feat: add SourceParser (#300)

Allows persisting a shared
[`solar_sema::Compiler`](paradigmxyz/solar#397)
throughout compilation and storing it inside of the output.

---------

Co-authored-by: onbjerg <[email protected]>

Author: DaniPopes

Cargo.toml

@@ -54,8 +54,8 @@ semver = { version = "1.0", features = ["serde"] }
 serde = { version = "1", features = ["derive", "rc"] }
 serde_json = "1.0"
 similar-asserts = "1"
-solar-parse = { version = "=0.1.5", default-features = false }
-solar-sema = { version = "=0.1.5", default-features = false }
+solar-parse = { version = "=0.1.6", default-features = false }
+solar-sema = { version = "=0.1.6", default-features = false }
 svm = { package = "svm-rs", version = "0.5", default-features = false }
 tempfile = "3.20"
 thiserror = "2"
@@ -69,7 +69,7 @@ tokio = { version = "1.47", features = ["rt-multi-thread"] }
 
 snapbox = "0.6.21"
 
-# [patch.crates-io]
+[patch.crates-io]
 # solar-parse = { git = "https://github.com/paradigmxyz/solar", branch = "main" }
 # solar-sema = { git = "https://github.com/paradigmxyz/solar", branch = "main" }
 # solar-ast = { git = "https://github.com/paradigmxyz/solar", branch = "main" }

crates/artifacts/solc/src/sources.rs

@@ -1,4 +1,4 @@
-use foundry_compilers_core::error::SolcIoError;
+use foundry_compilers_core::error::{SolcError, SolcIoError};
 use serde::{Deserialize, Serialize};
 use std::{
     collections::BTreeMap,
@@ -137,6 +137,30 @@ impl Source {
         Ok(Self::new(content))
     }
 
+    /// [`read`](Self::read) + mapping error to [`SolcError`].
+    pub fn read_(file: &Path) -> Result<Self, SolcError> {
+        Self::read(file).map_err(|err| {
+            let exists = err.path().exists();
+            if !exists && err.path().is_symlink() {
+                return SolcError::ResolveBadSymlink(err);
+            }
+
+            // This is an additional check useful on OS that have case-sensitive paths,
+            // see also <https://docs.soliditylang.org/en/v0.8.17/path-resolution.html#import-callback>
+            // check if there exists a file with different case
+            #[cfg(feature = "walkdir")]
+            if !exists {
+                if let Some(existing_file) =
+                    foundry_compilers_core::utils::find_case_sensitive_existing_file(file)
+                {
+                    return SolcError::ResolveCaseSensitiveFileName { error: err, existing_file };
+                }
+            }
+
+            SolcError::Resolve(err)
+        })
+    }
+
     /// Returns `true` if the source should be compiled with full output selection.
     pub fn is_dirty(&self) -> bool {
         self.kind.is_dirty()

crates/compilers/src/cache.rs

@@ -6,7 +6,7 @@ use crate::{
     output::Builds,
     resolver::GraphEdges,
     ArtifactFile, ArtifactOutput, Artifacts, ArtifactsMap, Graph, OutputContext, Project,
-    ProjectPaths, ProjectPathsConfig, SourceCompilationKind,
+    ProjectPaths, ProjectPathsConfig, SourceCompilationKind, SourceParser,
 };
 use foundry_compilers_artifacts::{
     sources::{Source, Sources},
@@ -658,7 +658,7 @@ pub(crate) struct ArtifactsCacheInner<
     pub cached_builds: Builds<C::Language>,
 
     /// Relationship between all the files.
-    pub edges: GraphEdges<C::ParsedSource>,
+    pub edges: GraphEdges<C::Parser>,
 
     /// The project.
     pub project: &'a Project<C, T>,
@@ -723,6 +723,7 @@ impl<T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
     /// Gets or calculates the interface representation hash for the given source file.
     fn interface_repr_hash(&mut self, source: &Source, file: &Path) -> &str {
         self.interface_repr_hashes.entry(file.to_path_buf()).or_insert_with(|| {
+            // TODO: use `interface_representation_ast` directly with `edges.parser()`.
             if let Some(r) = interface_repr_hash(&source.content, file) {
                 return r;
             }
@@ -823,10 +824,10 @@ impl<T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
 
     // Walks over all cache entries, detects dirty files and removes them from cache.
     fn find_and_remove_dirty(&mut self) {
-        fn populate_dirty_files<D>(
+        fn populate_dirty_files<P: SourceParser>(
             file: &Path,
             dirty_files: &mut HashSet<PathBuf>,
-            edges: &GraphEdges<D>,
+            edges: &GraphEdges<P>,
         ) {
             for file in edges.importers(file) {
                 // If file is marked as dirty we either have already visited it or it was marked as
@@ -890,7 +891,7 @@ impl<T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
 
         // Build a temporary graph for walking imports. We need this because `self.edges`
         // only contains graph data for in-scope sources but we are operating on cache entries.
-        if let Ok(graph) = Graph::<C::ParsedSource>::resolve_sources(&self.project.paths, sources) {
+        if let Ok(graph) = Graph::<C::Parser>::resolve_sources(&self.project.paths, sources) {
             let (sources, edges) = graph.into_sources();
 
             // Calculate content hashes for later comparison.
@@ -1020,7 +1021,7 @@ pub(crate) enum ArtifactsCache<
     C: Compiler,
 > {
     /// Cache nothing on disk
-    Ephemeral(GraphEdges<C::ParsedSource>, &'a Project<C, T>),
+    Ephemeral(GraphEdges<C::Parser>, &'a Project<C, T>),
     /// Handles the actual cached artifacts, detects artifacts that can be reused
     Cached(ArtifactsCacheInner<'a, T, C>),
 }
@@ -1032,7 +1033,7 @@ impl<'a, T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
     #[instrument(name = "ArtifactsCache::new", skip(project, edges))]
     pub fn new(
         project: &'a Project<C, T>,
-        edges: GraphEdges<C::ParsedSource>,
+        edges: GraphEdges<C::Parser>,
         preprocessed: bool,
     ) -> Result<Self> {
         /// Returns the [CompilerCache] to use
@@ -1117,7 +1118,7 @@ impl<'a, T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
     }
 
     /// Returns the graph data for this project
-    pub fn graph(&self) -> &GraphEdges<C::ParsedSource> {
+    pub fn graph(&self) -> &GraphEdges<C::Parser> {
         match self {
             ArtifactsCache::Ephemeral(graph, _) => graph,
             ArtifactsCache::Cached(inner) => &inner.edges,
@@ -1191,18 +1192,22 @@ impl<'a, T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
     ///
     /// Returns all the _cached_ artifacts.
     #[instrument(name = "ArtifactsCache::consume", skip_all)]
+    #[allow(clippy::type_complexity)]
     pub fn consume<A>(
         self,
         written_artifacts: &Artifacts<A>,
         written_build_infos: &Vec<RawBuildInfo<C::Language>>,
         write_to_disk: bool,
-    ) -> Result<(Artifacts<A>, Builds<C::Language>)>
+    ) -> Result<(Artifacts<A>, Builds<C::Language>, GraphEdges<C::Parser>)>
     where
         T: ArtifactOutput<Artifact = A>,
     {
-        let ArtifactsCache::Cached(cache) = self else {
-            trace!("no cache configured, ephemeral");
-            return Ok(Default::default());
+        let cache = match self {
+            ArtifactsCache::Ephemeral(edges, _project) => {
+                trace!("no cache configured, ephemeral");
+                return Ok((Default::default(), Default::default(), edges));
+            }
+            ArtifactsCache::Cached(cache) => cache,
         };
 
         let ArtifactsCacheInner {
@@ -1212,7 +1217,9 @@ impl<'a, T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
             dirty_sources,
             sources_in_scope,
             project,
-            ..
+            edges,
+            content_hashes: _,
+            interface_repr_hashes: _,
         } = cache;
 
         // Remove cached artifacts which are out of scope, dirty or appear in `written_artifacts`.
@@ -1264,7 +1271,7 @@ impl<'a, T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
             cache.write(project.cache_path())?;
         }
 
-        Ok((cached_artifacts, cached_builds))
+        Ok((cached_artifacts, cached_builds, edges))
     }
 
     /// Marks the cached entry as seen by the compiler, if it's cached.

crates/compilers/src/cache/iface.rs

@@ -1,5 +1,5 @@
 use crate::{parse_one_source, replace_source_content};
-use solar_sema::{
+use solar_parse::{
     ast::{self, Span},
     interface::diagnostics::EmittedDiagnostics,
 };
@@ -11,7 +11,7 @@ pub(crate) fn interface_repr_hash(content: &str, path: &Path) -> Option<String>
 }
 
 pub(crate) fn interface_repr(content: &str, path: &Path) -> Result<String, EmittedDiagnostics> {
-    parse_one_source(content, path, |ast| interface_representation_ast(content, &ast))
+    parse_one_source(content, path, |sess, ast| interface_representation_ast(content, sess, ast))
 }
 
 /// Helper function to remove parts of the contract which do not alter its interface:
@@ -21,6 +21,7 @@ pub(crate) fn interface_repr(content: &str, path: &Path) -> Result<String, Emitt
 /// Preserves all libraries and interfaces.
 pub(crate) fn interface_representation_ast(
     content: &str,
+    sess: &solar_sema::interface::Session,
     ast: &solar_parse::ast::SourceUnit<'_>,
 ) -> String {
     let mut spans_to_remove: Vec<Span> = Vec::new();
@@ -57,9 +58,9 @@ pub(crate) fn interface_representation_ast(
             }
         }
     }
-    let content =
-        replace_source_content(content, spans_to_remove.iter().map(|span| (span.to_range(), "")))
-            .replace("\n", "");
+    let updates =
+        spans_to_remove.iter().map(|&span| (sess.source_map().span_to_source(span).unwrap().1, ""));
+    let content = replace_source_content(content, updates).replace("\n", "");
     crate::utils::RE_TWO_OR_MORE_SPACES.replace_all(&content, "").into_owned()
 }
 

crates/compilers/src/compile/output/mod.rs

@@ -19,6 +19,7 @@ use crate::{
     compilers::{
         multi::MultiCompiler, CompilationError, Compiler, CompilerContract, CompilerOutput,
     },
+    resolver::GraphEdges,
     Artifact, ArtifactId, ArtifactOutput, Artifacts, ConfigurableArtifacts,
 };
 
@@ -62,7 +63,7 @@ impl<L> IntoIterator for Builds<L> {
 
 /// Contains a mixture of already compiled/cached artifacts and the input set of sources that still
 /// need to be compiled.
-#[derive(Clone, Debug, Default, PartialEq, Eq)]
+#[derive(Clone, Debug, Default)]
 pub struct ProjectCompileOutput<
     C: Compiler = MultiCompiler,
     T: ArtifactOutput<CompilerContract = C::CompilerContract> = ConfigurableArtifacts,
@@ -81,11 +82,23 @@ pub struct ProjectCompileOutput<
     pub(crate) compiler_severity_filter: Severity,
     /// all build infos that were just compiled
     pub(crate) builds: Builds<C::Language>,
+    /// The relationship between the source files and their imports
+    pub(crate) edges: GraphEdges<C::Parser>,
 }
 
 impl<T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
     ProjectCompileOutput<C, T>
 {
+    /// Returns the parser used to parse the sources.
+    pub fn parser(&self) -> &C::Parser {
+        self.edges.parser()
+    }
+
+    /// Returns the parser used to parse the sources.
+    pub fn parser_mut(&mut self) -> &mut C::Parser {
+        self.edges.parser_mut()
+    }
+
     /// Converts all `\\` separators in _all_ paths to `/`
     #[instrument(skip_all)]
     pub fn slash_paths(&mut self) {
@@ -460,6 +473,11 @@ impl<T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
     pub fn builds(&self) -> impl Iterator<Item = (&String, &BuildContext<C::Language>)> {
         self.builds.iter()
     }
+
+    /// Returns the source graph of the project.
+    pub fn graph(&self) -> &GraphEdges<C::Parser> {
+        &self.edges
+    }
 }
 
 impl<C: Compiler, T: ArtifactOutput<CompilerContract = C::CompilerContract>>

crates/compilers/src/compile/project.rs

@@ -146,7 +146,7 @@ pub struct ProjectCompiler<
     C: Compiler,
 > {
     /// Contains the relationship of the source files and their imports
-    edges: GraphEdges<C::ParsedSource>,
+    edges: GraphEdges<C::Parser>,
     project: &'a Project<C, T>,
     /// A mapping from a source file path to the primary profile name selected for it.
     primary_profiles: HashMap<PathBuf, &'a str>,
@@ -381,7 +381,7 @@ impl<T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
         let skip_write_to_disk = project.no_artifacts || has_error;
         trace!(has_error, project.no_artifacts, skip_write_to_disk, cache_path=?project.cache_path(),"prepare writing cache file");
 
-        let (cached_artifacts, cached_builds) =
+        let (cached_artifacts, cached_builds, edges) =
             cache.consume(&compiled_artifacts, &output.build_infos, !skip_write_to_disk)?;
 
         project.artifacts_handler().handle_cached_artifacts(&cached_artifacts)?;
@@ -404,6 +404,7 @@ impl<T: ArtifactOutput<CompilerContract = C::CompilerContract>, C: Compiler>
             ignored_file_paths,
             compiler_severity_filter,
             builds,
+            edges,
         })
     }
 }

crates/compilers/src/compilers/mod.rs

@@ -1,4 +1,4 @@
-use crate::ProjectPathsConfig;
+use crate::{resolver::Node, ProjectPathsConfig};
 use alloy_json_abi::JsonAbi;
 use core::fmt;
 use foundry_compilers_artifacts::{
@@ -9,6 +9,7 @@ use foundry_compilers_artifacts::{
     BytecodeObject, CompactContractRef, Contract, FileToContractsMap, Severity, SourceFile,
 };
 use foundry_compilers_core::error::Result;
+use rayon::prelude::*;
 use semver::{Version, VersionReq};
 use serde::{de::DeserializeOwned, Deserialize, Serialize};
 use std::{
@@ -139,12 +140,40 @@ pub trait CompilerInput: Serialize + Send + Sync + Sized + Debug {
     fn strip_prefix(&mut self, base: &Path);
 }
 
+/// [`ParsedSource`] parser.
+pub trait SourceParser: Clone + Debug + Send + Sync {
+    type ParsedSource: ParsedSource;
+
+    /// Creates a new parser for the given config.
+    fn new(config: &ProjectPathsConfig) -> Self;
+
+    /// Reads and parses the source file at the given path.
+    fn read(&mut self, path: &Path) -> Result<Node<Self::ParsedSource>> {
+        Node::read(path)
+    }
+
+    /// Parses the sources in the given sources map.
+    fn parse_sources(
+        &mut self,
+        sources: &mut Sources,
+    ) -> Result<Vec<(PathBuf, Node<Self::ParsedSource>)>> {
+        sources
+            .0
+            .par_iter()
+            .map(|(path, source)| {
+                let data = Self::ParsedSource::parse(source.as_ref(), path)?;
+                Ok((path.clone(), Node::new(path.clone(), source.clone(), data)))
+            })
+            .collect::<Result<_>>()
+    }
+}
+
 /// Parser of the source files which is used to identify imports and version requirements of the
 /// given source.
 ///
 /// Used by path resolver to resolve imports or determine compiler versions needed to compiler given
 /// sources.
-pub trait ParsedSource: Debug + Sized + Send + Clone {
+pub trait ParsedSource: Clone + Debug + Sized + Send {
     type Language: Language;
 
     /// Parses the content of the source file.
@@ -331,7 +360,7 @@ pub trait Compiler: Send + Sync + Clone {
     /// Output data for each contract
     type CompilerContract: CompilerContract;
     /// Source parser used for resolving imports and version requirements.
-    type ParsedSource: ParsedSource<Language = Self::Language>;
+    type Parser: SourceParser<ParsedSource: ParsedSource<Language = Self::Language>>;
     /// Compiler settings.
     type Settings: CompilerSettings;
     /// Enum of languages supported by the compiler.