Update the readme

Cargo.lock

@@ -5628,6 +5628,7 @@ dependencies = [
  "semver 1.0.26",
  "serde",
  "serde_json",
+ "serde_with",
  "strum",
  "temp-dir",
 ]
@@ -5668,6 +5669,7 @@ dependencies = [
  "alloy",
  "anyhow",
  "futures",
+ "itertools 0.14.0",
  "revive-common",
  "revive-dt-common",
  "schemars 1.0.4",

Cargo.toml

@@ -69,6 +69,7 @@ tracing-subscriber = { version = "0.3.19", default-features = false, features =
     "env-filter",
 ] }
 indexmap = { version = "2.10.0", default-features = false }
+itertools = { version = "0.14.0" }
 
 # revive compiler
 revive-solc-json-interface = { git = "https://github.com/paritytech/revive", rev = "3389865af7c3ff6f29a586d82157e8bc573c1a8e" }

README.md

@@ -52,195 +52,21 @@ All of the above need to be installed and available in the path in order for the
 
 This tool is being updated quite frequently. Therefore, it's recommended that you don't install the tool and then run it, but rather that you run it from the root of the directory using `cargo run --release`. The help command of the tool gives you all of the information you need to know about each of the options and flags that the tool offers.
 
-```bash
-$ cargo run --release -- execute-tests --help
-Error: Executes tests in the MatterLabs format differentially on multiple targets concurrently
-
-Usage: retester execute-tests [OPTIONS]
-
-Options:
-  -w, --working-directory <WORKING_DIRECTORY>
-          The working directory that the program will use for all of the temporary artifacts needed at runtime.
-
-          If not specified, then a temporary directory will be created and used by the program for all temporary artifacts.
-
-          [default: ]
-
-  -p, --platform <PLATFORMS>
-          The set of platforms that the differential tests should run on
-
-          [default: geth-evm-solc,revive-dev-node-polkavm-resolc]
-
-          Possible values:
-          - geth-evm-solc:                  The Go-ethereum reference full node EVM implementation with the solc compiler
-          - kitchensink-polkavm-resolc:     The kitchensink node with the PolkaVM backend with the resolc compiler
-          - kitchensink-revm-solc:          The kitchensink node with the REVM backend with the solc compiler
-          - revive-dev-node-polkavm-resolc: The revive dev node with the PolkaVM backend with the resolc compiler
-          - revive-dev-node-revm-solc:      The revive dev node with the REVM backend with the solc compiler
-
-  -c, --corpus <CORPUS>
-          A list of test corpus JSON files to be tested
-
-  -h, --help
-          Print help (see a summary with '-h')
-
-Solc Configuration:
-      --solc.version <VERSION>
-          Specifies the default version of the Solc compiler that should be used if there is no override specified by one of the test cases
-
-          [default: 0.8.29]
-
-Resolc Configuration:
-      --resolc.path <resolc.path>
-          Specifies the path of the resolc compiler to be used by the tool.
-
-          If this is not specified, then the tool assumes that it should use the resolc binary that's provided in the user's $PATH.
-
-          [default: resolc]
-
-Geth Configuration:
-      --geth.path <geth.path>
-          Specifies the path of the geth node to be used by the tool.
-
-          If this is not specified, then the tool assumes that it should use the geth binary that's provided in the user's $PATH.
-
-          [default: geth]
-
-      --geth.start-timeout-ms <geth.start-timeout-ms>
-          The amount of time to wait upon startup before considering that the node timed out
-
-          [default: 5000]
-
-Kitchensink Configuration:
-      --kitchensink.path <kitchensink.path>
-          Specifies the path of the kitchensink node to be used by the tool.
-
-          If this is not specified, then the tool assumes that it should use the kitchensink binary that's provided in the user's $PATH.
-
-          [default: substrate-node]
-
-      --kitchensink.start-timeout-ms <kitchensink.start-timeout-ms>
-          The amount of time to wait upon startup before considering that the node timed out
-
-          [default: 5000]
-
-      --kitchensink.dont-use-dev-node
-          This configures the tool to use Kitchensink instead of using the revive-dev-node
-
-Revive Dev Node Configuration:
-      --revive-dev-node.path <revive-dev-node.path>
-          Specifies the path of the revive dev node to be used by the tool.
-
-          If this is not specified, then the tool assumes that it should use the revive dev node binary that's provided in the user's $PATH.
-
-          [default: revive-dev-node]
-
-      --revive-dev-node.start-timeout-ms <revive-dev-node.start-timeout-ms>
-          The amount of time to wait upon startup before considering that the node timed out
-
-          [default: 5000]
-
-Eth RPC Configuration:
-      --eth-rpc.path <eth-rpc.path>
-          Specifies the path of the ETH RPC to be used by the tool.
-
-          If this is not specified, then the tool assumes that it should use the ETH RPC binary that's provided in the user's $PATH.
-
-          [default: eth-rpc]
-
-      --eth-rpc.start-timeout-ms <eth-rpc.start-timeout-ms>
-          The amount of time to wait upon startup before considering that the node timed out
-
-          [default: 5000]
-
-Genesis Configuration:
-      --genesis.path <genesis.path>
-          Specifies the path of the genesis file to use for the nodes that are started.
-
-          This is expected to be the path of a JSON geth genesis file.
-
-Wallet Configuration:
-      --wallet.default-private-key <DEFAULT_KEY>
-          The private key of the default signer
-
-          [default: 0x4f3edf983ac636a65a842ce7c78d9aa706d3b113bce9c46f30d7d21715b23b1d]
-
-      --wallet.additional-keys <ADDITIONAL_KEYS>
-          This argument controls which private keys the nodes should have access to and be added to its wallet signers. With a value of N, private keys (0, N] will be added to the signer set of the node
-
-          [default: 100000]
-
-Concurrency Configuration:
-      --concurrency.number-of-nodes <NUMBER_OF_NODES>
-          Determines the amount of nodes that will be spawned for each chain
-
-          [default: 5]
-
-      --concurrency.number-of-threads <NUMBER_OF_THREADS>
-          Determines the amount of tokio worker threads that will will be used
-
-          [default: 16]
-
-      --concurrency.number-of-concurrent-tasks <NUMBER_CONCURRENT_TASKS>
-          Determines the amount of concurrent tasks that will be spawned to run tests.
-
-          Defaults to 10 x the number of nodes.
-
-      --concurrency.ignore-concurrency-limit
-          Determines if the concurrency limit should be ignored or not
-
-Compilation Configuration:
-      --compilation.invalidate-cache
-          Controls if the compilation cache should be invalidated or not
-
-Report Configuration:
-      --report.include-compiler-input
-          Controls if the compiler input is included in the final report
-
-      --report.include-compiler-output
-          Controls if the compiler output is included in the final report
-```
-
-To run tests with this tool you need a corpus JSON file that defines the tests included in the corpus. The simplest corpus file looks like the following:
-
-```json
-{
-  "name": "MatterLabs Solidity Simple, Complex, and Semantic Tests",
-  "path": "resolc-compiler-tests/fixtures/solidity"
-}
-```
-
 > [!NOTE]  
 > Note that the tests can be found in the [`resolc-compiler-tests`](https://github.com/paritytech/resolc-compiler-tests) repository.
 
-The above corpus file instructs the tool to look for all of the test cases contained within all of the metadata files of the specified directory.
-
 The simplest command to run this tool is the following:
 
 ```bash
-RUST_LOG="info" cargo run --release -- execute-tests \
+RUST_LOG="info" cargo run --release -- test \
+    --test ./resolc-compiler-tests/fixtures/solidity \
     --platform geth-evm-solc \
-    --corpus corp.json \
     --working-directory workdir \
-    --concurrency.number-of-nodes 5 \
-    --concurrency.ignore-concurrency-limit \
     > logs.log \
     2> output.log
 ```
 
-The above command will run the tool executing every one of the tests discovered in the path specified in the corpus file. All of the logs from the execution will be persisted in the `logs.log` file and all of the output of the tool will be persisted to the `output.log` file. If all that you're looking for is to run the tool and check which tests succeeded and failed, then the `output.log` file is what you need to be looking at. However, if you're contributing the to the tool then the `logs.log` file will be very valuable.
-
-If you only want to run a subset of tests, then you can specify that in your corpus file. The following is an example:
-
-```json
-{
-  "name": "MatterLabs Solidity Simple, Complex, and Semantic Tests",
-  "paths": [
-    "path/to/a/single/metadata/file/I/want/to/run.json",
-    "path/to/a/directory/to/find/all/metadata/files/within"
-  ]
-}
-```
+The above command will run the tool executing every one of the tests discovered in the path provided to the tool. All of the logs from the execution will be persisted in the `logs.log` file and all of the output of the tool will be persisted to the `output.log` file. If all that you're looking for is to run the tool and check which tests succeeded and failed, then the `output.log` file is what you need to be looking at. However, if you're contributing the to the tool then the `logs.log` file will be very valuable.
 
 <details>
 <summary>User Managed Nodes</summary>
@@ -283,7 +109,7 @@ sleep 5
 # Run the tests (logs to files as before)
 RUST_LOG="info" retester test \
   --platform "$PLATFORM" \
-  --corpus ./corpus.json \
+  --corpus ./revive-differential-tests/fixtures/solidity \
   --working-directory ./workdir \
   --concurrency.number-of-nodes 1 \
   --concurrency.number-of-concurrent-tasks 5 \

crates/common/src/types/mod.rs

@@ -1,11 +1,13 @@
 mod identifiers;
 mod mode;
+mod parsed_test_specifier;
 mod private_key_allocator;
 mod round_robin_pool;
 mod version_or_requirement;
 
 pub use identifiers::*;
 pub use mode::*;
+pub use parsed_test_specifier::*;
 pub use private_key_allocator::*;
 pub use round_robin_pool::*;
 pub use version_or_requirement::*;

crates/common/src/types/parsed_test_specifier.rs

@@ -0,0 +1,133 @@
+use std::{fmt::Display, path::PathBuf, str::FromStr};
+
+use anyhow::{Context as _, bail};
+
+use crate::types::Mode;
+
+#[derive(Clone, Debug, PartialEq, Eq, Hash)]
+pub enum ParsedTestSpecifier {
+    /// All of the test cases in the file should be ran across all of the specified modes
+    FileOrDirectory {
+        /// The path of the metadata file containing the test cases.
+        metadata_or_directory_file_path: PathBuf,
+    },
+    /// Only a specific case within the metadata file should be ran across all of the modes in the
+    /// file.
+    Case {
+        /// The path of the metadata file containing the test cases.
+        metadata_file_path: PathBuf,
+
+        /// The index of the specific case to run.
+        case_idx: usize,
+    },
+    /// A specific case and a specific mode should be ran. This is the most specific out of all of
+    /// the specifier types.
+    CaseWithMode {
+        /// The path of the metadata file containing the test cases.
+        metadata_file_path: PathBuf,
+
+        /// The index of the specific case to run.
+        case_idx: usize,
+
+        /// The parsed mode that the test should be run in.
+        mode: Mode,
+    },
+}
+
+impl Display for ParsedTestSpecifier {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            ParsedTestSpecifier::FileOrDirectory {
+                metadata_or_directory_file_path,
+            } => {
+                write!(f, "{}", metadata_or_directory_file_path.display())
+            }
+            ParsedTestSpecifier::Case {
+                metadata_file_path,
+                case_idx,
+            } => {
+                write!(f, "{}::{}", metadata_file_path.display(), case_idx)
+            }
+            ParsedTestSpecifier::CaseWithMode {
+                metadata_file_path,
+                case_idx,
+                mode,
+            } => {
+                write!(
+                    f,
+                    "{}::{}::{}",
+                    metadata_file_path.display(),
+                    case_idx,
+                    mode
+                )
+            }
+        }
+    }
+}
+
+impl FromStr for ParsedTestSpecifier {
+    type Err = anyhow::Error;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        let mut split_iter = s.split("::");
+
+        let Some(path_string) = split_iter.next() else {
+            bail!("Could not find the path in the test specifier")
+        };
+        let path = PathBuf::from(path_string)
+            .canonicalize()
+            .context("Failed to canonicalize the path of the test")?;
+
+        let Some(case_idx_string) = split_iter.next() else {
+            return Ok(Self::FileOrDirectory {
+                metadata_or_directory_file_path: path,
+            });
+        };
+        let case_idx = usize::from_str(case_idx_string)
+            .context("Failed to parse the case idx of the test specifier from string")?;
+
+        // At this point the provided path must be a file.
+        if !path.is_file() {
+            bail!(
+                "Test specifier with a path and case idx must point to a file and not a directory"
+            )
+        }
+
+        let Some(mode_string) = split_iter.next() else {
+            return Ok(Self::Case {
+                metadata_file_path: path,
+                case_idx,
+            });
+        };
+        let mode = Mode::from_str(mode_string)
+            .context("Failed to parse the mode string in the parsed test specifier")?;
+
+        Ok(Self::CaseWithMode {
+            metadata_file_path: path,
+            case_idx,
+            mode,
+        })
+    }
+}
+
+impl From<ParsedTestSpecifier> for String {
+    fn from(value: ParsedTestSpecifier) -> Self {
+        value.to_string()
+    }
+}
+
+impl TryFrom<String> for ParsedTestSpecifier {
+    type Error = anyhow::Error;
+
+    fn try_from(value: String) -> Result<Self, Self::Error> {
+        value.parse()
+    }
+}
+
+impl TryFrom<&str> for ParsedTestSpecifier {
+    type Error = anyhow::Error;
+
+    fn try_from(value: &str) -> Result<Self, Self::Error> {
+        value.parse()
+    }
+}

crates/config/Cargo.toml

@@ -18,6 +18,7 @@ semver = { workspace = true }
 temp-dir = { workspace = true }
 serde = { workspace = true }
 serde_json = { workspace = true }
+serde_with = { workspace = true }
 strum = { workspace = true }
 
 [lints]

crates/config/src/lib.rs

@@ -17,7 +17,7 @@ use alloy::{
     signers::local::PrivateKeySigner,
 };
 use clap::{Parser, ValueEnum, ValueHint};
-use revive_dt_common::types::PlatformIdentifier;
+use revive_dt_common::types::{ParsedTestSpecifier, PlatformIdentifier};
 use semver::Version;
 use serde::{Deserialize, Serialize, Serializer};
 use strum::{AsRefStr, Display, EnumString, IntoStaticStr};
@@ -652,11 +652,24 @@ impl AsRef<WalletConfiguration> for ExportGenesisContext {
 }
 
 /// A set of configuration parameters for the corpus files to use for the execution.
+#[serde_with::serde_as]
 #[derive(Clone, Debug, Parser, Serialize, Deserialize)]
 pub struct CorpusConfiguration {
-    /// A list of test corpus JSON files to be tested.
-    #[arg(short = 'c', long = "corpus")]
-    pub paths: Vec<PathBuf>,
+    /// A list of test specifiers for the tests that the tool should run.
+    ///
+    /// Test specifiers follow the following format:
+    ///
+    /// - `{directory_path|metadata_file_path}`: A path to a metadata file where all of the cases
+    ///   live and should be run. Alternatively, it points to a directory instructing the framework
+    ///   to discover of the metadata files that live there an execute them.
+    /// - `{metadata_file_path}::{case_idx}`: The path to a metadata file and then a case idx
+    ///   separated by two colons. This specifies that only this specific test case within the
+    ///   metadata file should be executed.
+    /// - `{metadata_file_path}::{case_idx}::{mode}`: This is very similar to the above specifier
+    ///   with the exception that in this case the mode is specified and will be used in the test.
+    #[serde_as(as = "Vec<serde_with::DisplayFromStr>")]
+    #[arg(short = 't', long = "test")]
+    pub test_specifiers: Vec<ParsedTestSpecifier>,
 }
 
 /// A set of configuration parameters for Solc.

crates/core/src/differential_benchmarks/entry_point.rs

@@ -6,7 +6,10 @@ use anyhow::Context as _;
 use futures::{FutureExt, StreamExt};
 use revive_dt_common::types::PrivateKeyAllocator;
 use revive_dt_core::Platform;
-use revive_dt_format::steps::{Step, StepIdx, StepPath};
+use revive_dt_format::{
+    corpus::Corpus,
+    steps::{Step, StepIdx, StepPath},
+};
 use tokio::sync::Mutex;
 use tracing::{Instrument, error, info, info_span, instrument, warn};
 
@@ -15,7 +18,7 @@ use revive_dt_report::Reporter;
 
 use crate::{
     differential_benchmarks::{Driver, Watcher, WatcherEvent},
-    helpers::{CachedCompiler, NodePool, collect_metadata_files, create_test_definitions_stream},
+    helpers::{CachedCompiler, NodePool, create_test_definitions_stream},
 };
 
 /// Handles the differential testing executing it according to the information defined in the
@@ -39,9 +42,17 @@ pub async fn handle_differential_benchmarks(
     let full_context = Context::Benchmark(Box::new(context.clone()));
 
     // Discover all of the metadata files that are defined in the context.
-    let metadata_files = collect_metadata_files(&context)
-        .context("Failed to collect metadata files for differential testing")?;
-    info!(len = metadata_files.len(), "Discovered metadata files");
+    let corpus = context
+        .corpus_configuration
+        .test_specifiers
+        .clone()
+        .into_iter()
+        .try_fold(Corpus::default(), Corpus::with_test_specifier)
+        .context("Failed to parse the test corpus")?;
+    info!(
+        len = corpus.metadata_file_count(),
+        "Discovered metadata files"
+    );
 
     // Discover the list of platforms that the tests should run on based on the context.
     let platforms = context
@@ -84,7 +95,7 @@ pub async fn handle_differential_benchmarks(
     // Preparing test definitions for the execution.
     let test_definitions = create_test_definitions_stream(
         &full_context,
-        metadata_files.iter(),
+        &corpus,
         &platforms_and_nodes,
         None,
         reporter.clone(),

crates/core/src/differential_tests/entry_point.rs

@@ -12,6 +12,7 @@ use anyhow::Context as _;
 use futures::{FutureExt, StreamExt};
 use revive_dt_common::{cached_fs::read_to_string, types::PrivateKeyAllocator};
 use revive_dt_core::Platform;
+use revive_dt_format::corpus::Corpus;
 use tokio::sync::{Mutex, RwLock, Semaphore};
 use tracing::{Instrument, error, info, info_span, instrument};
 
@@ -20,7 +21,7 @@ use revive_dt_report::{Reporter, ReporterEvent, TestCaseStatus};
 
 use crate::{
     differential_tests::Driver,
-    helpers::{CachedCompiler, NodePool, collect_metadata_files, create_test_definitions_stream},
+    helpers::{CachedCompiler, NodePool, create_test_definitions_stream},
 };
 
 /// Handles the differential testing executing it according to the information defined in the
@@ -33,9 +34,17 @@ pub async fn handle_differential_tests(
     let reporter_clone = reporter.clone();
 
     // Discover all of the metadata files that are defined in the context.
-    let metadata_files = collect_metadata_files(&context)
-        .context("Failed to collect metadata files for differential testing")?;
-    info!(len = metadata_files.len(), "Discovered metadata files");
+    let corpus = context
+        .corpus_configuration
+        .test_specifiers
+        .clone()
+        .into_iter()
+        .try_fold(Corpus::default(), Corpus::with_test_specifier)
+        .context("Failed to parse the test corpus")?;
+    info!(
+        len = corpus.metadata_file_count(),
+        "Discovered metadata files"
+    );
 
     // Discover the list of platforms that the tests should run on based on the context.
     let platforms = context
@@ -83,7 +92,7 @@ pub async fn handle_differential_tests(
     let full_context = Context::Test(Box::new(context.clone()));
     let test_definitions = create_test_definitions_stream(
         &full_context,
-        metadata_files.iter(),
+        &corpus,
         &platforms_and_nodes,
         only_execute_failed_tests.as_ref(),
         reporter.clone(),

crates/core/src/helpers/metadata.rs

@@ -1,33 +0,0 @@
-use revive_dt_config::CorpusConfiguration;
-use revive_dt_format::{corpus::Corpus, metadata::MetadataFile};
-use tracing::{info, info_span, instrument};
-
-/// Given an object that implements [`AsRef<CorpusConfiguration>`], this function finds all of the
-/// corpus files and produces a map containing all of the [`MetadataFile`]s discovered.
-#[instrument(level = "debug", name = "Collecting Corpora", skip_all)]
-pub fn collect_metadata_files(
-    context: impl AsRef<CorpusConfiguration>,
-) -> anyhow::Result<Vec<MetadataFile>> {
-    let mut metadata_files = Vec::new();
-
-    let corpus_configuration = AsRef::<CorpusConfiguration>::as_ref(&context);
-    for path in &corpus_configuration.paths {
-        let span = info_span!("Processing corpus file", path = %path.display());
-        let _guard = span.enter();
-
-        let corpus = Corpus::try_from_path(path)?;
-        info!(
-            name = corpus.name(),
-            number_of_contained_paths = corpus.path_count(),
-            "Deserialized corpus file"
-        );
-        metadata_files.extend(corpus.enumerate_tests());
-    }
-
-    // There's a possibility that there are certain paths that all lead to the same metadata files
-    // and therefore it's important that we sort them and then deduplicate them.
-    metadata_files.sort_by(|a, b| a.metadata_file_path.cmp(&b.metadata_file_path));
-    metadata_files.dedup_by(|a, b| a.metadata_file_path == b.metadata_file_path);
-
-    Ok(metadata_files)
-}

crates/core/src/helpers/mod.rs

@@ -1,9 +1,7 @@
 mod cached_compiler;
-mod metadata;
 mod pool;
 mod test;
 
 pub use cached_compiler::*;
-pub use metadata::*;
 pub use pool::*;
 pub use test::*;

crates/core/src/helpers/test.rs

@@ -4,9 +4,9 @@ use std::{borrow::Cow, path::Path};
 
 use futures::{Stream, StreamExt, stream};
 use indexmap::{IndexMap, indexmap};
-use revive_dt_common::iterators::EitherIter;
-use revive_dt_common::types::{ParsedMode, PlatformIdentifier};
+use revive_dt_common::types::PlatformIdentifier;
 use revive_dt_config::Context;
+use revive_dt_format::corpus::Corpus;
 use serde_json::{Value, json};
 
 use revive_dt_compiler::Mode;
@@ -27,35 +27,17 @@ pub async fn create_test_definitions_stream<'a>(
     // This is only required for creating the compiler objects and is not used anywhere else in the
     // function.
     context: &Context,
-    metadata_files: impl IntoIterator<Item = &'a MetadataFile>,
+    corpus: &'a Corpus,
     platforms_and_nodes: &'a BTreeMap<PlatformIdentifier, (&dyn Platform, NodePool)>,
     only_execute_failed_tests: Option<&Report>,
     reporter: Reporter,
 ) -> impl Stream<Item = TestDefinition<'a>> {
     stream::iter(
-        metadata_files
-            .into_iter()
-            // Flatten over the cases.
-            .flat_map(|metadata_file| {
-                metadata_file
-                    .cases
-                    .iter()
-                    .enumerate()
-                    .map(move |(case_idx, case)| (metadata_file, case_idx, case))
-            })
-            // Flatten over the modes, prefer the case modes over the metadata file modes.
-            .flat_map(move |(metadata_file, case_idx, case)| {
+        corpus
+            .cases_iterator()
+            .map(move |(metadata_file, case_idx, case, mode)| {
                 let reporter = reporter.clone();
 
-                let modes = case.modes.as_ref().or(metadata_file.modes.as_ref());
-                let modes = match modes {
-                    Some(modes) => EitherIter::A(
-                        ParsedMode::many_to_modes(modes.iter()).map(Cow::<'static, _>::Owned),
-                    ),
-                    None => EitherIter::B(Mode::all().map(Cow::<'static, _>::Borrowed)),
-                };
-
-                modes.into_iter().map(move |mode| {
                 (
                     metadata_file,
                     case_idx,
@@ -68,7 +50,6 @@ pub async fn create_test_definitions_stream<'a>(
                     })),
                 )
             })
-            })
             // Inform the reporter of each one of the test cases that were discovered which we expect to
             // run.
             .inspect(|(_, _, _, _, reporter)| {

crates/format/Cargo.toml

@@ -21,6 +21,7 @@ schemars = { workspace = true }
 semver = { workspace = true }
 serde = { workspace = true, features = ["derive"] }
 serde_json = { workspace = true }
+itertools = { workspace = true }
 
 [dev-dependencies]
 tokio = { workspace = true }

crates/format/src/corpus.rs

@@ -1,56 +1,167 @@
 use std::{
-    fs::File,
+    borrow::Cow,
+    collections::HashMap,
     path::{Path, PathBuf},
 };
 
-use revive_dt_common::iterators::FilesWithExtensionIterator;
-use serde::{Deserialize, Serialize};
-use tracing::{debug, info};
+use itertools::Itertools;
+use revive_dt_common::{
+    iterators::{EitherIter, FilesWithExtensionIterator},
+    types::{Mode, ParsedMode, ParsedTestSpecifier},
+};
+use tracing::{debug, warn};
 
-use crate::metadata::{Metadata, MetadataFile};
-use anyhow::Context as _;
+use crate::{
+    case::{Case, CaseIdx},
+    metadata::{Metadata, MetadataFile},
+};
 
-#[derive(Clone, Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Serialize, Deserialize)]
-#[serde(untagged)]
-pub enum Corpus {
-    SinglePath { name: String, path: PathBuf },
-    MultiplePaths { name: String, paths: Vec<PathBuf> },
+#[derive(Default)]
+pub struct Corpus {
+    test_specifiers: HashMap<ParsedTestSpecifier, Vec<PathBuf>>,
+    metadata_files: HashMap<PathBuf, MetadataFile>,
 }
 
 impl Corpus {
-    pub fn try_from_path(file_path: impl AsRef<Path>) -> anyhow::Result<Self> {
-        let mut corpus = File::open(file_path.as_ref())
-            .map_err(anyhow::Error::from)
-            .and_then(|file| serde_json::from_reader::<_, Corpus>(file).map_err(Into::into))
-            .with_context(|| {
-                format!(
-                    "Failed to open and deserialize corpus file at {}",
-                    file_path.as_ref().display()
+    pub fn new() -> Self {
+        Default::default()
+    }
+
+    pub fn with_test_specifier(
+        mut self,
+        test_specifier: ParsedTestSpecifier,
+    ) -> anyhow::Result<Self> {
+        match &test_specifier {
+            ParsedTestSpecifier::FileOrDirectory {
+                metadata_or_directory_file_path: metadata_file_path,
+            }
+            | ParsedTestSpecifier::Case {
+                metadata_file_path, ..
+            }
+            | ParsedTestSpecifier::CaseWithMode {
+                metadata_file_path, ..
+            } => {
+                let metadata_files = enumerate_metadata_files(metadata_file_path);
+                self.test_specifiers.insert(
+                    test_specifier,
+                    metadata_files
+                        .iter()
+                        .map(|metadata_file| metadata_file.metadata_file_path.clone())
+                        .collect(),
+                );
+                for metadata_file in metadata_files.into_iter() {
+                    self.metadata_files
+                        .insert(metadata_file.metadata_file_path.clone(), metadata_file);
+                }
+            }
+        };
+
+        Ok(self)
+    }
+
+    pub fn cases_iterator(
+        &self,
+    ) -> impl Iterator<Item = (&'_ MetadataFile, CaseIdx, &'_ Case, Cow<'_, Mode>)> + '_ {
+        let mut iterator = Box::new(std::iter::empty())
+            as Box<dyn Iterator<Item = (&'_ MetadataFile, CaseIdx, &'_ Case, Cow<'_, Mode>)> + '_>;
+
+        for (test_specifier, metadata_file_paths) in self.test_specifiers.iter() {
+            for metadata_file_path in metadata_file_paths {
+                let metadata_file = self
+                    .metadata_files
+                    .get(metadata_file_path)
+                    .expect("Must succeed");
+
+                match test_specifier {
+                    ParsedTestSpecifier::FileOrDirectory { .. } => {
+                        for (case_idx, case) in metadata_file.cases.iter().enumerate() {
+                            let case_idx = CaseIdx::new(case_idx);
+
+                            let modes = case.modes.as_ref().or(metadata_file.modes.as_ref());
+                            let modes = match modes {
+                                Some(modes) => EitherIter::A(
+                                    ParsedMode::many_to_modes(modes.iter())
+                                        .map(Cow::<'static, _>::Owned),
+                                ),
+                                None => EitherIter::B(Mode::all().map(Cow::<'static, _>::Borrowed)),
+                            };
+
+                            iterator = Box::new(
+                                iterator.chain(
+                                    modes
+                                        .into_iter()
+                                        .map(move |mode| (metadata_file, case_idx, case, mode)),
+                                ),
                             )
-            })?;
+                        }
+                    }
+                    ParsedTestSpecifier::Case { case_idx, .. } => {
+                        let Some(case) = metadata_file.cases.get(*case_idx) else {
+                            warn!(
+                                test_specifier = %test_specifier,
+                                metadata_file_path = %metadata_file_path.display(),
+                                case_idx = case_idx,
+                                case_count = metadata_file.cases.len(),
+                                "Specified case not found in metadata file"
+                            );
+                            continue;
+                        };
+                        let case_idx = CaseIdx::new(*case_idx);
 
-        let corpus_directory = file_path
-            .as_ref()
-            .canonicalize()
-            .context("Failed to canonicalize the path to the corpus file")?
-            .parent()
-            .context("Corpus file has no parent")?
-            .to_path_buf();
+                        let modes = case.modes.as_ref().or(metadata_file.modes.as_ref());
+                        let modes = match modes {
+                            Some(modes) => EitherIter::A(
+                                ParsedMode::many_to_modes(modes.iter())
+                                    .map(Cow::<'static, Mode>::Owned),
+                            ),
+                            None => EitherIter::B(Mode::all().map(Cow::<'static, _>::Borrowed)),
+                        };
 
-        for path in corpus.paths_iter_mut() {
-            *path = corpus_directory.join(path.as_path())
+                        iterator = Box::new(
+                            iterator.chain(
+                                modes
+                                    .into_iter()
+                                    .map(move |mode| (metadata_file, case_idx, case, mode)),
+                            ),
+                        )
+                    }
+                    ParsedTestSpecifier::CaseWithMode { case_idx, mode, .. } => {
+                        let Some(case) = metadata_file.cases.get(*case_idx) else {
+                            warn!(
+                                test_specifier = %test_specifier,
+                                metadata_file_path = %metadata_file_path.display(),
+                                case_idx = case_idx,
+                                case_count = metadata_file.cases.len(),
+                                "Specified case not found in metadata file"
+                            );
+                            continue;
+                        };
+                        let case_idx = CaseIdx::new(*case_idx);
+
+                        let mode = Cow::Borrowed(mode);
+                        iterator = Box::new(iterator.chain(std::iter::once((
+                            metadata_file,
+                            case_idx,
+                            case,
+                            mode,
+                        ))))
+                    }
+                }
+            }
         }
 
-        Ok(corpus)
+        iterator.unique_by(|item| (&item.0.metadata_file_path, item.1, item.3.clone()))
     }
 
-    pub fn enumerate_tests(&self) -> Vec<MetadataFile> {
-        let mut tests = self
-            .paths_iter()
-            .flat_map(|root_path| {
-                if !root_path.is_dir() {
-                    Box::new(std::iter::once(root_path.to_path_buf()))
-                        as Box<dyn Iterator<Item = _>>
+    pub fn metadata_file_count(&self) -> usize {
+        self.metadata_files.len()
+    }
+}
+
+fn enumerate_metadata_files(path: impl AsRef<Path>) -> Vec<MetadataFile> {
+    let root_path = path.as_ref();
+    let mut tests = if !root_path.is_dir() {
+        Box::new(std::iter::once(root_path.to_path_buf())) as Box<dyn Iterator<Item = _>>
     } else {
         Box::new(
             FilesWithExtensionIterator::new(root_path)
@@ -60,7 +171,6 @@ impl Corpus {
         )
     }
     .map(move |metadata_file_path| (root_path, metadata_file_path))
-            })
     .filter_map(|(root_path, metadata_file_path)| {
         Metadata::try_from_file(&metadata_file_path)
             .or_else(|| {
@@ -86,46 +196,5 @@ impl Corpus {
     .collect::<Vec<_>>();
     tests.sort_by(|a, b| a.metadata_file_path.cmp(&b.metadata_file_path));
     tests.dedup_by(|a, b| a.metadata_file_path == b.metadata_file_path);
-        info!(
-            len = tests.len(),
-            corpus_name = self.name(),
-            "Found tests in Corpus"
-        );
     tests
 }
-
-    pub fn name(&self) -> &str {
-        match self {
-            Corpus::SinglePath { name, .. } | Corpus::MultiplePaths { name, .. } => name.as_str(),
-        }
-    }
-
-    pub fn paths_iter(&self) -> impl Iterator<Item = &Path> {
-        match self {
-            Corpus::SinglePath { path, .. } => {
-                Box::new(std::iter::once(path.as_path())) as Box<dyn Iterator<Item = _>>
-            }
-            Corpus::MultiplePaths { paths, .. } => {
-                Box::new(paths.iter().map(|path| path.as_path())) as Box<dyn Iterator<Item = _>>
-            }
-        }
-    }
-
-    pub fn paths_iter_mut(&mut self) -> impl Iterator<Item = &mut PathBuf> {
-        match self {
-            Corpus::SinglePath { path, .. } => {
-                Box::new(std::iter::once(path)) as Box<dyn Iterator<Item = _>>
-            }
-            Corpus::MultiplePaths { paths, .. } => {
-                Box::new(paths.iter_mut()) as Box<dyn Iterator<Item = _>>
-            }
-        }
-    }
-
-    pub fn path_count(&self) -> usize {
-        match self {
-            Corpus::SinglePath { .. } => 1,
-            Corpus::MultiplePaths { paths, .. } => paths.len(),
-        }
-    }
-}

crates/report/src/aggregator.rs

@@ -11,10 +11,10 @@ use std::{
 use alloy::primitives::Address;
 use anyhow::{Context as _, Result};
 use indexmap::IndexMap;
-use revive_dt_common::types::PlatformIdentifier;
+use revive_dt_common::types::{ParsedTestSpecifier, PlatformIdentifier};
 use revive_dt_compiler::{CompilerInput, CompilerOutput, Mode};
 use revive_dt_config::Context;
-use revive_dt_format::{case::CaseIdx, corpus::Corpus, metadata::ContractInstance};
+use revive_dt_format::{case::CaseIdx, metadata::ContractInstance};
 use semver::Version;
 use serde::{Deserialize, Serialize};
 use serde_with::{DisplayFromStr, serde_as};
@@ -67,7 +67,7 @@ impl ReportAggregator {
                 RunnerEvent::SubscribeToEvents(event) => {
                     self.handle_subscribe_to_events_event(*event);
                 }
-                RunnerEvent::CorpusFileDiscovery(event) => {
+                RunnerEvent::CorpusDiscovery(event) => {
                     self.handle_corpus_file_discovered_event(*event)
                 }
                 RunnerEvent::MetadataFileDiscovery(event) => {
@@ -152,8 +152,8 @@ impl ReportAggregator {
         let _ = event.tx.send(self.listener_tx.subscribe());
     }
 
-    fn handle_corpus_file_discovered_event(&mut self, event: CorpusFileDiscoveryEvent) {
-        self.report.corpora.push(event.corpus);
+    fn handle_corpus_file_discovered_event(&mut self, event: CorpusDiscoveryEvent) {
+        self.report.corpora.extend(event.test_specifiers);
     }
 
     fn handle_metadata_file_discovery_event(&mut self, event: MetadataFileDiscoveryEvent) {
@@ -420,7 +420,8 @@ pub struct Report {
     /// The context that the tool was started up with.
     pub context: Context,
     /// The list of corpus files that the tool found.
-    pub corpora: Vec<Corpus>,
+    #[serde_as(as = "Vec<DisplayFromStr>")]
+    pub corpora: Vec<ParsedTestSpecifier>,
     /// The list of metadata files that were found by the tool.
     pub metadata_files: BTreeSet<MetadataFilePath>,
     /// Information relating to each test case.

crates/report/src/runner_event.rs

@@ -6,10 +6,11 @@ use std::{collections::BTreeMap, path::PathBuf, sync::Arc};
 use alloy::primitives::Address;
 use anyhow::Context as _;
 use indexmap::IndexMap;
+use revive_dt_common::types::ParsedTestSpecifier;
 use revive_dt_common::types::PlatformIdentifier;
 use revive_dt_compiler::{CompilerInput, CompilerOutput};
+use revive_dt_format::metadata::ContractInstance;
 use revive_dt_format::metadata::Metadata;
-use revive_dt_format::{corpus::Corpus, metadata::ContractInstance};
 use semver::Version;
 use tokio::sync::{broadcast, oneshot};
 
@@ -481,9 +482,9 @@ define_event! {
             tx: oneshot::Sender<broadcast::Receiver<ReporterEvent>>
         },
         /// An event emitted by runners when they've discovered a corpus file.
-        CorpusFileDiscovery {
+        CorpusDiscovery {
             /// The contents of the corpus file.
-            corpus: Corpus
+            test_specifiers: Vec<ParsedTestSpecifier>
         },
         /// An event emitted by runners when they've discovered a metadata file.
         MetadataFileDiscovery {

run_tests.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 # Revive Differential Tests - Quick Start Script
-# This script clones the test repository, sets up the corpus file, and runs the tool
+# This script clones the test repository, and runs the tool
 
 set -e  # Exit on any error
 
@@ -14,7 +14,6 @@ NC='\033[0m' # No Color
 # Configuration
 TEST_REPO_URL="https://github.com/paritytech/resolc-compiler-tests"
 TEST_REPO_DIR="resolc-compiler-tests"
-CORPUS_FILE="./corpus.json"
 WORKDIR="workdir"
 
 # Optional positional argument: path to polkadot-sdk directory
@@ -68,21 +67,6 @@ else
     echo -e "${YELLOW}No polkadot-sdk path provided. Using binaries from $PATH.${NC}"
 fi
 
-# Create corpus file with absolute path resolved at runtime
-echo -e "${GREEN}Creating corpus file...${NC}"
-ABSOLUTE_PATH=$(realpath "$TEST_REPO_DIR/fixtures/solidity/")
-
-cat > "$CORPUS_FILE" << EOF
-{
-  "name": "MatterLabs Solidity Simple, Complex, and Semantic Tests",
-  "paths": [
-    "$(realpath "$TEST_REPO_DIR/fixtures/solidity")"
-  ]
-}
-EOF
-
-echo -e "${GREEN}Corpus file created: $CORPUS_FILE${NC}"
-
 # Create workdir if it doesn't exist
 mkdir -p "$WORKDIR"
 
@@ -94,7 +78,7 @@ echo ""
 cargo build --release;
 RUST_LOG="info,alloy_pubsub::service=error" ./target/release/retester test \
     --platform revive-dev-node-polkavm-resolc \
-    --corpus "$CORPUS_FILE" \
+    --test $(realpath "$TEST_REPO_DIR/fixtures/solidity") \
     --working-directory "$WORKDIR" \
     --concurrency.number-of-nodes 10 \
     --concurrency.number-of-threads 5 \