fix: pass build info as buffer

.changeset/honest-crews-drum.md

@@ -0,0 +1,5 @@
+---
+"@nomicfoundation/edr": minor
+---
+
+Improved provider initialization performance by passing build info as buffer which avoids FFI copy overhead

crates/edr_napi/index.d.ts

@@ -254,6 +254,27 @@ export interface ProviderConfig {
   /** The network ID of the blockchain */
   networkId: bigint
 }
+/** Tracing config for Solidity stack trace generation. */
+export interface TracingConfigWithBuffers {
+  /**
+   * Build information to use for decoding contracts. Either a Hardhat v2
+   * build info file that contains both input and output or a Hardhat v3
+   * build info file that doesn't contain output and a separate output file.
+   */
+  buildInfos?: Array<Uint8Array> | Array<BuildInfoAndOutput>
+  /** Whether to ignore contracts whose name starts with "Ignored". */
+  ignoreContracts?: boolean
+}
+/**
+ * Hardhat V3 build info where the compiler output is not part of the build
+ * info file.
+ */
+export interface BuildInfoAndOutput {
+  /** The build info input file */
+  buildInfo: Uint8Array
+  /** The build info output file */
+  output: Uint8Array
+}
 /** The possible reasons for successful termination of the EVM. */
 export const enum SuccessReason {
   /** The opcode `STOP` was called */
@@ -595,7 +616,7 @@ export declare class EdrContext {
 /** A JSON-RPC provider for Ethereum. */
 export declare class Provider {
   /**Constructs a new provider with the provided configuration. */
-  static withConfig(context: EdrContext, config: ProviderConfig, loggerConfig: LoggerConfig, tracingConfig: any, subscriberCallback: (event: SubscriptionEvent) => void): Promise<Provider>
+  static withConfig(context: EdrContext, config: ProviderConfig, loggerConfig: LoggerConfig, tracingConfig: TracingConfigWithBuffers, subscriberCallback: (event: SubscriptionEvent) => void): Promise<Provider>
   /**Handles a JSON-RPC request and returns a JSON-RPC response. */
   handleRequest(jsonRequest: string): Promise<Response>
   setCallOverrideCallback(callOverrideCallback: (contract_address: Buffer, data: Buffer) => Promise<CallOverrideResult | undefined>): void

crates/edr_napi/src/provider.rs

@@ -5,7 +5,9 @@ use std::sync::Arc;
 use edr_provider::{time::CurrentTime, InvalidRequestReason};
 use edr_rpc_eth::jsonrpc;
 use edr_solidity::contract_decoder::ContractDecoder;
-use napi::{tokio::runtime, Either, Env, JsFunction, JsObject, Status};
+use napi::{
+    bindgen_prelude::Uint8Array, tokio::runtime, Either, Env, JsFunction, JsObject, Status,
+};
 use napi_derive::napi;
 
 use self::config::ProviderConfig;
@@ -37,16 +39,17 @@ impl Provider {
         _context: &EdrContext,
         config: ProviderConfig,
         logger_config: LoggerConfig,
-        tracing_config: serde_json::Value,
+        tracing_config: TracingConfigWithBuffers,
         #[napi(ts_arg_type = "(event: SubscriptionEvent) => void")] subscriber_callback: JsFunction,
     ) -> napi::Result<JsObject> {
         let runtime = runtime::Handle::current();
 
         let config = edr_provider::ProviderConfig::try_from(config)?;
 
         // TODO https://github.com/NomicFoundation/edr/issues/760
-        let build_info_config: edr_solidity::contract_decoder::BuildInfoConfig =
-            serde_json::from_value(tracing_config)?;
+        let build_info_config =
+            edr_solidity::artifacts::BuildInfoConfig::parse_from_buffers((&tracing_config).into())
+                .map_err(|err| napi::Error::from_reason(err.to_string()))?;
         let contract_decoder = ContractDecoder::new(&build_info_config)
             .map_err(|error| napi::Error::from_reason(error.to_string()))?;
         let contract_decoder = Arc::new(contract_decoder);
@@ -245,6 +248,66 @@ impl Provider {
     }
 }
 
+/// Tracing config for Solidity stack trace generation.
+#[napi(object)]
+pub struct TracingConfigWithBuffers {
+    /// Build information to use for decoding contracts. Either a Hardhat v2
+    /// build info file that contains both input and output or a Hardhat v3
+    /// build info file that doesn't contain output and a separate output file.
+    pub build_infos: Option<Either<Vec<Uint8Array>, Vec<BuildInfoAndOutput>>>,
+    /// Whether to ignore contracts whose name starts with "Ignored".
+    pub ignore_contracts: Option<bool>,
+}
+
+/// Hardhat V3 build info where the compiler output is not part of the build
+/// info file.
+#[napi(object)]
+pub struct BuildInfoAndOutput {
+    /// The build info input file
+    pub build_info: Uint8Array,
+    /// The build info output file
+    pub output: Uint8Array,
+}
+
+impl<'a> From<&'a BuildInfoAndOutput>
+    for edr_solidity::artifacts::BuildInfoBufferSeparateOutput<'a>
+{
+    fn from(value: &'a BuildInfoAndOutput) -> Self {
+        Self {
+            build_info: value.build_info.as_ref(),
+            output: value.output.as_ref(),
+        }
+    }
+}
+
+impl<'a> From<&'a TracingConfigWithBuffers>
+    for edr_solidity::artifacts::BuildInfoConfigWithBuffers<'a>
+{
+    fn from(value: &'a TracingConfigWithBuffers) -> Self {
+        use edr_solidity::artifacts::{BuildInfoBufferSeparateOutput, BuildInfoBuffers};
+
+        let build_infos = value.build_infos.as_ref().map(|infos| match infos {
+            Either::A(with_output) => BuildInfoBuffers::WithOutput(
+                with_output
+                    .iter()
+                    .map(std::convert::AsRef::as_ref)
+                    .collect(),
+            ),
+            Either::B(separate_output) => BuildInfoBuffers::SeparateInputOutput(
+                separate_output
+                    .iter()
+                    .map(BuildInfoBufferSeparateOutput::from)
+                    .collect(),
+            ),
+        });
+
+        Self {
+            build_infos,
+            ignore_contracts: value.ignore_contracts,
+        }
+    }
+}
+
 #[derive(Debug)]
 struct SolidityTraceData {
     trace: Arc<edr_evm::trace::Trace>,

crates/edr_solidity/Cargo.toml

@@ -18,6 +18,7 @@ serde_json = { version = "1.0.89", features = ["preserve_order"] }
 strum = { version = "0.26.0", features = ["derive"] }
 semver = "1.0.23"
 thiserror = "1.0.58"
+itertools = "0.10.5"
 
 [dev-dependencies]
 criterion = "0.5"

crates/edr_solidity/benches/contracts_identifier.rs

@@ -13,8 +13,8 @@ use std::{fs, path::PathBuf, time::Duration};
 
 use criterion::{black_box, criterion_group, criterion_main, Criterion};
 use edr_solidity::{
-    artifacts::BuildInfo,
-    contract_decoder::{BuildInfoConfig, ContractDecoder},
+    artifacts::{BuildInfoConfig, BuildInfoWithOutput},
+    contract_decoder::ContractDecoder,
 };
 
 const FORGE_STD_ARTIFACTS_DIR: &str = "EDR_FORGE_STD_ARTIFACTS_DIR";
@@ -33,13 +33,13 @@ fn load_build_info_config() -> anyhow::Result<Option<BuildInfoConfig>> {
 
         if path.is_file() && path.extension().and_then(|ext| ext.to_str()) == Some("json") {
             let contents = fs::read(&path)?;
-            let build_info = serde_json::from_slice::<BuildInfo>(&contents)?;
+            let build_info = serde_json::from_slice::<BuildInfoWithOutput>(&contents)?;
             build_infos.push(build_info);
         }
     }
 
     Ok(Some(BuildInfoConfig {
-        build_infos: Some(build_infos),
+        build_infos,
         ignore_contracts: None,
     }))
 }
@@ -51,8 +51,6 @@ pub fn criterion_benchmark(c: &mut Criterion) {
 
     let contracts = &build_info_config
         .build_infos
-        .as_ref()
-        .expect("loaded build info")
         .first()
         .expect("there is at least one build info")
         .output

crates/edr_solidity/src/artifacts.rs

@@ -7,11 +7,148 @@
 use std::collections::HashMap;
 
 use indexmap::IndexMap;
+use itertools::Itertools;
 use serde::{Deserialize, Serialize};
 
-/// A `BuildInfo` is a file that contains all the information of a solc run. It
+/// Error in the build info config
+#[derive(Debug, thiserror::Error)]
+pub enum BuildInfoConfigError {
+    /// JSON deserialization error
+    #[error(transparent)]
+    Json(#[from] serde_json::Error),
+    /// Invalid semver in the build info
+    #[error(transparent)]
+    Semver(#[from] semver::Error),
+    /// Input output file mismatch
+    #[error("Input output mismatch. Input id: '{input_id}'. Output id: '{output_id}'")]
+    InputOutputMismatch { input_id: String, output_id: String },
+}
+
+/// Configuration for the [`crate::contract_decoder::ContractDecoder`].
+#[derive(Clone, Debug, Default, Deserialize, Serialize)]
+#[serde(rename_all = "camelCase")]
+pub struct BuildInfoConfig {
+    /// Build information to use for decoding contracts.
+    pub build_infos: Vec<BuildInfoWithOutput>,
+    /// Whether to ignore contracts whose name starts with "Ignored".
+    pub ignore_contracts: Option<bool>,
+}
+
+impl BuildInfoConfig {
+    /// Parse the config from bytes. This is a performance intensive operation
+    /// which is why it's not a `TryFrom` implementation.
+    pub fn parse_from_buffers(
+        config: BuildInfoConfigWithBuffers<'_>,
+    ) -> Result<Self, BuildInfoConfigError> {
+        let BuildInfoConfigWithBuffers {
+            build_infos,
+            ignore_contracts,
+        } = config;
+
+        let build_infos = build_infos.map_or_else(|| Ok(Vec::default()), |bi| bi.parse())?;
+
+        Ok(Self {
+            build_infos,
+            ignore_contracts,
+        })
+    }
+}
+
+/// Configuration for the [`crate::contract_decoder::ContractDecoder`] unparsed
+/// build infos.
+#[derive(Clone, Debug)]
+pub struct BuildInfoConfigWithBuffers<'a> {
+    /// Build information to use for decoding contracts.
+    pub build_infos: Option<BuildInfoBuffers<'a>>,
+    /// Whether to ignore contracts whose name starts with "Ignored".
+    pub ignore_contracts: Option<bool>,
+}
+
+/// Unparsed build infos.
+#[derive(Clone, Debug)]
+pub enum BuildInfoBuffers<'a> {
+    /// Deserializes to `BuildInfoWithOutput`.
+    WithOutput(Vec<&'a [u8]>),
+    /// Separate build info input and output files.
+    SeparateInputOutput(Vec<BuildInfoBufferSeparateOutput<'a>>),
+}
+
+impl BuildInfoBuffers<'_> {
+    fn parse(&self) -> Result<Vec<BuildInfoWithOutput>, BuildInfoConfigError> {
+        fn filter_on_solc_version(
+            build_info: BuildInfoWithOutput,
+        ) -> Result<Option<BuildInfoWithOutput>, BuildInfoConfigError> {
+            let solc_version = build_info.solc_version.parse::<semver::Version>()?;
+
+            if crate::compiler::FIRST_SOLC_VERSION_SUPPORTED <= solc_version {
+                Ok(Some(build_info))
+            } else {
+                Ok(None)
+            }
+        }
+
+        match self {
+            BuildInfoBuffers::WithOutput(build_infos_with_output) => build_infos_with_output
+                .iter()
+                .map(|item| {
+                    let build_info: BuildInfoWithOutput = serde_json::from_slice(item)?;
+                    filter_on_solc_version(build_info)
+                })
+                .flatten_ok()
+                .collect::<Result<Vec<BuildInfoWithOutput>, _>>(),
+            BuildInfoBuffers::SeparateInputOutput(separate_output) => separate_output
+                .iter()
+                .map(|item| {
+                    let input: BuildInfo = serde_json::from_slice(item.build_info)?;
+                    let output: BuildInfoOutput = serde_json::from_slice(item.output)?;
+                    // Make sure we get the output matching the input.
+                    if input.id != output.id {
+                        return Err(BuildInfoConfigError::InputOutputMismatch {
+                            input_id: input.id,
+                            output_id: output.id,
+                        });
+                    }
+                    filter_on_solc_version(BuildInfoWithOutput {
+                        _format: input._format,
+                        id: input.id,
+                        solc_version: input.solc_version,
+                        solc_long_version: input.solc_long_version,
+                        input: input.input,
+                        output: output.output,
+                    })
+                })
+                .flatten_ok()
+                .collect::<Result<Vec<BuildInfoWithOutput>, _>>(),
+        }
+    }
+}
+
+/// Separate build info input and output files.
+#[derive(Clone, Debug)]
+pub struct BuildInfoBufferSeparateOutput<'a> {
+    /// Deserializes to `BuildInfo`
+    pub build_info: &'a [u8],
+    /// Deserializes to `BuildInfoOutput`
+    pub output: &'a [u8],
+}
+
+/// A `BuildInfoWithOutput` contains all the information of a solc run. It
 /// includes all the necessary information to recreate that exact same run, and
-/// all of its output.
+/// the output of the run.
+#[derive(Clone, Debug, PartialEq, Eq, Deserialize, Serialize)]
+#[serde(rename_all = "camelCase")]
+pub struct BuildInfoWithOutput {
+    #[serde(rename = "_format")]
+    pub _format: String,
+    pub id: String,
+    pub solc_version: String,
+    pub solc_long_version: String,
+    pub input: CompilerInput,
+    pub output: CompilerOutput,
+}
+
+/// A `BuildInfo` contains all the input information of a solc run. It
+/// includes all the necessary information to recreate that exact same run.
 #[derive(Clone, Debug, PartialEq, Eq, Deserialize, Serialize)]
 #[serde(rename_all = "camelCase")]
 pub struct BuildInfo {
@@ -21,6 +158,15 @@ pub struct BuildInfo {
     pub solc_version: String,
     pub solc_long_version: String,
     pub input: CompilerInput,
+}
+
+/// A `BuildInfoOutput` contains all the output of a solc run.
+#[derive(Clone, Debug, PartialEq, Eq, Deserialize, Serialize)]
+#[serde(rename_all = "camelCase")]
+pub struct BuildInfoOutput {
+    #[serde(rename = "_format")]
+    pub _format: String,
+    pub id: String,
     pub output: CompilerOutput,
 }
 

crates/edr_solidity/src/compiler.rs

@@ -19,6 +19,9 @@ use crate::{
     source_map::decode_instructions,
 };
 
+/// First Solc version supported for stack trace generation
+pub const FIRST_SOLC_VERSION_SUPPORTED: semver::Version = semver::Version::new(0, 5, 1);
+
 /// For the Solidity compiler version and its standard JSON input and
 /// output[^1], creates the source model, decodes the bytecode with source
 /// mapping and links them to the source files.