_fs::read_to_string;
 use revive_dt_common::types::VersionOrRequirement;
 use revive_dt_config::Arguments;
 

crates/core/src/main.rs

@@ -714,6 +714,7 @@ async fn compile_contracts<P: Platform>(
         // library.
         compiler = FilesWithExtensionIterator::new(metadata.directory()?)
             .with_allowed_extension("sol")
+            .with_use_cached_fs(true)
             .fold(compiler, |compiler, path| {
                 compiler.with_library(&path, library_ident.as_str(), *library_address)
             });

crates/format/src/corpus.rs

@@ -3,6 +3,7 @@ use std::{
     path::{Path, PathBuf},
 };
 
+use revive_dt_common::cached_fs::read_dir;
 use serde::{Deserialize, Serialize};
 
 use crate::metadata::MetadataFile;
@@ -54,7 +55,7 @@ impl Corpus {
 /// `path` is expected to be a directory.
 pub fn collect_metadata(path: &Path, tests: &mut Vec<MetadataFile>) {
     if path.is_dir() {
-        let dir_entry = match std::fs::read_dir(path) {
+        let dir_entry = match read_dir(path) {
             Ok(dir_entry) => dir_entry,
             Err(error) => {
                 tracing::error!("failed to read dir '{}': {error}", path.display());
@@ -62,8 +63,8 @@ pub fn collect_metadata(path: &Path, tests: &mut Vec<MetadataFile>) {
             }
         };
 
-        for entry in dir_entry {
-            let entry = match entry {
+        for path in dir_entry {
+            let path = match path {
                 Ok(entry) => entry,
                 Err(error) => {
                     tracing::error!("error reading dir entry: {error}");
@@ -71,7 +72,6 @@ pub fn collect_metadata(path: &Path, tests: &mut Vec<MetadataFile>) {
                 }
             };
 
-            let path = entry.path();
             if path.is_dir() {
                 collect_metadata(&path, tests);
                 continue;

crates/format/src/metadata.rs

@@ -2,7 +2,7 @@ use std::{
     cmp::Ordering,
     collections::BTreeMap,
     fmt::Display,
-    fs::{File, read_to_string},
+    fs::File,
     ops::Deref,
     path::{Path, PathBuf},
     str::FromStr,
@@ -11,7 +11,9 @@ use std::{
 use serde::{Deserialize, Serialize};
 
 use revive_common::EVMVersion;
-use revive_dt_common::{iterators::FilesWithExtensionIterator, macros::define_wrapper_type};
+use revive_dt_common::{
+    cached_fs::read_to_string, iterators::FilesWithExtensionIterator, macros::define_wrapper_type,
+};
 
 use crate::{
     case::Case,
@@ -259,7 +261,9 @@ impl Metadata {
             Ok(Box::new(std::iter::once(metadata_file_path.clone())))
         } else {
             Ok(Box::new(
-                FilesWithExtensionIterator::new(self.directory()?).with_allowed_extension("sol"),
+                FilesWithExtensionIterator::new(self.directory()?)
+                    .with_allowed_extension("sol")
+                    .with_use_cached_fs(true),
             ))
         }
     }

crates/node/src/pool.rs

@@ -1,11 +1,12 @@
 //! This crate implements concurrent handling of testing node.
 
 use std::{
-    fs::read_to_string,
     sync::atomic::{AtomicUsize, Ordering},
     thread,
 };
 
+use revive_dt_common::cached_fs::read_to_string;
+
 use anyhow::Context;
 use revive_dt_config::Arguments;