set up grammar structure

grammar/MXLexer.g4

@@ -1,10 +1,22 @@
 lexer grammar MXLexer;
 
 // Keywords
-INT: 'int';
 VOID: 'void';
+BOOL: 'bool';
+INT: 'int';
+STRING: 'string';
+NEW: 'new';
+CLASS: 'class';
+NULL: 'null';
+TRUE: 'true';
+FALSE: 'false';
+THIS: 'this';
 IF: 'if';
 ELSE: 'else';
+FOR: 'for';
+WHILE: 'while';
+BREAK: 'break';
+CONTINUE: 'continue';
 RETURN: 'return';
 
 // Operators
@@ -12,21 +24,57 @@ PLUS: '+';
 MINUS: '-';
 MULTIPLY: '*';
 DIVIDE: '/';
+MOD: '%';
+GN: '>=';
+GE: '>';
+LN: '<=';
+LE: '<';
+NE: '!=';
+EQ: '==';
+LAND: '&&';
+LOR: '||';
+LNOT: '!';
+ARS: '>>';
+ALS: '<<';
+BAND: '&';
+BOR: '|';
+BXOR: '^';
+BNOT: '~';
 ASSIGN: '=';
+SELF_PLUS: '++';
+SELF_MINUS: '--';
+DOT: '.';
+QUESTION_MARK: '?';
+COLON: ':';
 
 // Punctuation
 LPAREN: '(';
 RPAREN: ')';
+LBRACKET: '[';
+RBRACKET: ']';
 LBRACE: '{';
 RBRACE: '}';
 SEMICOLON: ';';
+COMMA: ',';
+FORMAT_STRING_WHOLE: 'f"' FORMAT_STR_INTERNAL '"';
+FORMAT_STRING_HEAD: 'f"' FORMAT_STR_INTERNAL '$';
+FORMAT_STRING_TAIL: '$' FORMAT_STR_INTERNAL '"';
+FORMAT_STRING_BODY: '$' FORMAT_STR_INTERNAL '$';
+fragment FORMAT_STR_INTERNAL : (DOLLAR_IN_FORMAT_STR | ESC | ~[$"\\])*?;
+fragment FORMAT_STR_BEG: 'f"';
+fragment FORMAT_STR_END: '"';
+fragment DOLLAR_IN_FORMAT_STR: '$$';
+fragment DOLLAR: '$';
 
 // Identifiers
-ID: [a-zA-Z_][a-zA-Z_0-9]*;
+ID: [a-zA-Z][a-zA-Z_0-9]*;
 
 // Literals
 INT_LITERAL: [0-9]+;
+STRING_LITERAL: '"' (ESC|~["\\])*? '"';
+fragment ESC : '\\"' | '\\\\';
 
 // Whitespace and comments
 WS: [ \t\r\n]+ -> skip;
-COMMENT: '//' ~[\r\n]* -> skip;
+LINE_COMMENT : '//' .*? '\r'? '\n' -> skip;
+BLOCK_COMMENT: '/*' .*? '*/'       -> skip;

grammar/MXParser.g4

@@ -3,39 +3,74 @@ parser grammar MXParser;
 options { tokenVocab=MXLexer; }
 
 mxprog
-    : function* EOF
+    : (class_def|define_statement|function_def)+ EOF
     ;
 
-function
-    : type ID LPAREN RPAREN block
+function_def
+    : type ID LPAREN ((type ID)(COMMA type ID)*)? RPAREN suite
     ;
 
+class_def : CLASS ID LBRACE (class_var_def|class_constructor|function_def)* RBRACE SEMICOLON;
+class_var_def : type (LBRACKET RBRACKET)* ID (COMMA ID )* SEMICOLON;
+class_constructor : ID LPAREN RPAREN suite;
+suite : LBRACE statement* RBRACE;
+statement
+    : SEMICOLON
+    | define_statement
+    | expr SEMICOLON
+    | IF LPAREN expr RPAREN statement (ELSE statement)?
+    | WHILE LPAREN expr RPAREN statement
+    | FOR LPAREN (define_statement|(expr SEMICOLON)|SEMICOLON) expr? SEMICOLON (define_statement|expr)? RPAREN statement
+    | (BREAK|CONTINUE|(RETURN expr?)) SEMICOLON
+    | suite
+    ;
+define_statement : type (LBRACKET RBRACKET)* ID (ASSIGN expr)? (COMMA ID (ASSIGN expr)?)* SEMICOLON;
+expr
+    : basic_expr
+    | LPAREN expr RPAREN
+    | expr DOT (ID|(ID LPAREN (expr (COMMA expr)*)? RPAREN))
+    | expr (LBRACKET expr RBRACKET)+
+    | expr (SELF_PLUS | SELF_MINUS)
+    | (SELF_PLUS | SELF_MINUS) expr
+    | MINUS expr
+    | LNOT expr
+    | BNOT expr
+    | expr (MULTIPLY|DIVIDE|MOD) expr
+    | expr (PLUS|MINUS) expr
+    | expr (ARS|ALS) expr
+    | expr (GN|GE|LN|LE) expr
+    | expr (NE|EQ) expr
+    | expr BAND expr
+    | expr BXOR expr
+    | expr BOR expr
+    | expr LAND expr
+    | expr LOR expr
+    | expr QUESTION_MARK expr COLON expr
+    | <assoc=right> expr ASSIGN expr
+    | NEW ID
+    | NEW ID LPAREN RPAREN
+    | NEW ID (LBRACKET expr RBRACKET)*(LBRACKET RBRACKET)*constant?
+    ;
+
+basic_expr
+    : THIS
+    | ID
+    | ID LPAREN (expr (COMMA expr)*)? RPAREN
+    | formatted_string
+    | constant
+    ;
+
+formatted_string : FORMAT_STRING_WHOLE | (FORMAT_STRING_HEAD expr (FORMAT_STRING_BODY expr)*? FORMAT_STRING_TAIL);
+constant : TRUE
+          |FALSE
+          |INT_LITERAL
+          |STRING_LITERAL
+          |NULL
+          |LBRACE (constant (COMMA constant)*)? RBRACE;
 type
     : INT
+    | BOOL
+    | STRING
     | VOID
-    ;
-
-block
-    : LBRACE statement* RBRACE
-    ;
-
-statement
-    : expression SEMICOLON
-    | returnStmt
-    | ifStmt
-    ;
-
-expression
-    : INT_LITERAL
     | ID
-    | expression (PLUS | MINUS | MULTIPLY | DIVIDE) expression
-    | LPAREN expression RPAREN
-    ;
-
-returnStmt
-    : RETURN expression? SEMICOLON
-    ;
-
-ifStmt
-    : IF LPAREN expression RPAREN statement (ELSE statement)?
-    ;
+    ;

grammar/build.sh

@@ -9,5 +9,7 @@ OUTPUT_DIR="../src/semantic/antlr-generated"
 mkdir -p "$OUTPUT_DIR"
 # Run ANTLR to generate lexer and parser
 antlr4 -Dlanguage=Cpp -no-listener -visitor MXLexer.g4 MXParser.g4 -o "$OUTPUT_DIR"
+rm "$OUTPUT_DIR/MXParserBaseVisitor.cpp"
+rm "$OUTPUT_DIR/MXParserBaseVisitor.h"
 # Return to the original directory
 cd - > /dev/null