Scan datatypes. Parse datatype annotations

CHANGELOG.md

@@ -11,6 +11,11 @@
 - [ ] Stdlib
 - [ ] Document code
 
+
+## v0.0.4
+
+- Explicit datatype of variables
+
 ## v0.0.3
 
 - Get datatype of an identifier from the symbol table

src/ast_types.rs

@@ -9,11 +9,13 @@ pub enum Binding<'a> {
 }
 
 pub struct ValBinding<'a> {
+    pub datatype: Option<String>,
     pub identifier: &'a String,
     pub expression: Expression<'a>,
 }
 
 pub struct VarBinding<'a> {
+    pub datatype: Option<String>,
     pub identifier: &'a String,
     pub expression: Expression<'a>,
 }

src/codegen/binding.rs

@@ -31,6 +31,7 @@ mod tests {
         let id = String::from("identifier");
         let value = String::from("322");
         let binding = Binding::Val(ValBinding {
+            datatype: None,
             identifier: &id,
             expression: Expression::Number(&value),
         });

src/codegen/module_ast.rs

@@ -22,6 +22,7 @@ mod tests {
         let id = String::from("identifier");
         let value = String::from("322");
         let binding = Binding::Val(ValBinding {
+            datatype: None,
             identifier: &id,
             expression: Expression::Number(&value),
         });

src/lexic/mod.rs

@@ -75,6 +75,7 @@ fn next_token(chars: &Chars, current_pos: usize) -> LexResult {
     None
         .or_else(|| scanner::number(next_char, chars, current_pos))
         .or_else(|| scanner::identifier(next_char, chars, current_pos))
+        .or_else(|| scanner::datatype(next_char, chars, current_pos))
         .or_else(|| scanner::string(next_char, chars, current_pos))
         .or_else(|| scanner::operator(next_char, chars, current_pos))
         .or_else(|| scanner::grouping_sign(next_char, chars, current_pos))
@@ -203,6 +204,14 @@ mod tests {
         assert_eq!("]", t.value);
     }
     
+    #[test]
+    fn should_scan_datatype() {
+        let input = String::from("Num");
+        let tokens = get_tokens(&input).unwrap();
+
+        assert_eq!(TokenType::Datatype, tokens[0].token_type);
+    }
+    
     #[test]
     fn should_scan_new_line() {
         let input = String::from("3\n22");

src/lexic/scanner/identifier.rs

@@ -13,19 +13,32 @@ fn str_is_keyword(s: &String) -> Option<TokenType> {
 /// a valid identifier
 pub fn scan(start_char: char, chars: &Vec<char>, start_pos: usize) -> LexResult {
     // The scanning is done by this recursive function
-    scan_impl(chars, start_pos + 1, format!("{}", start_char))
+    scan_impl(
+        chars,
+        start_pos + 1,
+        format!("{}", start_char),
+        utils::is_uppercase(start_char),
+    )
 }
 
 /// Recursive funtion that scans the identifier
-fn scan_impl(chars: &Vec<char>, start_pos: usize, current: String) -> LexResult {
+fn scan_impl(chars: &Vec<char>, start_pos: usize, current: String, is_datatype: bool) -> LexResult {
     match chars.get(start_pos) {
         Some(c) if utils::is_identifier_char(*c) => {
-            scan_impl(chars, start_pos + 1, utils::str_append(current, *c))
+            scan_impl(
+                chars,
+                start_pos + 1,
+                utils::str_append(current, *c),
+                is_datatype,
+            )
         },
         _ => {
             if let Some(token_type) = str_is_keyword(&current) {
                 LexResult::Some(token::new(current, start_pos as i32, token_type), start_pos)
             }
+            else if is_datatype {
+                LexResult::Some(token::new_datatype(current, start_pos as i32), start_pos)
+            }
             else {
                 LexResult::Some(token::new_identifier(current, start_pos as i32), start_pos)
             }

src/lexic/scanner/mod.rs

@@ -48,6 +48,13 @@ pub fn identifier(c: char, chars: &Vec<char>, start_pos: usize) -> Option<LexRes
         .then(|| identifier::scan(c, chars, start_pos))
 }
 
+/// Attempts to scan a datatype. If not found returns None to be able to chain other scanner
+pub fn datatype(c: char, chars: &Vec<char>, start_pos: usize) -> Option<LexResult> {
+    // Since the only difference with an identifier is that the fist character is an
+    // uppercase letter, reuse the identifier scanner
+    utils::is_uppercase(c)
+        .then(|| identifier::scan(c, chars, start_pos))
+}
 
 /// Attempts to scan a string. If not found returns None to be able to chain other scanner
 pub fn string(c: char, chars: &Vec<char>, start_pos: usize) -> Option<LexResult> {

src/syntax/binding.rs

@@ -2,11 +2,26 @@ use crate::token::{Token, TokenType};
 use super::ast_types::{ValBinding, VarBinding, Binding};
 use super::expression;
 
-// Should return a 3 state value:
+// TODO: Should return a 3 state value:
 // - Success: binding parsed successfully
 // - NotFound: the first token (var | val) was not found, so the parser should try other options
 // - Error: token (var | val) was found, but then other expected tokens were not found
 pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> Option<Binding> {
+    let mut pos = pos;
+    
+    // Optional datatype annotation
+    let datatype_annotation = {
+        match tokens.get(pos) {
+            Some(t) if t.token_type == TokenType::Datatype => {
+                pos += 1;
+                Some(String::from(&t.value))
+            }
+            Some(_) => None,
+            None => return None
+        }
+    };
+    
+    // var/val keyword
     let is_val = {
         let res1 = try_token_type(tokens, pos, TokenType::VAL);
         match res1 {
@@ -27,7 +42,6 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> Option<Binding> {
 
     let equal_operator = try_operator(tokens, pos + 2, String::from("="));
     if equal_operator.is_none() { return None }
-    let _ = equal_operator.unwrap();
 
     let expression = expression::try_parse(tokens, pos + 3);
     if expression.is_none() { return None }
@@ -35,12 +49,14 @@ pub fn try_parse<'a>(tokens: &'a Vec<Token>, pos: usize) -> Option<Binding> {
 
     if is_val {
         Some(Binding::Val(ValBinding {
+            datatype: datatype_annotation,
             identifier: &identifier.value,
             expression,
         }))
     }
     else {
         Some(Binding::Var(VarBinding {
+            datatype: datatype_annotation,
             identifier: &identifier.value,
             expression,
         }))
@@ -80,9 +96,7 @@ mod tests {
             Binding::Val(binding) => {
                 assert_eq!("identifier", binding.identifier);
             }
-            Binding::Var(binding) => {
-                assert_eq!("identifier", binding.identifier);
-            }
+            _ => panic!()
         }
     }
 
@@ -110,4 +124,30 @@ mod tests {
 
         assert_eq!("=", token.value);
     }
+    
+    #[test]
+    fn should_parse_binding_with_datatype() {
+        let tokens = get_tokens(&String::from("Num val identifier = 20")).unwrap();
+        let binding = try_parse(&tokens, 0).unwrap();
+
+        match binding {
+            Binding::Val(binding) => {
+                assert_eq!(Some(String::from("Num")), binding.datatype);
+                assert_eq!("identifier", binding.identifier);
+            }
+            _ => panic!()
+        }
+        
+        
+        let tokens = get_tokens(&String::from("Bool var identifier = true")).unwrap();
+        let binding = try_parse(&tokens, 0).unwrap();
+
+        match binding {
+            Binding::Var(binding) => {
+                assert_eq!(Some(String::from("Bool")), binding.datatype);
+                assert_eq!("identifier", binding.identifier);
+            }
+            _ => panic!()
+        }
+    }
 }

src/syntax/grammar.md

@@ -11,13 +11,25 @@ A module is (commonly) a single source file.
 
 A declaration with `var` or `val`.
 
-- `var = "var"`
-- `val = "val"`
-- `variable_binding = (var | val), identifier, "=", expression`
+```ebnf
+var = "var"
+val = "val"
+variable_binding = (var | val), identifier, "=", expression
+```
+
 
 ### `expression`
 
-For now just a number
+For now just a number, string or boolean
 
-- `expression = number`
+```ebnf
+expression = number | string | boolean
+```
+
+
+## Type annotations
+
+```ebnf
+variable_binding = Datatype, (var | val), identifier, "=", expression
+```
 

src/syntax/mod.rs

@@ -1,14 +1,14 @@
 use super::token::Token;
 
 mod expression;
-mod val_binding;
+mod binding;
 use super::ast_types;
 
 use ast_types::ModuleAST;
 
 /// Constructs the Misti AST from a vector of tokens
 pub fn construct_ast<'a>(tokens: &'a Vec<Token>) -> Result<ModuleAST<'a>, String> {
-    let maybe_binding = val_binding::try_parse(tokens, 0);
+    let maybe_binding = binding::try_parse(tokens, 0);
 
     match maybe_binding {
         Some(binding) => {

src/token.rs

@@ -1,6 +1,7 @@
 #[derive(PartialEq, Debug, Clone)]
 pub enum TokenType {
     Identifier,
+    Datatype,
     Number,
     String,
     Operator,
@@ -76,3 +77,11 @@ pub fn new_semicolon(position: i32) -> Token {
         _position: position,
     }
 }
+
+pub fn new_datatype(value: String, position: i32) -> Token {
+    Token {
+        token_type: TokenType::Datatype,
+        value,
+        _position: position,
+    }
+}