Get rid of corpus files (#190)

* Get rid of corpus files

* Update the readme

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;
@@ -27,47 +27,28 @@ 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,
-                        case,
-                        mode.clone(),
-                        reporter.test_specific_reporter(Arc::new(TestSpecifier {
-                            solc_mode: mode.as_ref().clone(),
-                            metadata_file_path: metadata_file.metadata_file_path.clone(),
-                            case_idx: CaseIdx::new(case_idx),
-                        })),
-                    )
-                })
+                (
+                    metadata_file,
+                    case_idx,
+                    case,
+                    mode.clone(),
+                    reporter.test_specific_reporter(Arc::new(TestSpecifier {
+                        solc_mode: mode.as_ref().clone(),
+                        metadata_file_path: metadata_file.metadata_file_path.clone(),
+                        case_idx: CaseIdx::new(case_idx),
+                    })),
+                )
             })
             // Inform the reporter of each one of the test cases that were discovered which we expect to
             // run.