feat(solc): emit build info files if configured (#1338)

* feat(solc): emit build info files if configured

* feat: add content hashing

ethers-solc/src/artifacts/serde_helpers.rs

@@ -78,7 +78,7 @@ pub mod json_string_opt {
 pub mod empty_json_object_opt {
     use serde::{
         de::{self, DeserializeOwned},
-        ser, Deserialize, Deserializer, Serialize, Serializer,
+        Deserialize, Deserializer, Serialize, Serializer,
     };
 
     pub fn serialize<T, S>(value: &Option<T>, serializer: S) -> Result<S::Ok, S::Error>
@@ -87,8 +87,7 @@ pub mod empty_json_object_opt {
         T: Serialize,
     {
         if let Some(value) = value {
-            let value = serde_json::to_string(value).map_err(ser::Error::custom)?;
+            value.serialize(serializer)
-            serializer.serialize_str(&value)
         } else {
             let empty = serde_json::Value::Object(Default::default());
             serde_json::Value::serialize(&empty, serializer)

ethers-solc/src/buildinfo.rs

@@ -0,0 +1,114 @@
+//! Represents an entire build
+
+use crate::{utils, CompilerInput, CompilerOutput, SolcError};
+use md5::Digest;
+use semver::Version;
+use serde::{ser::SerializeStruct, Deserialize, Serialize, Serializer};
+use std::{cell::RefCell, path::Path, rc::Rc};
+
+pub const ETHERS_FORMAT_VERSION: &str = "ethers-rs-sol-build-info-1";
+
+// A hardhat compatible build info representation
+#[derive(Serialize, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct BuildInfo {
+    pub id: String,
+    #[serde(rename = "_format")]
+    pub format: String,
+    pub solc_version: Version,
+    pub solc_long_version: Version,
+    pub input: CompilerInput,
+    pub output: CompilerOutput,
+}
+
+impl BuildInfo {
+    /// Deserializes the `BuildInfo` object from the given file
+    pub fn read(path: impl AsRef<Path>) -> Result<Self, SolcError> {
+        utils::read_json_file(path)
+    }
+}
+
+/// Represents `BuildInfo` object
+#[derive(Debug, Clone, Eq, PartialEq)]
+pub struct RawBuildInfo {
+    /// The hash that identifies the BuildInfo
+    pub id: String,
+    /// serialized `BuildInfo` json
+    pub build_info: String,
+}
+
+// === impl RawBuildInfo ===
+
+impl RawBuildInfo {
+    /// Serializes a `BuildInfo` object
+    pub fn new(
+        input: &CompilerInput,
+        output: &CompilerOutput,
+        version: &Version,
+    ) -> serde_json::Result<RawBuildInfo> {
+        let mut hasher = md5::Md5::new();
+        let w = BuildInfoWriter { buf: Rc::new(RefCell::new(Vec::with_capacity(128))) };
+        let mut buf = w.clone();
+        let mut serializer = serde_json::Serializer::pretty(&mut buf);
+        let mut s = serializer.serialize_struct("BuildInfo", 6)?;
+        s.serialize_field("_format", &ETHERS_FORMAT_VERSION)?;
+        let solc_short = format!("{}.{}.{}", version.major, version.minor, version.patch);
+        s.serialize_field("solcVersion", &solc_short)?;
+        s.serialize_field("solcLongVersion", &version)?;
+        s.serialize_field("input", input)?;
+
+        // create the hash for `{_format,solcVersion,solcLongVersion,input}`
+        // N.B. this is not exactly the same as hashing the json representation of these values but
+        // the must efficient one
+        hasher.update(&*w.buf.borrow());
+        let result = hasher.finalize();
+        let id = hex::encode(result);
+
+        s.serialize_field("id", &id)?;
+        s.serialize_field("output", output)?;
+        s.end()?;
+
+        drop(buf);
+
+        let build_info = unsafe {
+            // serde_json does not emit non UTF8
+            String::from_utf8_unchecked(w.buf.take())
+        };
+
+        Ok(RawBuildInfo { id, build_info })
+    }
+}
+
+#[derive(Clone)]
+struct BuildInfoWriter {
+    buf: Rc<RefCell<Vec<u8>>>,
+}
+
+impl std::io::Write for BuildInfoWriter {
+    fn write(&mut self, buf: &[u8]) -> std::io::Result<usize> {
+        self.buf.borrow_mut().write(buf)
+    }
+
+    fn flush(&mut self) -> std::io::Result<()> {
+        self.buf.borrow_mut().flush()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::Source;
+    use std::{collections::BTreeMap, path::PathBuf};
+
+    #[test]
+    fn build_info_serde() {
+        let inputs = CompilerInput::with_sources(BTreeMap::from([(
+            PathBuf::from("input.sol"),
+            Source { content: "".to_string() },
+        )]));
+        let output = CompilerOutput::default();
+        let v: Version = "0.8.4+commit.c7e474f2".parse().unwrap();
+        let raw_info = RawBuildInfo::new(&inputs[0], &output, &v).unwrap();
+        let _info: BuildInfo = serde_json::from_str(&raw_info.build_info).unwrap();
+    }
+}

ethers-solc/src/compile/output/mod.rs

@@ -5,12 +5,14 @@ use crate::{
         contract::{CompactContractBytecode, CompactContractRef, Contract},
         Error,
     },
+    buildinfo::RawBuildInfo,
     sources::{VersionedSourceFile, VersionedSourceFiles},
-    ArtifactId, ArtifactOutput, Artifacts, CompilerOutput, ConfigurableArtifacts,
+    ArtifactId, ArtifactOutput, Artifacts, CompilerOutput, ConfigurableArtifacts, SolcIoError,
 };
 use contracts::{VersionedContract, VersionedContracts};
 use semver::Version;
 use std::{collections::BTreeMap, fmt, path::Path};
+use tracing::trace;
 
 pub mod contracts;
 pub mod sources;
@@ -243,6 +245,8 @@ pub struct AggregatedCompilerOutput {
     pub sources: VersionedSourceFiles,
     /// All compiled contracts combined with the solc version used to compile them
     pub contracts: VersionedContracts,
+    // All the `BuildInfo`s of solc invocations.
+    pub build_infos: BTreeMap<Version, RawBuildInfo>,
 }
 
 impl AggregatedCompilerOutput {
@@ -302,6 +306,29 @@ impl AggregatedCompilerOutput {
         }
     }
 
+    /// Creates all `BuildInfo` files in the given `build_info_dir`
+    ///
+    /// There can be multiple `BuildInfo`, since we support multiple versions.
+    ///
+    /// The created files have the md5 hash `{_format,solcVersion,solcLongVersion,input}` as their
+    /// file name
+    pub fn write_build_infos(&self, build_info_dir: impl AsRef<Path>) -> Result<(), SolcIoError> {
+        if self.build_infos.is_empty() {
+            return Ok(())
+        }
+        let build_info_dir = build_info_dir.as_ref();
+        std::fs::create_dir_all(build_info_dir)
+            .map_err(|err| SolcIoError::new(err, build_info_dir))?;
+        for (version, build_info) in &self.build_infos {
+            trace!("writing build info file for solc {}", version);
+            let file_name = format!("{}.json", build_info.id);
+            let file = build_info_dir.join(file_name);
+            std::fs::write(&file, &build_info.build_info)
+                .map_err(|err| SolcIoError::new(err, file))?;
+        }
+        Ok(())
+    }
+
     /// Finds the _first_ contract with the given name
     ///
     /// # Example

ethers-solc/src/compile/project.rs

@@ -104,6 +104,7 @@
 use crate::{
     artifact_output::Artifacts,
     artifacts::{Settings, VersionedFilteredSources, VersionedSources},
+    buildinfo::RawBuildInfo,
     cache::ArtifactsCache,
     error::Result,
     filter::SparseOutputFilter,
@@ -240,11 +241,13 @@ impl<'a, T: ArtifactOutput> PreprocessedState<'a, T> {
     /// advance to the next state by compiling all sources
     fn compile(self) -> Result<CompiledState<'a, T>> {
         let PreprocessedState { sources, cache, sparse_output } = self;
+        let project = cache.project();
         let mut output = sources.compile(
-            &cache.project().solc_config.settings,
+            &project.solc_config.settings,
-            &cache.project().paths,
+            &project.paths,
             sparse_output,
             cache.graph(),
+            project.build_info,
         )?;
 
         // source paths get stripped before handing them over to solc, so solc never uses absolute
@@ -288,11 +291,16 @@ impl<'a, T: ArtifactOutput> CompiledState<'a, T> {
                 output.sources.len()
             );
             // this emits the artifacts via the project's artifacts handler
-            project.artifacts_handler().on_output(
+            let artifacts = project.artifacts_handler().on_output(
                 &output.contracts,
                 &output.sources,
                 &project.paths,
-            )?
+            )?;
+
+            // emits all the build infos, if they exist
+            output.write_build_infos(project.build_info_path())?;
+
+            artifacts
         };
 
         Ok(ArtifactsState { output, cache, compiled_artifacts })
@@ -391,13 +399,14 @@ impl FilteredCompilerSources {
         paths: &ProjectPathsConfig,
         sparse_output: SparseOutputFilter,
         graph: &GraphEdges,
+        create_build_info: bool,
     ) -> Result<AggregatedCompilerOutput> {
         match self {
             FilteredCompilerSources::Sequential(input) => {
-                compile_sequential(input, settings, paths, sparse_output, graph)
+                compile_sequential(input, settings, paths, sparse_output, graph, create_build_info)
             }
             FilteredCompilerSources::Parallel(input, j) => {
-                compile_parallel(input, j, settings, paths, sparse_output, graph)
+                compile_parallel(input, j, settings, paths, sparse_output, graph, create_build_info)
             }
         }
     }
@@ -419,6 +428,7 @@ fn compile_sequential(
     paths: &ProjectPathsConfig,
     sparse_output: SparseOutputFilter,
     graph: &GraphEdges,
+    create_build_info: bool,
 ) -> Result<AggregatedCompilerOutput> {
     let mut aggregated = AggregatedCompilerOutput::default();
     tracing::trace!("compiling {} jobs sequentially", input.len());
@@ -484,6 +494,13 @@ fn compile_sequential(
             report::solc_success(&solc, &version, &output, &start.elapsed());
             tracing::trace!("compiled input, output has error: {}", output.has_error());
             tracing::trace!("received compiler output: {:?}", output.contracts.keys());
+
+            // if configured also create the build info
+            if create_build_info {
+                let build_info = RawBuildInfo::new(&input, &output, &version)?;
+                aggregated.build_infos.insert(version.clone(), build_info);
+            }
+
             aggregated.extend(version.clone(), output);
         }
     }
@@ -498,6 +515,7 @@ fn compile_parallel(
     paths: &ProjectPathsConfig,
     sparse_output: SparseOutputFilter,
     graph: &GraphEdges,
+    create_build_info: bool,
 ) -> Result<AggregatedCompilerOutput> {
     debug_assert!(num_jobs > 1);
     tracing::trace!(
@@ -580,14 +598,21 @@ fn compile_parallel(
                 report::solc_spawn(&solc, &version, &input, &actually_dirty);
                 solc.compile(&input).map(move |output| {
                     report::solc_success(&solc, &version, &output, &start.elapsed());
-                    (version, output)
+                    (version, input, output)
                 })
             })
             .collect::<Result<Vec<_>>>()
     })?;
 
     let mut aggregated = AggregatedCompilerOutput::default();
-    aggregated.extend_all(outputs);
+    for (version, input, output) in outputs {
+        // if configured also create the build info
+        if create_build_info {
+            let build_info = RawBuildInfo::new(&input, &output, &version)?;
+            aggregated.build_infos.insert(version.clone(), build_info);
+        }
+        aggregated.extend(version, output);
+    }
 
     Ok(aggregated)
 }

ethers-solc/src/lib.rs

@@ -6,6 +6,7 @@ pub use artifacts::{CompilerInput, CompilerOutput, EvmVersion};
 use std::collections::{BTreeMap, HashSet};
 
 mod artifact_output;
+pub mod buildinfo;
 pub mod cache;
 pub mod hh;
 pub use artifact_output::*;
@@ -59,6 +60,8 @@ pub struct Project<T: ArtifactOutput = ConfigurableArtifacts> {
     pub solc_config: SolcConfig,
     /// Whether caching is enabled
     pub cached: bool,
+    /// Whether to output build information with each solc call.
+    pub build_info: bool,
     /// Whether writing artifacts to disk is enabled
     pub no_artifacts: bool,
     /// Whether writing artifacts to disk is enabled
@@ -121,6 +124,11 @@ impl<T: ArtifactOutput> Project<T> {
         &self.paths.cache
     }
 
+    /// Returns the path to the `build-info` directory nested in the artifacts dir
+    pub fn build_info_path(&self) -> PathBuf {
+        self.paths.artifacts.join("build-info")
+    }
+
     /// Returns the root directory of the project
     pub fn root(&self) -> &PathBuf {
         &self.paths.root
@@ -513,6 +521,8 @@ pub struct ProjectBuilder<T: ArtifactOutput = ConfigurableArtifacts> {
     solc_config: Option<SolcConfig>,
     /// Whether caching is enabled, default is true.
     cached: bool,
+    /// Whether to output build information with each solc call.
+    build_info: bool,
     /// Whether writing artifacts to disk is enabled, default is true.
     no_artifacts: bool,
     /// Whether automatic solc version detection is enabled
@@ -536,6 +546,7 @@ impl<T: ArtifactOutput> ProjectBuilder<T> {
             solc: None,
             solc_config: None,
             cached: true,
+            build_info: false,
             no_artifacts: false,
             auto_detect: true,
             offline: false,
@@ -591,6 +602,13 @@ impl<T: ArtifactOutput> ProjectBuilder<T> {
         self
     }
 
+    /// Sets the build info value
+    #[must_use]
+    pub fn set_build_info(mut self, build_info: bool) -> Self {
+        self.build_info = build_info;
+        self
+    }
+
     /// Activates offline mode
     ///
     /// Prevents network possible access to download/check solc installs
@@ -663,6 +681,7 @@ impl<T: ArtifactOutput> ProjectBuilder<T> {
             allowed_paths,
             solc_jobs,
             offline,
+            build_info,
             ..
         } = self;
         ProjectBuilder {
@@ -677,6 +696,7 @@ impl<T: ArtifactOutput> ProjectBuilder<T> {
             ignored_error_codes,
             allowed_paths,
             solc_jobs,
+            build_info,
         }
     }
 
@@ -713,6 +733,7 @@ impl<T: ArtifactOutput> ProjectBuilder<T> {
             mut allowed_paths,
             solc_jobs,
             offline,
+            build_info,
         } = self;
 
         let paths = paths.map(Ok).unwrap_or_else(ProjectPathsConfig::current_hardhat)?;
@@ -730,6 +751,7 @@ impl<T: ArtifactOutput> ProjectBuilder<T> {
             solc,
             solc_config,
             cached,
+            build_info,
             no_artifacts,
             auto_detect,
             artifacts,

ethers-solc/tests/project.rs

@@ -13,6 +13,7 @@ use ethers_solc::{
         BytecodeHash, DevDoc, ErrorDoc, EventDoc, Libraries, MethodDoc, ModelCheckerEngine::CHC,
         ModelCheckerSettings, UserDoc, UserDocNotice,
     },
+    buildinfo::BuildInfo,
     cache::{SolFilesCache, SOLIDITY_FILES_CACHE_FILENAME},
     project_util::*,
     remappings::Remapping,
@@ -311,6 +312,45 @@ fn can_compile_dapp_detect_changes_in_sources() {
     }
 }
 
+#[test]
+fn can_emit_build_info() {
+    let mut project = TempProject::dapptools().unwrap();
+    project.project_mut().build_info = true;
+    project
+        .add_source(
+            "A",
+            r#"
+pragma solidity ^0.8.10;
+import "./B.sol";
+contract A { }
+"#,
+        )
+        .unwrap();
+
+    project
+        .add_source(
+            "B",
+            r#"
+pragma solidity ^0.8.10;
+contract B { }
+"#,
+        )
+        .unwrap();
+
+    let compiled = project.compile().unwrap();
+    assert!(!compiled.has_compiler_errors());
+
+    let info_dir = project.project().build_info_path();
+    assert!(info_dir.exists());
+
+    let mut build_info_count = 0;
+    for entry in fs::read_dir(info_dir).unwrap() {
+        let _info = BuildInfo::read(entry.unwrap().path()).unwrap();
+        build_info_count += 1;
+    }
+    assert_eq!(build_info_count, 1);
+}
+
 #[test]
 fn can_compile_dapp_sample_with_cache() {
     let tmp_dir = tempfile::tempdir().unwrap();