fix: correctly construct first and follow tables

umut-sahin · Nov 19, 2024 · 7bd9e90 · 7bd9e90
1 parent 78c098e
commit 7bd9e90
Show file tree

Hide file tree

Showing 4 changed files with 104 additions and 1 deletion.
diff --git a/assets/grammars/correct/indirect_empty.lr b/assets/grammars/correct/indirect_empty.lr
@@ -0,0 +1,7 @@
+S -> A B
+
+A -> C
+C -> D
+D -> ''
+
+B -> 'x'
diff --git a/src/tables.rs b/src/tables.rs
@@ -12,6 +12,7 @@ impl FirstTable {
     /// Constructs the first table from the grammar.
     pub fn construct(grammar: &Grammar) -> FirstTable {
         let mut first_table = IndexMap::new();
+        let mut indirectly_empty_symbols = IndexSet::new();
 
         let mut done = false;
         while !done {
@@ -32,7 +33,9 @@ impl FirstTable {
                                         .cloned(),
                                 );
                             }
-                            if !grammar.empty_symbols().contains(symbol) {
+                            if !grammar.empty_symbols().contains(symbol)
+                                && !indirectly_empty_symbols.contains(symbol)
+                            {
                                 break;
                             }
                         },
@@ -42,6 +45,7 @@ impl FirstTable {
                         },
                     }
                     if index == rule.pattern().len() - 1 {
+                        indirectly_empty_symbols.insert(rule.symbol());
                         possible_first_tokens.insert(Token::Empty);
                     }
                 }

diff --git a/tests/common.rs b/tests/common.rs
@@ -12,6 +12,7 @@ pub mod grammars {
     pub const G9: &str = include_str!("../assets/grammars/correct/g9.lr");
     pub const G10: &str = include_str!("../assets/grammars/correct/g10.lr");
     pub const G11: &str = include_str!("../assets/grammars/correct/g11.lr");
+    pub const INDIRECT_EMPTY: &str = include_str!("../assets/grammars/correct/indirect_empty.lr");
     pub const JSON: &str = include_str!("../assets/grammars/correct/json.lr");
     pub const NOT_LALR: &str = include_str!("../assets/grammars/correct/not-lalr.lr");
     pub const OPTIONAL: &str = include_str!("../assets/grammars/correct/optional.lr");

diff --git a/tests/parser.rs b/tests/parser.rs
@@ -144,6 +144,97 @@ fn raising_correct_error_when_creating_lalr_parser_for_non_lalr_grammar() {
 }
 
 
+#[test]
+#[cfg_attr(target_family = "wasm", wasm_bindgen_test)]
+fn correctly_creating_first_and_follow_sets_for_indirectly_empty_grammar() {
+    let grammar = Grammar::parse(common::grammars::INDIRECT_EMPTY).unwrap();
+    let parser = Parser::lr(grammar).unwrap();
+
+    let first_table = parser.first_table();
+    {
+        // +--------+-----------+
+        // | Symbol | First Set |
+        // +--------+-----------+
+        // | C      | { ε }     |
+        // | D      | { ε }     |
+        // | B      | { 'x' }   |
+        // | A      | { ε }     |
+        // | S      | { 'x' }   |
+        // +--------+-----------+
+
+        #[rustfmt::skip]
+    assert_eq!(
+        *first_table.deref(),
+        [
+            (
+                Symbol::from("C"),
+                [Token::Empty.into()].into(),
+            ),
+            (
+                Symbol::from("D"),
+                [Token::Empty.into()].into(),
+            ),
+            (
+                Symbol::from("B"),
+                [ConstantToken::from("x").into()].into(),
+            ),
+            (
+                Symbol::from("A"),
+                [Token::Empty.into()].into(),
+            ),
+            (
+                Symbol::from("S"),
+                [ConstantToken::from("x").into()].into(),
+            )
+        ]
+            .into_iter()
+            .collect::<IndexMap<_, _>>()
+    );
+    }
+
+    let follow_table = parser.follow_table();
+    {
+        // +--------+------------+
+        // | Symbol | Follow Set |
+        // +--------+------------+
+        // | C      | { 'x' }    |
+        // | D      | { 'x' }    |
+        // | B      | { $ }      |
+        // | A      | { 'x' }    |
+        // | S      | { $ }      |
+        // +--------+------------+
+
+        #[rustfmt::skip]
+    assert_eq!(
+        *follow_table.deref(),
+        [
+            (
+                Symbol::from("C"),
+                [ConstantToken::from("x").into()].into(),
+            ),
+            (
+                Symbol::from("D"),
+                [ConstantToken::from("x").into()].into(),
+            ),
+            (
+                Symbol::from("B"),
+                [Token::Eof].into(),
+            ),
+            (
+                Symbol::from("A"),
+                [ConstantToken::from("x").into()].into(),
+            ),
+            (
+                Symbol::from("S"),
+                [Token::Eof].into(),
+            ),
+        ]
+            .into_iter()
+            .collect::<IndexMap<_, _>>()
+    );
+    }
+}
+
 #[test]
 #[cfg_attr(target_family = "wasm", wasm_bindgen_test)]
 fn correctly_creating_lr_parser_for_binary_addition_grammar() {