Use sml-file, warn on .sig/.fun not containing 1

crates/cm-syntax/Cargo.toml

@@ -16,3 +16,4 @@ text-size-util.workspace = true
 
 lex-util.path = "../lex-util"
 slash-var-path.path = "../slash-var-path"
+sml-file.path = "../sml-file"

crates/cm-syntax/src/lower.rs

@@ -15,7 +15,7 @@ pub(crate) fn get(root: ParseRoot) -> Result<CmFile> {
     };
     let kind = match cls {
       Some(class) => match class.val {
-        Class::Sml => PathKind::Sml,
+        Class::Sml(k) => PathKind::Sml(k),
         Class::Cm => PathKind::Cm,
         Class::Other(s) => {
           return Err(Error::new(ErrorKind::UnsupportedClass(path, s), class.range))

crates/cm-syntax/src/types.rs

@@ -121,7 +121,7 @@ pub enum CmFileKind {
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum PathKind {
   /// SML files.
-  Sml,
+  Sml(sml_file::Kind),
   /// CM files.
   Cm,
 }
@@ -220,7 +220,7 @@ impl Member {
 #[derive(Debug, Clone)]
 pub enum Class {
   /// SML files.
-  Sml,
+  Sml(sml_file::Kind),
   /// CM files.
   Cm,
   /// Some other class.
@@ -229,11 +229,8 @@ pub enum Class {
 
 impl Class {
   fn from_path(path: &Path) -> Option<Self> {
-    let ret = match path.extension()?.to_str()? {
-      "sig" | "sml" | "fun" => Self::Sml,
-      "cm" => Self::Cm,
-      _ => return None,
-    };
+    let ext = path.extension()?.to_str()?;
+    let ret = if ext == "cm" { Self::Cm } else { Self::Sml(ext.parse().ok()?) };
     Some(ret)
   }
 }
@@ -243,7 +240,7 @@ impl FromStr for Class {
 
   fn from_str(s: &str) -> Result<Self, Self::Err> {
     let ret = match s.to_ascii_lowercase().as_str() {
-      "sml" => Self::Sml,
+      "sml" => Self::Sml(sml_file::Kind::Sml),
       "cm" | "cmfile" => Self::Cm,
       s => Self::Other(s.to_owned()),
     };
@@ -254,7 +251,7 @@ impl FromStr for Class {
 impl fmt::Display for Class {
   fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
     match self {
-      Class::Sml => f.write_str("sml"),
+      Class::Sml(_) => f.write_str("sml"),
       Class::Cm => f.write_str("cm"),
       Class::Other(s) => f.write_str(s),
     }

crates/input/Cargo.toml

@@ -26,6 +26,7 @@ mlb-hir.path = "../mlb-hir"
 mlb-syntax.path = "../mlb-syntax"
 slash-var-path.path = "../slash-var-path"
 sml-file-syntax.path = "../sml-file-syntax"
+sml-file.path = "../sml-file"
 sml-fixity.path = "../sml-fixity"
 sml-namespace.path = "../sml-namespace"
 sml-path.path = "../sml-path"

crates/input/src/lower_cm.rs

@@ -66,7 +66,7 @@ struct CmFile {
   /// only optional so this can derive default.
   pos_db: Option<text_pos::PositionDb>,
   cm_paths: Vec<paths::PathId>,
-  sml_paths: FxHashSet<paths::PathId>,
+  sml_paths: FxHashSet<(paths::PathId, sml_file::Kind)>,
   exports: NameExports,
 }
 
@@ -125,7 +125,7 @@ fn get_one_cm_file<F>(
       }
     };
     match pp.val.kind() {
-      cm_syntax::PathKind::Sml => {
+      cm_syntax::PathKind::Sml(kind) => {
         let contents = match read_file(st.fs, source, path.as_path()) {
           Ok(x) => x,
           Err(e) => {
@@ -134,7 +134,7 @@ fn get_one_cm_file<F>(
           }
         };
         st.sources.insert(path_id, contents);
-        ret.sml_paths.insert(path_id);
+        ret.sml_paths.insert((path_id, kind));
       }
       cm_syntax::PathKind::Cm => {
         let cur = GroupPathToProcess { parent: cur_path_id, range: source.range, path: path_id };
@@ -179,7 +179,7 @@ fn get_one_cm_file<F>(
 struct ExportCx<'a> {
   group: &'a StartedGroup,
   cm_paths: &'a [paths::PathId],
-  sml_paths: &'a FxHashSet<paths::PathId>,
+  sml_paths: &'a FxHashSet<(paths::PathId, sml_file::Kind)>,
   cur_path_id: paths::PathId,
 }
 
@@ -273,7 +273,7 @@ fn get_all_sources<F>(st: &mut St<'_, F>, cx: ExportCx<'_>, range: TextRange, ac
 where
   F: paths::FileSystem,
 {
-  for path_id in cx.sml_paths {
+  for (path_id, _) in cx.sml_paths {
     let contents = st.sources.get(path_id).expect("sml file should be set").as_str();
     get_top_defs(contents, ac, range);
   }

crates/input/src/lower_mlb.rs

@@ -121,7 +121,7 @@ where
         }
       };
       let kind = match pp.val.kind() {
-        mlb_syntax::PathKind::Sml => {
+        mlb_syntax::PathKind::Sml(file_kind) => {
           let contents = match read_file(st.fs, source, path.as_path()) {
             Ok(x) => x,
             Err(e) => {
@@ -131,7 +131,7 @@ where
             }
           };
           st.sources.insert(path_id, contents);
-          mlb_hir::PathKind::Source
+          mlb_hir::PathKind::Source(file_kind)
         }
         mlb_syntax::PathKind::Mlb => {
           st.stack.push(GroupPathToProcess {

crates/input/src/topo.rs

@@ -24,7 +24,7 @@ fn bas_dec_paths(ac: &mut BTreeSet<PathId>, dec: &mlb_hir::BasDec) {
     mlb_hir::BasDec::Path(p, _) => {
       ac.insert(*p);
     }
-    mlb_hir::BasDec::SourcePathSet(paths) => ac.extend(paths.iter().copied()),
+    mlb_hir::BasDec::SourcePathSet(paths) => ac.extend(paths.iter().map(|&(x, _)| x)),
     mlb_hir::BasDec::Basis(_, exp) => bas_exp_paths(ac, exp),
     mlb_hir::BasDec::Local(local_dec, in_dec) => {
       bas_dec_paths(ac, local_dec);

crates/mlb-hir/Cargo.toml

@@ -15,4 +15,5 @@ paths.workspace = true
 str-util.workspace = true
 text-size-util.workspace = true
 
+sml-file.path = "../sml-file"
 sml-namespace.path = "../sml-namespace"

crates/mlb-hir/src/lib.rs

@@ -22,7 +22,7 @@ pub enum BasDec {
   /// A file path.
   Path(paths::PathId, PathKind),
   /// Used by CM only.
-  SourcePathSet(FxHashSet<paths::PathId>),
+  SourcePathSet(FxHashSet<(paths::PathId, sml_file::Kind)>),
   /// A sequence of declarations.
   Seq(Vec<BasDec>),
 }
@@ -58,7 +58,7 @@ pub enum BasExp {
 #[derive(Debug, Clone, Copy)]
 pub enum PathKind {
   /// An SML source path.
-  Source,
+  Source(sml_file::Kind),
   /// A group path, like MLB or CM.
   Group,
 }

crates/mlb-statics/Cargo.toml

@@ -22,6 +22,7 @@ config.path = "../config"
 mlb-hir.path = "../mlb-hir"
 sml-comment.path = "../sml-comment"
 sml-file-syntax.path = "../sml-file-syntax"
+sml-file.path = "../sml-file"
 sml-fixity.path = "../sml-fixity"
 sml-hir-lower.path = "../sml-hir-lower"
 sml-hir.path = "../sml-hir"

crates/mlb-statics/src/lib.rs

@@ -245,10 +245,10 @@ fn get_bas_dec(
       }
     },
     mlb_hir::BasDec::Path(path, kind) => match kind {
-      mlb_hir::PathKind::Source => {
+      mlb_hir::PathKind::Source(file_kind) => {
         let contents = cx.source_file_contents.get(path).expect("no source file");
         let mut fix_env = scope.fix_env.clone();
-        let syntax = SourceFileSyntax::new(&mut fix_env, cx.lang, contents);
+        let syntax = SourceFileSyntax::new(&mut fix_env, cx.lang, *file_kind, contents);
         get_source_file(st, cx.lang, *path, scope, ac, fix_env, syntax);
       }
       mlb_hir::PathKind::Group => match st.bases.get(path) {
@@ -259,11 +259,11 @@ fn get_bas_dec(
     mlb_hir::BasDec::SourcePathSet(paths) => {
       let mut syntaxes: paths::PathMap<_> = paths
         .iter()
-        .map(|path| {
-          let contents = cx.source_file_contents.get(path).expect("no source file");
+        .map(|&(path, kind)| {
+          let contents = cx.source_file_contents.get(&path).expect("no source file");
           let mut fix_env = scope.fix_env.clone();
-          let syntax = SourceFileSyntax::new(&mut fix_env, cx.lang, contents);
-          (*path, (fix_env, syntax))
+          let syntax = SourceFileSyntax::new(&mut fix_env, cx.lang, kind, contents);
+          (path, (fix_env, syntax))
         })
         .collect();
       let hir_roots: paths::PathMap<_> = syntaxes
@@ -386,7 +386,7 @@ pub fn update_one(
   contents: &str,
 ) {
   let mut fix_env = sml_fixity::STD_BASIS.clone();
-  sf.syntax = sml_file_syntax::SourceFileSyntax::new(&mut fix_env, lang, contents);
+  sf.syntax = sml_file_syntax::SourceFileSyntax::new(&mut fix_env, lang, sf.syntax.kind, contents);
   let mode = sml_statics_types::mode::Mode::Regular(Some(path));
   let checked =
     sml_statics::get(syms_tys, &sf.scope, mode, &sf.syntax.lower.arenas, &sf.syntax.lower.root);

crates/mlb-statics/src/std_basis.rs

@@ -97,7 +97,9 @@ where
       contents = &owned_contents;
     }
     let mut fix_env = sml_fixity::STD_BASIS.clone();
-    let started = SourceFileSyntax::new(&mut fix_env, &lang, contents);
+    let ext = std::path::Path::new(name).extension().unwrap().to_str().unwrap();
+    let file_kind: sml_file::Kind = ext.parse().unwrap();
+    let started = SourceFileSyntax::new(&mut fix_env, &lang, file_kind, contents);
     if let Some(e) = started.lex_errors.first() {
       panic!("{name}: lex error: {e}");
     }

crates/mlb-syntax/Cargo.toml

@@ -16,4 +16,5 @@ text-size-util.workspace = true
 
 lex-util.path = "../lex-util"
 slash-var-path.path = "../slash-var-path"
+sml-file.path = "../sml-file"
 sml-namespace.path = "../sml-namespace"

crates/mlb-syntax/src/parse.rs

@@ -171,11 +171,8 @@ fn bas_dec_one(p: &mut Parser<'_>) -> Result<BasDecOne> {
 }
 
 fn path_kind(path: &Path) -> Option<PathKind> {
-  let ret = match path.extension()?.to_str()? {
-    "sml" | "sig" | "fun" => PathKind::Sml,
-    "mlb" => PathKind::Mlb,
-    _ => return None,
-  };
+  let ext = path.extension()?.to_str()?;
+  let ret = if ext == "mlb" { PathKind::Mlb } else { PathKind::Sml(ext.parse().ok()?) };
   Some(ret)
 }
 

crates/mlb-syntax/src/types.rs

@@ -131,7 +131,7 @@ pub enum BasExp {
 #[derive(Debug, Clone, Copy)]
 pub enum PathKind {
   /// SML paths.
-  Sml,
+  Sml(sml_file::Kind),
   /// ML Basis paths.
   Mlb,
 }

crates/sml-dynamics-tests/Cargo.toml

@@ -15,6 +15,7 @@ str-util.workspace = true
 config.path = "../config"
 sml-dynamics.path = "../sml-dynamics"
 sml-file-syntax.path = "../sml-file-syntax"
+sml-file.path = "../sml-file"
 sml-fixity.path = "../sml-fixity"
 sml-hir.path = "../sml-hir"
 sml-path.path = "../sml-path"

crates/sml-dynamics-tests/src/lib.rs

@@ -19,7 +19,7 @@ fn check(s: &str, steps: &[&str]) {
   let check_steps = !env_var_enabled("MILLET_NO_CHECK_STEPS");
   let mut fix_env = sml_fixity::STD_BASIS.clone();
   let lang = config::lang::Language::default();
-  let sf = sml_file_syntax::SourceFileSyntax::new(&mut fix_env, &lang, s);
+  let sf = sml_file_syntax::SourceFileSyntax::new(&mut fix_env, &lang, sml_file::Kind::Sml, s);
   if let Some(e) = sf.lex_errors.first() {
     panic!("lex error: {e}");
   }

crates/sml-file-syntax/Cargo.toml

@@ -14,6 +14,7 @@ elapsed.workspace = true
 text-pos.workspace = true
 
 config.path = "../config"
+sml-file.path = "../sml-file"
 sml-fixity.path = "../sml-fixity"
 sml-hir-lower.path = "../sml-hir-lower"
 sml-lex.path = "../sml-lex"

crates/sml-file-syntax/src/lib.rs

@@ -13,16 +13,23 @@ pub struct SourceFileSyntax {
   pub parse: sml_parse::Parse,
   /// The lowered HIR.
   pub lower: sml_hir_lower::Lower,
+  /// The kind of source file this is.
+  pub kind: sml_file::Kind,
 }
 
 impl SourceFileSyntax {
   /// Starts processing a single source file.
-  pub fn new(fix_env: &mut sml_fixity::Env, lang: &config::lang::Language, contents: &str) -> Self {
+  pub fn new(
+    fix_env: &mut sml_fixity::Env,
+    lang: &config::lang::Language,
+    kind: sml_file::Kind,
+    contents: &str,
+  ) -> Self {
     elapsed::log("SourceFileSyntax::new", || {
       let (lex_errors, parse) = Self::lex_and_parse(fix_env, contents);
-      let mut lower = sml_hir_lower::get(lang, &parse.root);
+      let mut lower = sml_hir_lower::get(lang, kind, &parse.root);
       sml_ty_var_scope::get(&mut lower.arenas, &lower.root);
-      Self { pos_db: text_pos::PositionDb::new(contents), lex_errors, parse, lower }
+      Self { pos_db: text_pos::PositionDb::new(contents), lex_errors, parse, lower, kind }
     })
   }
 

crates/sml-file/Cargo.toml

@@ -0,0 +1,10 @@
+[package]
+name = "sml-file"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+publish.workspace = true
+
+[lib]
+doctest = false
+test = false