baml-language: set up ppir

Author: sxlijin

baml_language/Cargo.lock

@@ -33,6 +33,7 @@ baml_codegen_types = { path = "crates/baml_codegen_types" }
 baml_compiler_analysis = { path = "crates/baml_compiler_analysis" }
 baml_compiler2_ast = { path = "crates/baml_compiler2_ast" }
 baml_compiler2_hir = { path = "crates/baml_compiler2_hir" }
+baml_compiler2_ppir = { path = "crates/baml_compiler2_ppir" }
 baml_compiler2_tir = { path = "crates/baml_compiler2_tir" }
 baml_compiler2_mir = { path = "crates/baml_compiler2_mir" }
 baml_compiler2_emit = { path = "crates/baml_compiler2_emit" }

baml_language/Cargo.toml

@@ -29,7 +29,7 @@ use crate::{
     semantic_index::{DefinitionSite, FileSemanticIndex, ScopeBindings, SemanticIndexExtra},
 };
 
-pub(crate) struct SemanticIndexBuilder<'db> {
+pub struct SemanticIndexBuilder<'db> {
     db: &'db dyn crate::Db,
     file: SourceFile,
 
@@ -51,7 +51,7 @@ pub(crate) struct SemanticIndexBuilder<'db> {
 }
 
 impl<'db> SemanticIndexBuilder<'db> {
-    pub(crate) fn new(db: &'db dyn crate::Db, file: SourceFile) -> Self {
+    pub fn new(db: &'db dyn crate::Db, file: SourceFile) -> Self {
         Self {
             db,
             file,
@@ -71,7 +71,7 @@ impl<'db> SemanticIndexBuilder<'db> {
     ///
     /// `file_range` is the full text range of the file (used for
     /// Project/Package/Namespace/File scopes).
-    pub(crate) fn build(
+    pub fn build(
         mut self,
         items: Vec<ast::Item>,
         file_range: TextRange,

baml_language/crates/baml_compiler2_hir/src/builder.rs

@@ -13,7 +13,7 @@
 //! - Cross-file aggregation: `namespace_items`, `package_items`
 
 pub mod body;
-mod builder;
+pub mod builder;
 pub mod contributions;
 pub mod diagnostic;
 pub mod file_package;
@@ -30,8 +30,9 @@ use std::sync::Arc;
 
 use baml_base::SourceFile;
 
+pub use builder::SemanticIndexBuilder;
+
 use crate::{
-    builder::SemanticIndexBuilder,
     contributions::FileSymbolContributions,
     item_tree::ItemTree,
     semantic_index::{FileSemanticIndex, ScopeBindings},

baml_language/crates/baml_compiler2_hir/src/lib.rs

@@ -0,0 +1,18 @@
+[package]
+name = "baml_compiler2_ppir"
+version.workspace = true
+edition.workspace = true
+
+[lints]
+workspace = true
+
+[dependencies]
+baml_base = { workspace = true }
+baml_compiler_parser = { workspace = true }
+baml_compiler2_ast = { workspace = true }
+baml_compiler2_hir = { workspace = true }
+baml_workspace = { workspace = true }
+rustc-hash = { workspace = true }
+salsa = { workspace = true }
+smol_str = { workspace = true }
+text-size = { workspace = true }

baml_language/crates/baml_compiler2_ppir/Cargo.toml

@@ -0,0 +1,70 @@
+//! Pre-Processed Intermediate Representation (PPIR) for compiler2.
+//!
+//! Pipeline: CST -> AST -> HIR (raw) -> PPIR (expansion + canonical) -> TIR.
+//!
+//! PPIR uses HIR's `package_items` for symbol classification (class vs enum vs alias),
+//! then synthesizes `stream_*` AST items and provides canonical queries that include
+//! both original and synthetic items.
+//!
+//! **No union simplification in PPIR.** Deferred to TIR.
+//!
+//! Currently a passthrough scaffold — all queries delegate directly to HIR.
+
+use std::sync::Arc;
+
+use baml_base::SourceFile;
+use baml_compiler2_hir::{
+    contributions::FileSymbolContributions,
+    item_tree::ItemTree,
+    namespace::NamespaceId,
+    package::{PackageId, PackageItems},
+    scope::ScopeId,
+    semantic_index::{FileSemanticIndex, ScopeBindings},
+};
+
+// -- Db trait -----------------------------------------------------------------
+
+#[salsa::db]
+pub trait Db: baml_compiler2_hir::Db {}
+
+// -- Canonical queries (passthrough to HIR) -----------------------------------
+
+/// Canonical file_semantic_index — currently delegates to HIR.
+/// Will later merge synthetic stream_* items into the index.
+pub fn file_semantic_index<'db>(
+    db: &'db dyn Db,
+    file: SourceFile,
+) -> &'db FileSemanticIndex<'db> {
+    baml_compiler2_hir::file_semantic_index(db, file)
+}
+
+/// Canonical file_item_tree — currently delegates to HIR.
+pub fn file_item_tree(db: &dyn Db, file: SourceFile) -> Arc<ItemTree> {
+    baml_compiler2_hir::file_item_tree(db, file)
+}
+
+/// Canonical symbol contributions — currently delegates to HIR.
+pub fn file_symbol_contributions<'db>(
+    db: &'db dyn Db,
+    file: SourceFile,
+) -> Arc<FileSymbolContributions<'db>> {
+    baml_compiler2_hir::file_symbol_contributions(db, file)
+}
+
+/// Canonical scope_bindings_query — currently delegates to HIR.
+pub fn scope_bindings_query<'db>(db: &'db dyn Db, scope_id: ScopeId<'db>) -> ScopeBindings {
+    baml_compiler2_hir::scope_bindings_query(db, scope_id)
+}
+
+/// Canonical namespace_items — currently delegates to HIR.
+pub fn namespace_items<'db>(
+    db: &'db dyn Db,
+    namespace_id: NamespaceId<'db>,
+) -> &'db baml_compiler2_hir::namespace::NamespaceItems<'db> {
+    baml_compiler2_hir::namespace::namespace_items(db, namespace_id)
+}
+
+/// Canonical package_items — currently delegates to HIR.
+pub fn package_items<'db>(db: &'db dyn Db, package_id: PackageId<'db>) -> &'db PackageItems<'db> {
+    baml_compiler2_hir::package::package_items(db, package_id)
+}

baml_language/crates/baml_compiler2_ppir/src/lib.rs

@@ -10,6 +10,7 @@ workspace = true
 baml_base = { workspace = true }
 baml_compiler2_ast = { workspace = true }
 baml_compiler2_hir = { workspace = true }
+baml_compiler2_ppir = { workspace = true }
 baml_compiler_analysis = { workspace = true }
 baml_compiler_diagnostics = { workspace = true }
 baml_workspace = { workspace = true }

baml_language/crates/baml_compiler2_tir/Cargo.toml

@@ -2023,7 +2023,7 @@ impl<'db> TypeInferenceBuilder<'db> {
             first.clone()
         };
         let pkg_id = baml_compiler2_hir::package::PackageId::new(db, pkg_name);
-        let pkg_items = baml_compiler2_hir::package::package_items(db, pkg_id);
+        let pkg_items = baml_compiler2_ppir::package_items(db, pkg_id);
 
         if pkg_items.namespaces.is_empty() {
             return Ty::Unknown;
@@ -2247,7 +2247,7 @@ impl<'db> TypeInferenceBuilder<'db> {
                     let db = self.context.db();
                     let pkg_info =
                         baml_compiler2_hir::file_package::file_package(db, class_loc.file(db));
-                    let item_tree = baml_compiler2_hir::file_item_tree(db, class_loc.file(db));
+                    let item_tree = baml_compiler2_ppir::file_item_tree(db, class_loc.file(db));
                     let class_data = &item_tree[class_loc.id(db)];
                     self.resolutions.insert(
                         at,
@@ -2504,7 +2504,7 @@ impl<'db> TypeInferenceBuilder<'db> {
                 let ns_context =
                     baml_compiler2_hir::file_package::file_package(self.context.db(), file)
                         .namespace_path;
-                let item_tree = baml_compiler2_hir::file_item_tree(self.context.db(), file);
+                let item_tree = baml_compiler2_ppir::file_item_tree(self.context.db(), file);
                 let class_data = &item_tree[class_loc.id(self.context.db())];
                 for field in &class_data.fields {
                     let mut diags = Vec::new();
@@ -2547,7 +2547,7 @@ impl<'db> TypeInferenceBuilder<'db> {
             self.package_items
         } else {
             let foreign_pkg_id = baml_compiler2_hir::package::PackageId::new(db, class_pkg.clone());
-            baml_compiler2_hir::package::package_items(db, foreign_pkg_id)
+            baml_compiler2_ppir::package_items(db, foreign_pkg_id)
         }
     }
 
@@ -2575,7 +2575,7 @@ impl<'db> TypeInferenceBuilder<'db> {
         let db = self.context.db();
         let file = class_loc.file(db);
         let ns_context = baml_compiler2_hir::file_package::file_package(db, file).namespace_path;
-        let item_tree = baml_compiler2_hir::file_item_tree(db, file);
+        let item_tree = baml_compiler2_ppir::file_item_tree(db, file);
         let class_data = &item_tree[class_loc.id(db)];
 
         for &method_id in &class_data.methods {
@@ -2700,7 +2700,7 @@ impl<'db> TypeInferenceBuilder<'db> {
             first.clone()
         };
         let pkg_id = baml_compiler2_hir::package::PackageId::new(db, resolved_pkg_name);
-        let pkg_items = baml_compiler2_hir::package::package_items(db, pkg_id);
+        let pkg_items = baml_compiler2_ppir::package_items(db, pkg_id);
 
         // Check that this package actually has items (non-empty = real package)
         if pkg_items.namespaces.is_empty() {
@@ -2840,7 +2840,7 @@ impl<'db> TypeInferenceBuilder<'db> {
                 let db = self.context.db();
                 let pkg_info =
                     baml_compiler2_hir::file_package::file_package(db, class_loc.file(db));
-                let item_tree = baml_compiler2_hir::file_item_tree(db, class_loc.file(db));
+                let item_tree = baml_compiler2_ppir::file_item_tree(db, class_loc.file(db));
                 let class_data = &item_tree[class_loc.id(db)];
                 self.resolutions.insert(
                     at,
@@ -2868,7 +2868,7 @@ impl<'db> TypeInferenceBuilder<'db> {
         let db = self.context.db();
         let baml_pkg_id =
             baml_compiler2_hir::package::PackageId::new(db, baml_base::Name::new("baml"));
-        let baml_items = baml_compiler2_hir::package::package_items(db, baml_pkg_id);
+        let baml_items = baml_compiler2_ppir::package_items(db, baml_pkg_id);
 
         // Look up the class by path (e.g. &["Array"] or &["media", "Image"]).
         let path: Vec<Name> = class_path.iter().map(|s| baml_base::Name::new(s)).collect();
@@ -2880,7 +2880,7 @@ impl<'db> TypeInferenceBuilder<'db> {
         let file = class_loc.file(db);
         let stub_pkg = baml_compiler2_hir::file_package::file_package(db, file);
         let stub_ns: &[Name] = &stub_pkg.namespace_path;
-        let item_tree = baml_compiler2_hir::file_item_tree(db, file);
+        let item_tree = baml_compiler2_ppir::file_item_tree(db, file);
         let class_data = &item_tree[class_loc.id(db)];
 
         // Bind generic type variables: e.g. {T → int} for Array<int>.
@@ -3031,7 +3031,7 @@ impl<'db> TypeInferenceBuilder<'db> {
             // Cross-package enum: look it up in the enum's own package.
             let pkg_id =
                 baml_compiler2_hir::package::PackageId::new(db, enum_name.package().clone());
-            Some(baml_compiler2_hir::package::package_items(db, pkg_id))
+            Some(baml_compiler2_ppir::package_items(db, pkg_id))
         };
 
         // Build the lookup path: namespace segments + enum name.
@@ -3042,7 +3042,7 @@ impl<'db> TypeInferenceBuilder<'db> {
         if let Some(def) = items.lookup_type(&lookup_path) {
             if let Definition::Enum(enum_loc) = def {
                 let file = enum_loc.file(db);
-                let item_tree = baml_compiler2_hir::file_item_tree(db, file);
+                let item_tree = baml_compiler2_ppir::file_item_tree(db, file);
                 let enum_data = &item_tree[enum_loc.id(db)];
                 return enum_data.variants.iter().map(|v| v.name.clone()).collect();
             }

baml_language/crates/baml_compiler2_tir/src/builder.rs

@@ -193,7 +193,7 @@ pub fn infer_scope_types<'db>(
 ) -> ScopeInference<'db> {
     let file = scope_id.file(db);
     let file_scope = scope_id.file_scope_id(db);
-    let index = baml_compiler2_hir::file_semantic_index(db, file);
+    let index = baml_compiler2_ppir::file_semantic_index(db, file);
     let scope = &index.scopes[file_scope.index() as usize];
 
     // Get package items for cross-file resolution
@@ -211,7 +211,7 @@ pub fn infer_scope_types<'db>(
             // Find the function by matching scope range AND name against item_tree functions.
             // Both checks are required to disambiguate companion functions that
             // share the parent's span.
-            let item_tree = baml_compiler2_hir::file_item_tree(db, file);
+            let item_tree = baml_compiler2_ppir::file_item_tree(db, file);
             let mut found = false;
             for (local_id, func_data) in &item_tree.functions {
                 if func_data.span == scope.range && scope.name.as_ref() == Some(&func_data.name) {
@@ -360,7 +360,7 @@ pub fn infer_scope_types<'db>(
         ScopeKind::Let => {
             // Top-level let binding — find the matching let in the item tree
             // and type-infer its initializer expression.
-            let item_tree = baml_compiler2_hir::file_item_tree(db, file);
+            let item_tree = baml_compiler2_ppir::file_item_tree(db, file);
             for (local_id, let_data) in &item_tree.lets {
                 if let_data.span == scope.range && scope.name.as_ref() == Some(&let_data.name) {
                     let let_loc = LetLoc::new(db, file, *local_id);
@@ -532,7 +532,7 @@ pub fn resolve_class_fields<'db>(
     class_loc: ClassLoc<'db>,
 ) -> Arc<ResolvedClassFields> {
     let file = class_loc.file(db);
-    let item_tree = baml_compiler2_hir::file_item_tree(db, file);
+    let item_tree = baml_compiler2_ppir::file_item_tree(db, file);
     let pkg_info = baml_compiler2_hir::file_package::file_package(db, file);
     let pkg_id = PackageId::new(db, pkg_info.package.clone());
     let pkg_items = package_items(db, pkg_id);
@@ -581,7 +581,7 @@ pub fn resolve_type_alias<'db>(
     alias_loc: TypeAliasLoc<'db>,
 ) -> Arc<ResolvedTypeAlias> {
     let file = alias_loc.file(db);
-    let item_tree = baml_compiler2_hir::file_item_tree(db, file);
+    let item_tree = baml_compiler2_ppir::file_item_tree(db, file);
     let pkg_info = baml_compiler2_hir::file_package::file_package(db, file);
     let pkg_id = PackageId::new(db, pkg_info.package.clone());
     let pkg_items = package_items(db, pkg_id);
@@ -635,9 +635,9 @@ pub fn render_scope_diagnostics<'db>(
     // Find the source map by matching scope range against item_tree functions.
     let file = scope_id.file(db);
     let file_scope = scope_id.file_scope_id(db);
-    let index = baml_compiler2_hir::file_semantic_index(db, file);
+    let index = baml_compiler2_ppir::file_semantic_index(db, file);
     let scope = &index.scopes[file_scope.index() as usize];
-    let item_tree = baml_compiler2_hir::file_item_tree(db, file);
+    let item_tree = baml_compiler2_ppir::file_item_tree(db, file);
 
     let source_map = item_tree
         .functions
@@ -675,7 +675,7 @@ pub fn collect_file_diagnostics<'db>(
     db: &'db dyn crate::Db,
     file: baml_base::SourceFile,
 ) -> TypeCheckDiagnostics<'db> {
-    let index = baml_compiler2_hir::file_semantic_index(db, file);
+    let index = baml_compiler2_ppir::file_semantic_index(db, file);
     let mut all_diagnostics = TypeCheckDiagnostics::default();
 
     for scope_id in &index.scope_ids {

baml_language/crates/baml_compiler2_tir/src/inference.rs

@@ -35,4 +35,4 @@ pub mod ty;
 /// Extends `baml_compiler2_hir::Db`. Use `infer_scope_types` for type
 /// inference queries, `resolve_name_at` for name resolution.
 #[salsa::db]
-pub trait Db: baml_compiler2_hir::Db {}
+pub trait Db: baml_compiler2_ppir::Db {}