chore: refactor global type system to accommodate further extensions (#8253)

Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
Co-authored-by: Arend van Beelen jr. <[email protected]>

Author: 3 people

Cargo.lock

@@ -22,6 +22,7 @@ biome_resolver            = { workspace = true }
 biome_rowan               = { workspace = true }
 camino                    = { workspace = true }
 hashbrown                 = { workspace = true }
+paste                     = "1.0"
 rustc-hash                = { workspace = true }
 
 [dev-dependencies]

crates/biome_js_type_info/Cargo.toml

@@ -589,8 +589,12 @@ impl Format<FormatTypeContext> for TypeReference {
                 let level = resolved.level();
                 let id = resolved.id();
                 if level == TypeResolverLevel::Global {
-                    if resolved.index() < NUM_PREDEFINED_TYPES {
-                        write!(f, [token(global_type_name(id))])
+                    // GlobalsResolverBuilder makes sure the type store is fully filled.
+                    // Every global TypeId whose index is less than NUM_PREDEFINED_TYPES
+                    // must have a name returned by global_type_name().
+                    // GLOBAL_TYPE_MEMBERS ensures this invariant.
+                    if let Some(name) = global_type_name(id) {
+                        write!(f, [token(name)])
                     } else {
                         // Start counting from `NUM_PREDEFINED_TYPES` so
                         // snapshots remain stable even if we add new predefined

crates/biome_js_type_info/src/format_type_info.rs

@@ -0,0 +1,55 @@
+//! Builder for constructing GlobalsResolver with forward references support.
+
+use std::sync::Arc;
+
+use crate::{NUM_PREDEFINED_TYPES, TypeData, TypeId, TypeStore};
+
+use super::globals::GlobalsResolver;
+
+/// Builder for constructing a GlobalsResolver
+pub struct GlobalsResolverBuilder {
+    /// Types being built. None = reserved but not yet filled.
+    types: Vec<Option<TypeData>>,
+}
+
+impl GlobalsResolverBuilder {
+    pub fn with_capacity(capacity: usize) -> Self {
+        Self {
+            types: vec![None; capacity],
+        }
+    }
+
+    /// Fill a previously reserved type slot with actual type data.
+    pub fn set_type_data(&mut self, id: TypeId, data: TypeData) {
+        let index = id.index();
+        debug_assert!(
+            index < self.types.len(),
+            "TypeId {index} out of bounds (len: {})",
+            self.types.len()
+        );
+        debug_assert!(
+            self.types[index].is_none(),
+            "Type at index {index} already set"
+        );
+        self.types[index] = Some(data);
+    }
+
+    /// Build the final GlobalsResolver.
+    pub fn build(self) -> GlobalsResolver {
+        let types: Vec<Arc<TypeData>> = self
+            .types
+            .into_iter()
+            .map(|opt| Arc::new(opt.unwrap_or(TypeData::Unknown)))
+            .collect();
+
+        GlobalsResolver {
+            types: TypeStore::from_types(types),
+        }
+    }
+}
+
+impl Default for GlobalsResolverBuilder {
+    fn default() -> Self {
+        Self::with_capacity(NUM_PREDEFINED_TYPES)
+    }
+}

crates/biome_js_type_info/src/globals.rs

@@ -0,0 +1,116 @@
+//! Type ID constants for global types.
+//!
+//! TODO(tidefield): Implement a codegen for this file from TypeScript .d.ts files.
+
+// use crate::define_global_type;
+use crate::{ResolvedTypeId, TypeId};
+
+use super::globals::GLOBAL_LEVEL;
+
+// FIXME(tidefield): Get rid of the macro when implementing the codegen to improve compile time
+// Right now, I'm preserving the names so that the snapshot tests don't break the snapshot tests
+// to make sure I'm not breaking anything.
+#[macro_export]
+macro_rules! define_global_type {
+    ($name:ident, $index:expr, $name_str:expr) => {
+        pub const $name: TypeId = TypeId::new($index);
+        paste::paste! {
+            pub const [<$name _NAME>]: &str = $name_str;
+        }
+    };
+}
+// define_global_type!(ARRAY_ID, 8, "Array");         // Creates ARRAY_ID and ARRAY_ID_NAME
+
+// Type ID constants with their names defined together
+define_global_type!(UNKNOWN_ID, 0, "unknown");
+define_global_type!(UNDEFINED_ID, 1, "undefined");
+define_global_type!(VOID_ID, 2, "void");
+define_global_type!(CONDITIONAL_ID, 3, "conditional");
+define_global_type!(NUMBER_ID, 4, "number");
+define_global_type!(STRING_ID, 5, "string");
+define_global_type!(INSTANCEOF_ARRAY_T_ID, 6, "instanceof Array<T>");
+define_global_type!(INSTANCEOF_ARRAY_U_ID, 7, "instanceof Array<U>");
+define_global_type!(ARRAY_ID, 8, "Array");
+define_global_type!(ARRAY_FILTER_ID, 9, "Array.prototype.filter");
+define_global_type!(ARRAY_FOREACH_ID, 10, "Array.prototype.forEach");
+define_global_type!(ARRAY_MAP_ID, 11, "Array.prototype.map");
+define_global_type!(GLOBAL_ID, 12, "globalThis");
+define_global_type!(INSTANCEOF_PROMISE_ID, 13, "instanceof Promise");
+define_global_type!(PROMISE_ID, 14, "Promise");
+define_global_type!(PROMISE_CONSTRUCTOR_ID, 15, "Promise.constructor");
+define_global_type!(PROMISE_CATCH_ID, 16, "Promise.prototype.catch");
+define_global_type!(PROMISE_FINALLY_ID, 17, "Promise.prototype.finally");
+define_global_type!(PROMISE_THEN_ID, 18, "Promise.prototype.then");
+define_global_type!(PROMISE_ALL_ID, 19, "Promise.all");
+define_global_type!(PROMISE_ALL_SETTLED_ID, 20, "Promise.allSettled");
+define_global_type!(PROMISE_ANY_ID, 21, "Promise.any");
+define_global_type!(PROMISE_RACE_ID, 22, "Promise.race");
+define_global_type!(PROMISE_REJECT_ID, 23, "Promise.reject");
+define_global_type!(PROMISE_RESOLVE_ID, 24, "Promise.resolve");
+define_global_type!(PROMISE_TRY_ID, 25, "Promise.try");
+define_global_type!(BIGINT_STRING_LITERAL_ID, 26, "\"bigint\"");
+define_global_type!(BOOLEAN_STRING_LITERAL_ID, 27, "\"boolean\"");
+define_global_type!(FUNCTION_STRING_LITERAL_ID, 28, "\"function\"");
+define_global_type!(NUMBER_STRING_LITERAL_ID, 29, "\"number\"");
+define_global_type!(OBJECT_STRING_LITERAL_ID, 30, "\"object\"");
+define_global_type!(STRING_STRING_LITERAL_ID, 31, "\"string\"");
+define_global_type!(SYMBOL_STRING_LITERAL_ID, 32, "\"symbol\"");
+define_global_type!(UNDEFINED_STRING_LITERAL_ID, 33, "\"undefined\"");
+define_global_type!(
+    TYPEOF_OPERATOR_RETURN_UNION_ID,
+    34,
+    "\"bigint\" | \"boolean\" | \"function\" | \"number\" | \"object\" | \"string\" | \"symbol\" | \"undefined\""
+);
+define_global_type!(T_ID, 35, "T");
+define_global_type!(U_ID, 36, "U");
+define_global_type!(CONDITIONAL_CALLBACK_ID, 37, "() => conditional");
+define_global_type!(MAP_CALLBACK_ID, 38, "<U>(item: T) => U");
+define_global_type!(VOID_CALLBACK_ID, 39, "() => void");
+define_global_type!(FETCH_ID, 40, "fetch");
+define_global_type!(INSTANCEOF_REGEXP_ID, 41, "instanceof RegExp");
+define_global_type!(REGEXP_ID, 42, "RegExp");
+define_global_type!(REGEXP_EXEC_ID, 43, "RegExp.exec");
+
+/// Total number of predefined types.
+/// Must be one more than the highest TypeId above.
+pub const NUM_PREDEFINED_TYPES: usize = 44;
+
+// Resolved type ID constants (TypeId wrapped with GlobalLevel)
+pub const GLOBAL_UNKNOWN_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, UNKNOWN_ID);
+pub const GLOBAL_UNDEFINED_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, UNDEFINED_ID);
+pub const GLOBAL_VOID_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, VOID_ID);
+pub const GLOBAL_CONDITIONAL_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, CONDITIONAL_ID);
+pub const GLOBAL_NUMBER_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, NUMBER_ID);
+pub const GLOBAL_STRING_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, STRING_ID);
+pub const GLOBAL_ARRAY_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, ARRAY_ID);
+pub const GLOBAL_GLOBAL_ID /* :smirk: */: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, GLOBAL_ID);
+pub const GLOBAL_INSTANCEOF_PROMISE_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, INSTANCEOF_PROMISE_ID);
+pub const GLOBAL_PROMISE_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, PROMISE_ID);
+pub const GLOBAL_PROMISE_CONSTRUCTOR_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, PROMISE_CONSTRUCTOR_ID);
+pub const GLOBAL_BIGINT_STRING_LITERAL_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, BIGINT_STRING_LITERAL_ID);
+pub const GLOBAL_BOOLEAN_STRING_LITERAL_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, BOOLEAN_STRING_LITERAL_ID);
+pub const GLOBAL_FUNCTION_STRING_LITERAL_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, FUNCTION_STRING_LITERAL_ID);
+pub const GLOBAL_NUMBER_STRING_LITERAL_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, NUMBER_STRING_LITERAL_ID);
+pub const GLOBAL_OBJECT_STRING_LITERAL_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, OBJECT_STRING_LITERAL_ID);
+pub const GLOBAL_STRING_STRING_LITERAL_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, STRING_STRING_LITERAL_ID);
+pub const GLOBAL_SYMBOL_STRING_LITERAL_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, SYMBOL_STRING_LITERAL_ID);
+pub const GLOBAL_UNDEFINED_STRING_LITERAL_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, UNDEFINED_STRING_LITERAL_ID);
+pub const GLOBAL_TYPEOF_OPERATOR_RETURN_UNION_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, TYPEOF_OPERATOR_RETURN_UNION_ID);
+pub const GLOBAL_T_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, T_ID);
+pub const GLOBAL_U_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, U_ID);
+pub const GLOBAL_FETCH_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, FETCH_ID);
+pub const GLOBAL_INSTANCEOF_REGEXP_ID: ResolvedTypeId =
+    ResolvedTypeId::new(GLOBAL_LEVEL, INSTANCEOF_REGEXP_ID);
+pub const GLOBAL_REGEXP_ID: ResolvedTypeId = ResolvedTypeId::new(GLOBAL_LEVEL, REGEXP_ID);

crates/biome_js_type_info/src/globals_builder.rs

@@ -4,6 +4,8 @@ mod conditionals;
 mod flattening;
 mod format_type_info;
 mod globals;
+mod globals_builder;
+pub(crate) mod globals_ids;
 mod helpers;
 mod local_inference;
 mod resolver;
@@ -13,7 +15,8 @@ mod type_store;
 
 pub use conditionals::*;
 pub use flattening::MAX_FLATTEN_DEPTH;
-pub use globals::{GLOBAL_RESOLVER, GLOBAL_UNKNOWN_ID, GlobalsResolver, NUM_PREDEFINED_TYPES};
+pub use globals::{GLOBAL_RESOLVER, GlobalsResolver};
+pub use globals_ids::{GLOBAL_UNKNOWN_ID, NUM_PREDEFINED_TYPES};
 pub use resolver::*;
 pub use r#type::Type;
 pub use type_data::*;

crates/biome_js_type_info/src/globals_ids.rs

@@ -34,8 +34,12 @@ pub struct ResolvedTypeId(ResolverId, TypeId);
 impl Debug for ResolvedTypeId {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         if self.0.level() == TypeResolverLevel::Global {
-            if self.1.index() < NUM_PREDEFINED_TYPES {
-                f.write_str(global_type_name(self.1))
+            // GlobalsResolverBuilder makes sure the type store is fully filled.
+            // Every global TypeId whose index less than NUM_PREDEFINED_TYPES
+            // must have a name returned by global_type_name().
+            // GLOBAL_TYPE_MEMBERS ensures this invariant.
+            if let Some(name) = global_type_name(self.1) {
+                f.write_str(name)
             } else {
                 let id = self.1.index() - NUM_PREDEFINED_TYPES;
                 f.write_fmt(format_args!("Global TypeId({id})"))