Split binding rules into multiple parts and process built-ins with a subset

crates/codegen/language/definition/src/model/manifest.rs

@@ -14,7 +14,7 @@ pub struct Language {
     pub name: Identifier,
 
     pub documentation_dir: PathBuf,
-    pub binding_rules_file: PathBuf,
+    pub binding_rules_dir: PathBuf,
     pub file_extension: Option<String>,
 
     pub root_item: Identifier,

crates/codegen/language/tests/src/fail/definitions/duplicate_expression_name/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/definitions/duplicate_item_name/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar1,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/definitions/duplicate_variant_name/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/definitions/operator_mismatch/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/parsing/duplicate_map_key/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/parsing/duplicate_set_entry/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/parsing/empty_string/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/parsing/missing_field/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/parsing/unrecognized_field/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/parsing/unrecognized_variant/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar1,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/reachability/unreachable/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar1,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/reachability/unused_version/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/references/disabled_too_late/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = One,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/references/enabled_too_early/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = One,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/references/invalid_filter/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/references/invalid_version/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/references/unknown_reference/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = Bar,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/references/unordered_version_pair/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = One,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/fail/references/version_not_found/test.rs

@@ -3,7 +3,7 @@
 codegen_language_macros::compile!(Language(
     name = Foo,
     documentation_dir = "foo/bar",
-    binding_rules_file = "bindings/rules.msgb",
+    binding_rules_dir = "bindings",
     root_item = One,
     leading_trivia = Sequence([]),
     trailing_trivia = Sequence([]),

crates/codegen/language/tests/src/pass/tiny_language.rs

@@ -10,7 +10,7 @@ use semver::Version;
 codegen_language_macros::compile!(Language(
     name = Tiny,
     documentation_dir = "tiny/docs",
-    binding_rules_file = "tiny/bindings/rules.msgb",
+    binding_rules_dir = "tiny/bindings",
     file_extension = ".tiny",
     root_item = Foo,
     leading_trivia = Sequence([]),
@@ -50,7 +50,7 @@ fn definition() {
         Language {
             name: "Tiny".into(),
             documentation_dir: Path::repo_path("tiny/docs"),
-            binding_rules_file: Path::repo_path("tiny/bindings/rules.msgb"),
+            binding_rules_dir: Path::repo_path("tiny/bindings"),
             file_extension: Some(".tiny".into()),
             root_item: "Foo".into(),
             leading_trivia: TriviaParser::Sequence { parsers: [].into() },

crates/codegen/runtime/cargo/crate/src/extensions/bindings/mod.rs

@@ -6,7 +6,7 @@ use crate::parser::ParserInitializationError;
 #[allow(clippy::needless_pass_by_value)]
 pub fn add_built_ins(
     _binding_graph: &mut BindingGraph,
-    _version: Version,
+    _version: &Version,
 ) -> Result<(), ParserInitializationError> {
     unreachable!("Built-ins are Solidity-specific")
 }

crates/codegen/runtime/cargo/crate/src/runtime/bindings/binding_rules.rs.jinja2

@@ -1,5 +1,9 @@
 #[allow(clippy::needless_raw_string_hashes)]
-#[allow(dead_code)] // TODO(#982): use to create the graph
-pub const BINDING_RULES_SOURCE: &str = r#####"
-    {{ model.bindings.binding_rules_source }}
+pub const USER_BINDING_RULES_SOURCE: &str = r#####"
+    {{ model.bindings.user_binding_rules_source }}
 "#####;
+
+#[allow(clippy::needless_raw_string_hashes)]
+pub const SYSTEM_BINDING_RULES_SOURCE: &str = r#####"
+    {{ model.bindings.system_binding_rules_source }}
+"#####;

crates/codegen/runtime/cargo/crate/src/runtime/bindings/mod.rs

@@ -26,22 +26,29 @@ pub enum BindingGraphInitializationError {
     ParserInitialization(#[from] ParserInitializationError),
 }
 
+#[allow(clippy::needless_pass_by_value)]
 pub fn create_with_resolver(
     version: Version,
     resolver: Rc<dyn PathResolver<KindTypes>>,
 ) -> Result<BindingGraph, ParserInitializationError> {
     let mut binding_graph = BindingGraph::create(
         version.clone(),
-        binding_rules::BINDING_RULES_SOURCE,
+        binding_rules::USER_BINDING_RULES_SOURCE,
+        binding_rules::SYSTEM_BINDING_RULES_SOURCE,
         resolver,
     );
 
-    crate::extensions::bindings::add_built_ins(&mut binding_graph, version)?;
+    crate::extensions::bindings::add_built_ins(&mut binding_graph, &version)?;
 
     Ok(binding_graph)
 }
 
 #[cfg(feature = "__private_testing_utils")]
-pub fn get_binding_rules() -> &'static str {
-    binding_rules::BINDING_RULES_SOURCE
+pub fn get_user_binding_rules() -> &'static str {
+    binding_rules::USER_BINDING_RULES_SOURCE
+}
+
+#[cfg(feature = "__private_testing_utils")]
+pub fn get_system_binding_rules() -> &'static str {
+    binding_rules::SYSTEM_BINDING_RULES_SOURCE
 }

crates/codegen/runtime/generator/src/bindings/mod.rs

@@ -1,32 +1,48 @@
 use std::collections::BTreeSet;
+use std::path::Path;
 
 use anyhow::Result;
 use codegen_language_definition::model;
 use infra_utils::codegen::CodegenFileSystem;
-use infra_utils::paths::PathExtensions;
 use semver::Version;
 use serde::Serialize;
 
 #[derive(Default, Serialize)]
 pub struct BindingsModel {
-    binding_rules_source: String,
+    user_binding_rules_source: String,
+    system_binding_rules_source: String,
     built_ins_versions: BTreeSet<Version>,
     file_extension: String,
 }
 
 impl BindingsModel {
     pub fn from_language(language: &model::Language) -> Result<Self> {
-        // We use `CodegenFileSystem` here to ensure the rules are rebuilt if the rules file changes
-        let binding_rules_dir = language.binding_rules_file.unwrap_parent();
-        let mut fs = CodegenFileSystem::new(binding_rules_dir)?;
-        let binding_rules_source = fs.read_file(&language.binding_rules_file)?;
+        let binding_rules_dir = &language.binding_rules_dir;
+        let user_binding_rules_source = assemble_rules(binding_rules_dir, "user-rules.parts")?;
+        let system_binding_rules_source = assemble_rules(binding_rules_dir, "system-rules.parts")?;
         let built_ins_versions = language.collect_built_ins_versions();
         let file_extension = language.file_extension.clone().unwrap_or_default();
 
         Ok(Self {
-            binding_rules_source,
+            user_binding_rules_source,
+            system_binding_rules_source,
             built_ins_versions,
             file_extension,
         })
     }
 }
+
+// Builds a rules file by concatenating the file parts listed in the given `parts_file`
+fn assemble_rules(rules_dir: &Path, parts_file: &str) -> Result<String> {
+    // We use `CodegenFileSystem` here to ensure the rules are rebuilt if the rules file changes
+    let mut fs = CodegenFileSystem::new(rules_dir)?;
+    let parts_contents = fs.read_file(rules_dir.join(parts_file))?;
+    let mut parts = Vec::new();
+    for part_name in parts_contents.lines() {
+        if part_name.is_empty() {
+            continue;
+        }
+        parts.push(fs.read_file(rules_dir.join(part_name))?);
+    }
+    Ok(parts.join("\n"))
+}