Add parsing support for inline variable declaration

Variables can be defined in the body using the syntax
```st
{def} VAR y : DINT := 0; (*Pragma to avoid confusion with var blocks *)
(*Pragma is optional once in body*)
VAR x := 0; (*Implicit type declaration to DINT*)
FOR VAR x := 0 TO 10 BEGIN
(*Implicit variable declaration as loop counter*)
END_FOR
```

Ref: #973

Author: GitExample

compiler/plc_ast/src/ast.rs

@@ -619,6 +619,8 @@ pub enum AstStatement {
     ReturnStatement(ReturnStatement),
     JumpStatement(JumpStatement),
     LabelStatement(LabelStatement),
+    InlineVariable(InlineVariable),
+    DataTypeDeclaration(Box<DataTypeDeclaration>),
 }
 
 impl Debug for AstNode {
@@ -738,6 +740,12 @@ impl Debug for AstNode {
             AstStatement::LabelStatement(LabelStatement { name, .. }) => {
                 f.debug_struct("LabelStatement").field("name", name).finish()
             }
+            AstStatement::InlineVariable(InlineVariable { name, datatype }) => {
+                f.debug_struct("InlineVariable").field("name", name).field("datatype", datatype).finish()
+            }
+            AstStatement::DataTypeDeclaration(decl) => {
+                f.debug_tuple("DataTypeDeclaration").field(decl).finish()
+            }
         }
     }
 }
@@ -1512,6 +1520,27 @@ impl AstFactory {
     pub fn create_label_statement(name: String, location: SourceLocation, id: AstId) -> AstNode {
         AstNode { stmt: AstStatement::LabelStatement(LabelStatement { name }), location, id }
     }
+
+    /// Creates a new inline declaration by boxing the name and datatype
+    pub fn create_inline_declaration(
+        name: AstNode,
+        datatype: Option<AstNode>,
+        id: AstId,
+        location: SourceLocation,
+    ) -> AstNode {
+        let name = Box::new(name);
+        let datatype = datatype.map(Box::new);
+        AstNode { stmt: AstStatement::InlineVariable(InlineVariable { name, datatype }), id, location }
+    }
+
+    pub fn create_type_declaration(
+        datatype: DataTypeDeclaration,
+        id: AstId,
+        location: SourceLocation,
+    ) -> AstNode {
+        let datatype = Box::new(datatype);
+        AstNode { stmt: AstStatement::DataTypeDeclaration(datatype), id, location }
+    }
 }
 #[derive(Debug, Clone, PartialEq)]
 pub struct EmptyStatement {}
@@ -1594,3 +1623,10 @@ pub struct JumpStatement {
 pub struct LabelStatement {
     pub name: String,
 }
+
+/// Represents a new vaiable declaration in the body
+#[derive(Clone, Debug, PartialEq)]
+pub struct InlineVariable {
+    pub name: Box<AstNode>,
+    pub datatype: Option<Box<AstNode>>,
+}

src/lexer/tests/lexer_tests.rs

@@ -58,14 +58,16 @@ fn undefined_pragmas_are_ignored_by_the_lexer() {
 #[test]
 fn registered_pragmas_parsed() {
     let mut lexer = lex(r"
-        {external}{ref}{sized}{not_registerd}
+        {external}{ref}{sized}{def}{not_registerd}
         ");
     assert_eq!(lexer.token, PropertyExternal, "Token : {}", lexer.slice());
     lexer.advance();
     assert_eq!(lexer.token, PropertyByRef, "Token : {}", lexer.slice());
     lexer.advance();
     assert_eq!(lexer.token, PropertySized, "Token : {}", lexer.slice());
     lexer.advance();
+    assert_eq!(lexer.token, PropertyDef, "Token : {}", lexer.slice());
+    lexer.advance();
     assert_eq!(lexer.token, End);
 }
 

src/lexer/tokens.rs

@@ -22,6 +22,9 @@ pub enum Token {
     #[token("{sized}")]
     PropertySized,
 
+    #[token("{def}")]
+    PropertyDef,
+
     #[token("PROGRAM", ignore(case))]
     KeywordProgram,
 

src/parser.rs

@@ -338,7 +338,7 @@ fn parse_super_class(lexer: &mut ParseSession) -> Option<String> {
 fn parse_return_type(lexer: &mut ParseSession, pou_type: &PouType) -> Option<DataTypeDeclaration> {
     let start_return_type = lexer.range().start;
     if lexer.try_consume(&KeywordColon) {
-        if let Some((declaration, initializer)) = parse_data_type_definition(lexer, None) {
+        if let Some((declaration, initializer)) = parse_datatype_with_initializer(lexer, None) {
             if let Some(init) = initializer {
                 lexer.accept_diagnostic(Diagnostic::unexpected_initializer_on_function_return(
                     init.get_location(),
@@ -587,7 +587,7 @@ fn parse_full_data_type_definition(
                 None,
             ))
         } else {
-            parse_data_type_definition(lexer, name).map(|(type_def, initializer)| {
+            parse_datatype_with_initializer(lexer, name).map(|(type_def, initializer)| {
                 if lexer.try_consume(&KeywordDotDotDot) {
                     (
                         DataTypeDeclaration::DataTypeDefinition {
@@ -605,23 +605,29 @@ fn parse_full_data_type_definition(
     })
 }
 
-// TYPE xxx : 'STRUCT' | '(' | IDENTIFIER
-fn parse_data_type_definition(
+fn parse_datatype_with_initializer(
     lexer: &mut ParseSession,
     name: Option<String>,
 ) -> Option<DataTypeWithInitializer> {
+    parse_data_type_definition(lexer, name).map(|type_def| {
+        let initializer =
+            if lexer.try_consume(&KeywordAssignment) { Some(parse_expression(lexer)) } else { None };
+
+        (type_def, initializer)
+    })
+}
+
+// TYPE xxx : 'STRUCT' | '(' | IDENTIFIER
+fn parse_data_type_definition(lexer: &mut ParseSession, name: Option<String>) -> Option<DataTypeDeclaration> {
     let start = lexer.location();
     if lexer.try_consume(&KeywordStruct) {
         // Parse struct
         let variables = parse_variable_list(lexer);
-        Some((
-            DataTypeDeclaration::DataTypeDefinition {
-                data_type: DataType::StructType { name, variables },
-                location: start.span(&lexer.location()),
-                scope: lexer.scope.clone(),
-            },
-            None,
-        ))
+        Some(DataTypeDeclaration::DataTypeDefinition {
+            data_type: DataType::StructType { name, variables },
+            location: start.span(&lexer.location()),
+            scope: lexer.scope.clone(),
+        })
     } else if lexer.try_consume(&KeywordArray) {
         parse_array_type_definition(lexer, name)
     } else if lexer.try_consume(&KeywordPointer) {
@@ -661,23 +667,18 @@ fn parse_pointer_definition(
     lexer: &mut ParseSession,
     name: Option<String>,
     start_pos: usize,
-) -> Option<(DataTypeDeclaration, Option<AstNode>)> {
-    parse_data_type_definition(lexer, None).map(|(decl, initializer)| {
-        (
-            DataTypeDeclaration::DataTypeDefinition {
-                data_type: DataType::PointerType { name, referenced_type: Box::new(decl) },
-                location: lexer.source_range_factory.create_range(start_pos..lexer.last_range.end),
-                scope: lexer.scope.clone(),
-            },
-            initializer,
-        )
+) -> Option<DataTypeDeclaration> {
+    parse_data_type_definition(lexer, None).map(|decl| DataTypeDeclaration::DataTypeDefinition {
+        data_type: DataType::PointerType { name, referenced_type: Box::new(decl) },
+        location: lexer.source_range_factory.create_range(start_pos..lexer.last_range.end),
+        scope: lexer.scope.clone(),
     })
 }
 
 fn parse_type_reference_type_definition(
     lexer: &mut ParseSession,
     name: Option<String>,
-) -> Option<(DataTypeDeclaration, Option<AstNode>)> {
+) -> Option<DataTypeDeclaration> {
     let start = lexer.range().start;
     //Subrange
     let referenced_type = lexer.slice_and_advance();
@@ -692,9 +693,6 @@ fn parse_type_reference_type_definition(
         None
     };
 
-    let initial_value =
-        if lexer.try_consume(&KeywordAssignment) { Some(parse_expression(lexer)) } else { None };
-
     let end = lexer.last_range.end;
     if name.is_some() || bounds.is_some() {
         let data_type = match bounds {
@@ -732,15 +730,12 @@ fn parse_type_reference_type_definition(
                 scope: lexer.scope.clone(),
             },
         };
-        Some((data_type, initial_value))
+        Some(data_type)
     } else {
-        Some((
-            DataTypeDeclaration::DataTypeReference {
-                referenced_type,
-                location: lexer.source_range_factory.create_range(start..end),
-            },
-            initial_value,
-        ))
+        Some(DataTypeDeclaration::DataTypeReference {
+            referenced_type,
+            location: lexer.source_range_factory.create_range(start..end),
+        })
     }
 }
 
@@ -778,7 +773,7 @@ fn parse_string_size_expression(lexer: &mut ParseSession) -> Option<AstNode> {
 fn parse_string_type_definition(
     lexer: &mut ParseSession,
     name: Option<String>,
-) -> Option<(DataTypeDeclaration, Option<AstNode>)> {
+) -> Option<DataTypeDeclaration> {
     let text = lexer.slice().to_string();
     let start = lexer.range().start;
     let is_wide = lexer.token == KeywordWideString;
@@ -805,34 +800,26 @@ fn parse_string_type_definition(
         }),
         _ => Some(DataTypeDeclaration::DataTypeReference { referenced_type: text, location }),
     }
-    .zip(Some(lexer.try_consume(&KeywordAssignment).then(|| parse_expression(lexer))))
 }
 
-fn parse_enum_type_definition(
-    lexer: &mut ParseSession,
-    name: Option<String>,
-) -> Option<(DataTypeDeclaration, Option<AstNode>)> {
+fn parse_enum_type_definition(lexer: &mut ParseSession, name: Option<String>) -> Option<DataTypeDeclaration> {
     let start = lexer.last_location();
     let elements = parse_any_in_region(lexer, vec![KeywordParensClose], |lexer| {
         // Parse Enum - we expect at least one element
         let elements = parse_expression_list(lexer);
         Some(elements)
     })?;
-    let initializer = lexer.try_consume(&KeywordAssignment).then(|| parse_expression(lexer));
-    Some((
-        DataTypeDeclaration::DataTypeDefinition {
-            data_type: DataType::EnumType { name, elements, numeric_type: DINT_TYPE.to_string() },
-            location: start.span(&lexer.last_location()),
-            scope: lexer.scope.clone(),
-        },
-        initializer,
-    ))
+    Some(DataTypeDeclaration::DataTypeDefinition {
+        data_type: DataType::EnumType { name, elements, numeric_type: DINT_TYPE.to_string() },
+        location: start.span(&lexer.last_location()),
+        scope: lexer.scope.clone(),
+    })
 }
 
 fn parse_array_type_definition(
     lexer: &mut ParseSession,
     name: Option<String>,
-) -> Option<(DataTypeDeclaration, Option<AstNode>)> {
+) -> Option<DataTypeDeclaration> {
     let start = lexer.last_range.start;
     let range = parse_any_in_region(lexer, vec![KeywordOf], |lexer| {
         // Parse Array range
@@ -849,7 +836,7 @@ fn parse_array_type_definition(
     })?;
 
     let inner_type_defintion = parse_data_type_definition(lexer, None);
-    inner_type_defintion.map(|(reference, initializer)| {
+    inner_type_defintion.map(|reference| {
         let reference_end = reference.get_location().to_range().map(|it| it.end).unwrap_or(0);
         let location = lexer.source_range_factory.create_range(start..reference_end);
 
@@ -876,19 +863,16 @@ fn parse_array_type_definition(
             }
         };
 
-        (
-            DataTypeDeclaration::DataTypeDefinition {
-                data_type: DataType::ArrayType {
-                    name,
-                    bounds: range,
-                    referenced_type: Box::new(reference),
-                    is_variable_length,
-                },
-                location,
-                scope: lexer.scope.clone(),
+        DataTypeDeclaration::DataTypeDefinition {
+            data_type: DataType::ArrayType {
+                name,
+                bounds: range,
+                referenced_type: Box::new(reference),
+                is_variable_length,
             },
-            initializer,
-        )
+            location,
+            scope: lexer.scope.clone(),
+        }
     })
 }
 

src/parser/expressions_parser.rs

@@ -15,7 +15,7 @@ use plc_source::source_location::SourceLocation;
 use regex::{Captures, Regex};
 use std::{ops::Range, str::FromStr};
 
-use super::parse_hardware_access;
+use super::{parse_data_type_definition, parse_hardware_access};
 
 macro_rules! parse_left_associative_expression {
     ($lexer: expr, $action : expr,
@@ -281,6 +281,12 @@ fn parse_atomic_leaf_expression(lexer: &mut ParseSession<'_>) -> Result<AstNode,
         LiteralNull => parse_null_literal(lexer),
         KeywordSquareParensOpen => parse_array_literal(lexer),
         DirectAccess(access) => parse_direct_access(lexer, access),
+        PropertyDef => {
+            //Just consume the {def} and go further, if it's a variable we parse it next
+            lexer.advance();
+            parse_atomic_leaf_expression(lexer)
+        }
+        KeywordVar => parse_inline_declaration(lexer),
         _ => {
             if lexer.closing_keywords.contains(&vec![KeywordParensClose])
                 && matches!(lexer.last_token, KeywordOutputAssignment | KeywordAssignment)
@@ -296,6 +302,27 @@ fn parse_atomic_leaf_expression(lexer: &mut ParseSession<'_>) -> Result<AstNode,
     }
 }
 
+fn parse_inline_declaration(lexer: &mut ParseSession<'_>) -> Result<AstNode, Diagnostic> {
+    //Consume the direct access
+    let location = lexer.location();
+    //Inline variable declaration
+    lexer.advance();
+    //Parse the name
+    let name = parse_identifier(lexer);
+    let datatype = if lexer.try_consume(&KeywordColon) {
+        //Parse datatype
+        let type_location = lexer.location();
+        parse_data_type_definition(lexer, None).map(|it| {
+            AstFactory::create_type_declaration(it, lexer.next_id(), type_location.span(&lexer.location()))
+        })
+    } else {
+        None
+    };
+    let location = location.span(&lexer.last_location());
+
+    Ok(AstFactory::create_inline_declaration(name, datatype, lexer.next_id(), location))
+}
+
 fn parse_identifier(lexer: &mut ParseSession<'_>) -> AstNode {
     AstFactory::create_identifier(&lexer.slice_and_advance(), &lexer.last_location(), lexer.next_id())
 }

src/parser/tests.rs

@@ -11,6 +11,7 @@ mod control_parser_tests;
 mod expressions_parser_tests;
 mod function_parser_tests;
 mod initializer_parser_tests;
+mod inline_variable_tests;
 mod misc_parser_tests;
 mod parse_errors;
 mod parse_generics;