feat: grammar for typed bnf

Devin-Yeung · Nov 10, 2024 · 5bc2185 · 5bc2185
1 parent 9099355
commit 5bc2185
Show file tree

Hide file tree

Showing 7 changed files with 989 additions and 68 deletions.
diff --git a/src/grammar/mod.rs b/src/grammar/mod.rs
@@ -33,6 +33,15 @@ mod test {
         insta::assert_debug_snapshot!(grammar);
     }
 
+    #[test]
+    fn typed() {
+        let text = r#"
+            <E> ::= <E: "int"> "+" <E: "int"> ;
+        "#;
+        let grammar = RawGrammar::parse(text).unwrap();
+        insta::assert_debug_snapshot!(grammar);
+    }
+
     #[test]
     fn repeat() {
         let text = r#"
@@ -52,7 +61,7 @@ mod test {
 
     #[test]
     fn invalid_token() {
-        let text = ":";
+        let text = "*";
         let err = RawGrammar::parse(text).err().unwrap();
         let ui = report_with_unnamed_source(err, text);
         insta::assert_snapshot!(ui);

diff --git a/src/grammar/snapshots/bnfgen__grammar__test__invalid_token.snap b/src/grammar/snapshots/bnfgen__grammar__test__invalid_token.snap
@@ -4,7 +4,7 @@ expression: ui
 ---
   × Invalid token
    ╭────
- 1 │ :
+ 1 │ *
    · ┬
    · ╰── this token is invalid
    ╰────
diff --git a/src/grammar/snapshots/bnfgen__grammar__test__typed.snap b/src/grammar/snapshots/bnfgen__grammar__test__typed.snap
@@ -0,0 +1,56 @@
+---
+source: src/grammar/mod.rs
+expression: grammar
+---
+RawGrammar {
+    rules: [
+        Rule {
+            name: "E",
+            production: WeightedProduction {
+                alts: [
+                    Alternative {
+                        span: Span {
+                            start: 21,
+                            end: 46,
+                        },
+                        weight: 1,
+                        invoke_limit: Unlimited,
+                        symbols: [
+                            Symbol {
+                                kind: NonTerminal(
+                                    "E",
+                                ),
+                                span: Span {
+                                    start: 21,
+                                    end: 31,
+                                },
+                            },
+                            Symbol {
+                                kind: Terminal(
+                                    "+",
+                                ),
+                                span: Span {
+                                    start: 32,
+                                    end: 35,
+                                },
+                            },
+                            Symbol {
+                                kind: NonTerminal(
+                                    "E",
+                                ),
+                                span: Span {
+                                    start: 36,
+                                    end: 46,
+                                },
+                            },
+                        ],
+                    },
+                ],
+            },
+            span: Span {
+                start: 13,
+                end: 48,
+            },
+        },
+    ],
+}
diff --git a/src/grammar/snapshots/bnfgen__grammar__test__unexpected_eof.snap b/src/grammar/snapshots/bnfgen__grammar__test__unexpected_eof.snap
@@ -6,5 +6,5 @@ expression: ui
    ╭────
  1 │ <start> ::= "Hello" | "World"
    ·                             ┬
-   ·                             ╰── expect "{", "|", ";", "re", "str", "nonterminal"
+   ·                             ╰── expect "{", "|", ";", "<", "re", "str"
    ╰────
diff --git a/src/parser.lalrpop b/src/parser.lalrpop
@@ -26,10 +26,13 @@ extern {
     "|" => Token::Or,
     "," => Token::Comma,
     "::=" => Token::Def,
+    ":" => Token::Colon,
     ";" => Token::Semi,
+    "<" => Token::LAngle,
+    ">" => Token::RAngle,
     "re" => Token::Re,
     "str" => Token::Str(<String>),
-    "nonterminal" => Token::NonTerminal(<String>),
+    "id" => Token::Id(<String>),
     "int" => Token::Int(<usize>),
   }
 }
@@ -43,7 +46,7 @@ pub RawGrammar: RawGrammar = {
 };
 
 Rule: Rule = {
-    <l: @L> <name: "nonterminal"> "::=" <alts: Alternatives> ";" <r: @R> => {
+    <l: @L> "<" <name: "id"> ">" "::=" <alts: Alternatives> ";" <r: @R> => {
         Rule {
             name,
             production: WeightedProduction {
@@ -96,7 +99,14 @@ Symbol: Symbol = {
             span: Span::new(l, r),
         }
     },
-    <l: @L> <nt: "nonterminal"> <r: @R> => {
+    <l: @L> "<" <nt: "id"> ">" <r: @R> => {
+        Symbol {
+            kind: SymbolKind::NonTerminal(Rc::new(nt)),
+            span: Span::new(l, r),
+        }
+    },
+    // typed non-terminal
+    <l: @L> "<" <nt: "id"> ":" <_ty: "str"> ">" <r: @R> => {
         Symbol {
             kind: SymbolKind::NonTerminal(Rc::new(nt)),
             span: Span::new(l, r),