Parse function return datatype

CHANGELOG.md

@@ -40,8 +40,8 @@
 - [ ] Formally define the top level constructs
 - [ ] Parse bindings and function declarations as top level constructs
 - [ ] Parse function declaration arguments (`Type id`)
-- [ ] Parse function return datatype (`fun f() -> Type`)
+- [x] Parse function return datatype (`fun f() -> Type`)
-- [ ] Return parsing to variables to var/val
+- [x] Return parsing to variables to var/val
 - [ ] Write tests
 
 

src/lexic/scanner/identifier.rs

@@ -4,8 +4,8 @@ use crate::lexic::{token::Token, utils, LexResult};
 /// Checks if a String is a keyword, and returns its TokenType
 fn str_is_keyword(s: &String) -> Option<TokenType> {
     match s.as_str() {
-        "let" => Some(TokenType::LET),
+        "val" => Some(TokenType::VAL),
-        "mut" => Some(TokenType::MUT),
+        "var" => Some(TokenType::VAR),
         "fun" => Some(TokenType::FUN),
         _ => None,
     }
@@ -141,23 +141,23 @@ mod tests {
     // Should scan keywords
     #[test]
     fn test_4() {
-        let input = str_to_vec("mut");
+        let input = str_to_vec("var");
         let start_pos = 0;
         if let LexResult::Some(token, next) = scan(*input.get(0).unwrap(), &input, start_pos) {
             assert_eq!(3, next);
-            assert_eq!(TokenType::MUT, token.token_type);
+            assert_eq!(TokenType::VAR, token.token_type);
-            assert_eq!("mut", token.value);
+            assert_eq!("var", token.value);
             assert_eq!(0, token.position);
         } else {
             panic!()
         }
 
-        let input = str_to_vec("let");
+        let input = str_to_vec("val");
         let start_pos = 0;
         if let LexResult::Some(token, next) = scan(*input.get(0).unwrap(), &input, start_pos) {
             assert_eq!(3, next);
-            assert_eq!(TokenType::LET, token.token_type);
+            assert_eq!(TokenType::VAL, token.token_type);
-            assert_eq!("let", token.value);
+            assert_eq!("val", token.value);
             assert_eq!(0, token.position);
         } else {
             panic!()

src/lexic/token.rs

@@ -16,13 +16,13 @@ pub enum TokenType {
     Comma,
     INDENT,
     DEDENT,
-    LET,
+    VAL,
-    MUT,
+    VAR,
     EOF,
     FUN,
 }
 
-#[derive(Debug, Clone)]
+#[derive(Debug, Clone, PartialEq)]
 pub struct Token {
     pub token_type: TokenType,
     // The token as a raw string

src/semantic/impls.rs

@@ -32,7 +32,7 @@ impl SemanticCheck for TopLevelDeclaration<'_> {
                         error_start: binding.identifier.position,
                         error_end: binding.identifier.get_end_position(),
                         reason: format!(
-                            "Duplicated function: A function with name {} was already defined",
+                            "Duplicated: A symbol with name {} was already defined",
                             binding_name
                         ),
                     };
@@ -72,7 +72,7 @@ impl SemanticCheck for TopLevelDeclaration<'_> {
                         error_start: function.identifier.position,
                         error_end: function.identifier.get_end_position(),
                         reason: format!(
-                            "Duplicated function: A function with name {} was already defined",
+                            "Duplicated: A symbol with name {} was already defined",
                             function_name
                         ),
                     };

src/syntax/ast/mod.rs

@@ -20,6 +20,7 @@ pub enum TopLevelDeclaration<'a> {
 #[derive(Debug)]
 pub struct FunctionDeclaration<'a> {
     pub identifier: &'a Token,
+    pub return_type: Option<&'a Token>,
     pub params_list: Box<ParamsList>,
     pub block: Box<Block<'a>>,
 }

src/syntax/binding.rs

@@ -14,16 +14,15 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Bindin
      * let keyword
      */
     let (is_mutable, binding_token, next_pos) = {
-        let let_token = parse_token_type(tokens, current_pos, TokenType::LET);
+        match parse_token_type(tokens, current_pos, TokenType::VAL) {
-        match let_token {
+            Ok((val_token, next_pos)) => (false, val_token, next_pos),
-            Ok((let_token, next_let)) => {
+            _ => {
-                let mut_token = parse_token_type(tokens, next_let, TokenType::MUT);
+                // If VAL is not found, search for VAR
-                match mut_token {
+                match parse_token_type(tokens, current_pos, TokenType::VAR) {
-                    Ok((_mut_token, next_mut)) => (true, let_token, next_mut),
+                    Ok((var_token, next_pos)) => (true, var_token, next_pos),
-                    _ => (false, let_token, next_let),
+                    _ => return Err(ParsingError::Unmatched),
                 }
             }
-            _ => return Err(ParsingError::Unmatched),
         }
     };
     current_pos = next_pos;
@@ -112,7 +111,7 @@ mod tests {
 
     #[test]
     fn should_parse_val_binding() {
-        let tokens = get_tokens(&String::from("let identifier = 20")).unwrap();
+        let tokens = get_tokens(&String::from("val identifier = 20")).unwrap();
         let Ok((binding, _)) = try_parse(&tokens, 0) else {
             panic!()
         };
@@ -122,11 +121,11 @@ mod tests {
 
     #[test]
     fn should_parse_val() {
-        let tokens = get_tokens(&String::from("let")).unwrap();
+        let tokens = get_tokens(&String::from("val")).unwrap();
-        let (token, _) = parse_token_type(&tokens, 0, TokenType::LET).unwrap();
+        let (token, _) = parse_token_type(&tokens, 0, TokenType::VAL).unwrap();
 
-        assert_eq!(TokenType::LET, token.token_type);
+        assert_eq!(TokenType::VAL, token.token_type);
-        assert_eq!("let", token.value);
+        assert_eq!("val", token.value);
     }
 
     #[test]
@@ -168,8 +167,8 @@ mod tests {
 
     #[test]
     fn should_return_correct_error() {
-        let tokens = get_tokens(&String::from("let")).unwrap();
+        let tokens = get_tokens(&String::from("val")).unwrap();
-        assert_eq!(TokenType::LET, tokens[0].token_type);
+        assert_eq!(TokenType::VAL, tokens[0].token_type);
         assert_eq!(0, tokens[0].position);
         let binding = try_parse(&tokens, 0);
 
@@ -184,8 +183,8 @@ mod tests {
 
     #[test]
     fn should_return_error_when_identifier_is_wrong() {
-        let tokens = get_tokens(&String::from("let 322")).unwrap();
+        let tokens = get_tokens(&String::from("val 322")).unwrap();
-        assert_eq!(TokenType::LET, tokens[0].token_type);
+        assert_eq!(TokenType::VAL, tokens[0].token_type);
         assert_eq!(0, tokens[0].position);
         let binding = try_parse(&tokens, 0);
 
@@ -197,7 +196,7 @@ mod tests {
             _ => panic!("Error expected"),
         }
 
-        let tokens = get_tokens(&String::from("let \"hello\"")).unwrap();
+        let tokens = get_tokens(&String::from("val \"hello\"")).unwrap();
         let binding = try_parse(&tokens, 0);
 
         match binding {
@@ -211,7 +210,7 @@ mod tests {
 
     #[test]
     fn should_return_error_when_equal_op_is_wrong() {
-        let tokens = get_tokens(&String::from("let id \"error\"")).unwrap();
+        let tokens = get_tokens(&String::from("val id \"error\"")).unwrap();
         let binding = try_parse(&tokens, 0);
 
         match binding {

src/syntax/functions/function_declaration.rs

@@ -1,7 +1,7 @@
 use crate::{
     error_handling::SyntaxError,
     lexic::token::{Token, TokenType},
-    syntax::{ParsingError, ParsingResult},
+    syntax::{utils::try_operator, ParsingError, ParsingResult},
 };
 
 use super::{
@@ -9,6 +9,13 @@ use super::{
     params_list::parse_params_list,
 };
 
+/*
+function declaration = "fun", identifier, params list, return type?, block;
+
+params list = "(", ")";
+
+return type = ;
+ */
 pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<FunctionDeclaration> {
     let mut current_pos = pos;
 
@@ -19,6 +26,7 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Functi
     };
     current_pos = next_pos;
 
+    // identifier
     let (identifier, next_pos) = match parse_token_type(tokens, current_pos, TokenType::Identifier)
     {
         Ok((id, next)) => (id, next),
@@ -40,19 +48,20 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Functi
     };
     current_pos = next_pos;
 
+    // Params list
     let (params_list, next_pos) = match parse_params_list(tokens, current_pos) {
         Ok((params, next_pos)) => (params, next_pos),
         Err(ParsingError::Err(err)) => return Err(ParsingError::Err(err)),
         Err(ParsingError::Mismatch(wrong_token)) => {
             return Err(ParsingError::Err(SyntaxError {
-                reason: String::from("Expected an opening paren afted the function identifier."),
+                reason: String::from("Expected an opening paren after the function identifier."),
                 error_start: wrong_token.position,
                 error_end: wrong_token.get_end_position(),
             }));
         }
         Err(ParsingError::Unmatched) => {
             return Err(ParsingError::Err(SyntaxError {
-                reason: String::from("Expected an opening paren afted the function identifier."),
+                reason: String::from("Expected an opening paren after the function identifier."),
                 error_start: identifier.position,
                 error_end: identifier.get_end_position(),
             }));
@@ -60,6 +69,37 @@ 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()) {
+            Ok((op, next)) => (op, next),
+            _ => break 'return_label (None, current_pos),
+        };
+
+        // At this point the '->' operator was matched, so we expect a datatype
+        match parse_token_type(tokens, next_pos, TokenType::Datatype) {
+            Ok((t, next)) => (Some(t), next),
+            Err(ParsingError::Err(err)) => return Err(ParsingError::Err(err)),
+            Err(ParsingError::Mismatch(wrong_token)) => {
+                return Err(ParsingError::Err(SyntaxError {
+                    reason: String::from("Expected a datatype after the arrow operator."),
+                    error_start: wrong_token.position,
+                    error_end: wrong_token.get_end_position(),
+                }));
+            }
+            Err(ParsingError::Unmatched) => {
+                return Err(ParsingError::Err(SyntaxError {
+                    reason: String::from("Expected a datatype after the arrow operator."),
+                    error_start: arrow_op.position,
+                    error_end: arrow_op.get_end_position(),
+                }));
+            }
+        }
+    };
+    current_pos = next_pos;
+
+    // Function body (block)
     let (block, next_pos) = match parse_block(tokens, current_pos) {
         Ok((block, next_pos)) => (block, next_pos),
         Err(ParsingError::Err(error)) => {
@@ -86,6 +126,7 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Functi
     Ok((
         FunctionDeclaration {
             identifier: &identifier,
+            return_type,
             params_list: Box::new(params_list),
             block: Box::new(block),
         },
@@ -93,6 +134,7 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> ParsingResult<Functi
     ))
 }
 
+
 #[cfg(test)]
 mod tests {
     use crate::lexic::get_tokens;
@@ -150,7 +192,7 @@ mod tests {
             Err(ParsingError::Err(err)) => {
                 assert_eq!(
                     err.reason,
-                    "Expected an opening paren afted the function identifier."
+                    "Expected an opening paren after the function identifier."
                 );
                 assert_eq!(err.error_start, 7);
                 assert_eq!(err.error_end, 8);
@@ -164,7 +206,7 @@ mod tests {
             Err(ParsingError::Err(err)) => {
                 assert_eq!(
                     err.reason,
-                    "Expected an opening paren afted the function identifier."
+                    "Expected an opening paren after the function identifier."
                 );
                 assert_eq!(err.error_start, 4);
                 assert_eq!(err.error_end, 6);
@@ -304,6 +346,46 @@ 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, None);
+    }
+
+    #[test]
+    fn should_parse_return_type() {
+        let tokens = get_tokens(&String::from("fun id() -> String {}")).unwrap();
+        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"));
+    }
+
+    #[test]
+    fn should_throw_error_on_return_type_1() {
+        let tokens = get_tokens(&String::from("fun id() -> {}")).unwrap();
+        let fun_decl = try_parse(&tokens, 0);
+
+        match fun_decl {
+            Err(ParsingError::Err(err)) => {
+                assert_eq!(err.reason, "Expected a datatype after the arrow operator.");
+                assert_eq!(err.error_start, 12);
+                assert_eq!(err.error_end, 13);
+            }
+            _ => panic!("Expected an error: {:?}", fun_decl),
+        }
+    }
+
+    #[test]
+    fn should_throw_error_on_return_type_2() {
+        let tokens = get_tokens(&String::from("fun id() -> ")).unwrap();
+        let fun_decl = try_parse(&tokens, 0);
+
+        match fun_decl {
+            Err(ParsingError::Err(err)) => {
+                assert_eq!(err.reason, "Expected a datatype after the arrow operator.");
+                assert_eq!(err.error_start, 9);
+                assert_eq!(err.error_end, 11);
+            }
+            _ => panic!("Expected an error: {:?}", fun_decl),
+        }
     }
 }
 

src/syntax/grammar.md

@@ -16,6 +16,7 @@ simple function calls, and then implement other features top down
 ```ebnf
 top level statement = expression
                     | function declaration
+                    | binding
 ```
 
 

src/syntax/mod.rs

@@ -81,6 +81,22 @@ fn next_construct<'a>(
         _ => {}
     }
 
+    // Try to parse a binding
+    match binding::try_parse(tokens, current_pos) {
+        Ok((binding, next_pos)) => return Ok((TopLevelDeclaration::Binding(binding), next_pos)),
+        Err(ParsingError::Err(err)) => return Err(ParsingError::Err(err)),
+        _ => {}
+    }
+
+    // Try to parse an expression
+    match expression::try_parse(tokens, current_pos) {
+        Ok((expression, next_pos)) => {
+            return Ok((TopLevelDeclaration::Expression(expression), next_pos))
+        }
+        Err(ParsingError::Err(err)) => return Err(ParsingError::Err(err)),
+        _ => {}
+    }
+
     // No top level construct was found, return unmatched
     Err(ParsingError::Unmatched)
 }

src/syntax/statement.rs

@@ -49,7 +49,7 @@ mod tests {
 
     #[test]
     fn should_parse_binding() {
-        let input = String::from("let identifier = 20");
+        let input = String::from("val identifier = 20");
         let tokens = crate::lexic::get_tokens(&input).unwrap();
         let statement = try_parse(&tokens, 0);
 

src/syntax/utils.rs

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