Parse val/var binding & implicit val binding

CHANGELOG.md

@@ -32,13 +32,13 @@
 
 ## v0.0.11
 
-- [ ] Parse binding of form `val Type variable = value`
+- [x] Parse binding of form `val Type variable = value`
-- [ ] Parse binding of form `Type variable = value`
+- [x] Parse binding of form `Type variable = value`
 - [ ] Infer datatype of `value` in the above for a simple expression
 - [ ] Ensure that the anotated datatype matches the datatype of `value` in the above
 - [ ] Infer datatype of a `val variable = value` in the AST: Use the infered datatype
-- [ ] Formally define the top level constructs
+- [x] Formally define the top level constructs
-- [ ] Parse bindings and function declarations as top level constructs
+- [x] Parse bindings and function declarations as top level constructs
 - [ ] Parse function declaration arguments (`Type id`)
 - [x] Parse function return datatype (`fun f() -> Type`)
 - [x] Return parsing to variables to var/val

src/syntax/binding.rs

@@ -4,29 +4,51 @@ use super::{expression, ParsingError, ParsingResult};
 use crate::error_handling::SyntaxError;
 use crate::lexic::token::{Token, TokenType};
 
+/*
+binding = val binding | var binding
+val binding = "val", datatype?, binding remainder
+            | datatype, binding remainder
+
+var binding = "var", datatype?, binding remainder
+
+binding remainder = identifier, "=", expression
+ */
 pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Binding> {
     let mut current_pos = pos;
 
-    // TODO: Detect if the binding starts with a datatype
-    // TODO: Revert to val/var
-
     /*
-     * let keyword
+     * val/var keyword
      */
-    let (is_mutable, binding_token, next_pos) = {
+    let (is_var, binding_token, next_pos) = 'token: {
-        match parse_token_type(tokens, current_pos, TokenType::VAL) {
+        // check for VAL
-            Ok((val_token, next_pos)) => (false, val_token, next_pos),
+        if let Ok((val_token, next_pos)) = parse_token_type(tokens, current_pos, TokenType::VAL) {
-            _ => {
+            break 'token (false, Some(val_token), next_pos);
-                // If VAL is not found, search for VAR
+        };
+
+        // check for VAR
         match parse_token_type(tokens, current_pos, TokenType::VAR) {
-                    Ok((var_token, next_pos)) => (true, var_token, next_pos),
+            Ok((var_token, next_pos)) => (true, Some(var_token), next_pos),
-                    _ => return Err(ParsingError::Unmatched),
+            // If a VAR is not found it is still possible that the binding is an implicit VAL
-                }
+            _ => (false, None, current_pos),
-            }
         }
     };
     current_pos = next_pos;
 
+    /*
+     * datatype
+     */
+    let (datatype, next_pos) = match parse_token_type(tokens, current_pos, TokenType::Datatype) {
+        Ok((t, next)) => (Some(t), next),
+        _ => (None, current_pos),
+    };
+    current_pos = next_pos;
+
+    // Here:
+    // If the binding is None and the datatype is None, then we didn't match a binding
+    if binding_token.is_none() && datatype.is_none() {
+        return Err(ParsingError::Unmatched);
+    }
+
     /*
      * identifier
      */
@@ -45,16 +67,30 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Bindin
             return Err(ParsingError::Err(error));
         }
         _ => {
-            // The parser didn't find an Identifier after VAL/VAR
+            // The parser didn't find an Identifier after VAL/VAR or the Datatype
+            match (binding_token, datatype) {
+                (Some(binding_token), _) => {
                     return Err(ParsingError::Err(SyntaxError {
                         reason: format!(
                             "There should be an identifier after a `{}` token",
-                    if is_mutable { "val" } else { "var" }
+                            if is_var { "val" } else { "var" }
                         ),
                         error_start: binding_token.position,
                         error_end: binding_token.get_end_position(),
                     }));
                 }
+                (None, Some(datatype_token)) => {
+                    return Err(ParsingError::Err(SyntaxError {
+                        reason: "There should be an identifier after the datatype".into(),
+                        error_start: datatype_token.position,
+                        error_end: datatype_token.get_end_position(),
+                    }));
+                }
+                _ => {
+                    panic!("Illegal parser state: binding_token and datatype are both None")
+                }
+            };
+        }
     };
     current_pos = next_pos;
 
@@ -82,6 +118,9 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Bindin
     };
     current_pos += 1;
 
+    /*
+     * Expression of the binding
+     */
     let (expression, next_pos) = match expression::try_parse(tokens, current_pos) {
         Ok((exp, next)) => (exp, next),
         _ => {
@@ -95,10 +134,10 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Bindin
     current_pos = next_pos;
 
     let binding = Binding {
-        datatype: None,
+        datatype,
         identifier: &identifier,
         expression,
-        is_mutable,
+        is_mutable: is_var,
     };
 
     Ok((binding, current_pos))
@@ -144,26 +183,51 @@ mod tests {
         assert_eq!("=", token.value);
     }
 
-    /*
     #[test]
-    fn should_parse_binding_with_datatype() {
+    fn should_parse_val_binding_with_datatype() {
-        let tokens = get_tokens(&String::from("Num val identifier = 20")).unwrap();
+        let tokens = get_tokens(&String::from("val Int identifier = 20")).unwrap();
-        let ParseResult::Ok(Binding::Val(binding), _) = try_parse(&tokens, 0) else {
+        let (binding, _) = try_parse(&tokens, 0).unwrap();
-            panic!()
-        };
 
-        assert_eq!(Some(String::from("Num")), binding.datatype);
+        assert!(!binding.is_mutable);
-        assert_eq!("identifier", format!("{}", binding.identifier));
+        assert_eq!("Int", binding.datatype.unwrap().value);
-
+        assert_eq!("identifier", binding.identifier.value);
-        let tokens = get_tokens(&String::from("Bool var identifier = 20")).unwrap();
+    }
-        let ParseResult::Ok(Binding::Var(binding), _) = try_parse(&tokens, 0) else {
+
-            panic!()
+    #[test]
-        };
+    fn should_parse_var_binding_with_datatype() {
-
+        let tokens = get_tokens(&String::from("var Int identifier = 20")).unwrap();
-        assert_eq!(Some(String::from("Bool")), binding.datatype);
+        let (binding, _) = try_parse(&tokens, 0).unwrap();
-        assert_eq!("identifier", format!("{}", binding.identifier));
+
+        assert!(binding.is_mutable);
+        assert!(binding.datatype.is_some());
+        assert_eq!("Int", binding.datatype.unwrap().value);
+        assert_eq!("identifier", binding.identifier.value);
+    }
+
+    #[test]
+    fn should_parse_implicit_val_binding() {
+        let tokens = get_tokens(&String::from("Int identifier = 20")).unwrap();
+        let (binding, _) = try_parse(&tokens, 0).unwrap();
+
+        assert!(!binding.is_mutable);
+        assert!(binding.datatype.is_some());
+        assert_eq!("Int", binding.datatype.unwrap().value);
+        assert_eq!("identifier", binding.identifier.value);
+    }
+
+    #[test]
+    fn should_return_error_on_implicit_val_binding() {
+        let tokens = get_tokens(&String::from("Int => 20")).unwrap();
+        let binding = try_parse(&tokens, 0);
+
+        match binding {
+            Err(ParsingError::Err(error)) => {
+                assert_eq!(4, error.error_start);
+                assert_eq!(6, error.error_end);
+            }
+            _ => panic!("Error expected"),
+        }
     }
-     */
 
     #[test]
     fn should_return_correct_error() {

src/syntax/functions/function_declaration.rs

@@ -69,7 +69,6 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Functi
     };
     current_pos = next_pos;
 
-
     // Try to parse a return type
     let (return_type, next_pos) = 'return_label: {
         let (arrow_op, next_pos) = match try_operator(tokens, current_pos, "->".into()) {
@@ -134,7 +133,6 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Functi
     ))
 }
 
-
 #[cfg(test)]
 mod tests {
     use crate::lexic::get_tokens;
@@ -355,7 +353,10 @@ mod tests {
         let (function_declaration, _) = try_parse(&tokens, 0).unwrap();
 
         assert_eq!(function_declaration.identifier.value, String::from("id"));
-        assert_eq!(function_declaration.return_type.unwrap().value, String::from("String"));
+        assert_eq!(
+            function_declaration.return_type.unwrap().value,
+            String::from("String")
+        );
     }
 
     #[test]

src/syntax/grammar.md

@@ -58,7 +58,13 @@ arguments list = "(", ")"
 ## Binding
 
 ```ebnf
-binding = ("val" | "var"), identifier, "=", expression
+binding = val binding | var binding
+val binding = "val", datatype?, binding remainder
+            | datatype, binding remainder
+
+var binding = "var", datatype?, binding remainder
+
+binding remainder = identifier, "=", expression
 ```
 
 

src/syntax/utils.rs

@@ -75,5 +75,3 @@ pub fn parse_token_type(
         None => Err(ParsingError::Unmatched),
     }
 }
-
-