diff options
| -rw-r--r-- | src/parser.zig | 138 | ||||
| -rw-r--r-- | src/tokenizer.zig | 2 |
2 files changed, 72 insertions, 68 deletions
diff --git a/src/parser.zig b/src/parser.zig index 5b6915f..f55f5a2 100644 --- a/src/parser.zig +++ b/src/parser.zig @@ -8,31 +8,30 @@ const ParserError = error{ const NodeType = enum { PROGRAM, + STATEMENT, VARIABLE_STATEMENT, PRINT_STATEMENT, - NUMBER, - IDENTIFIER, + EXPRESSION, }; -pub const Node = union(NodeType) { - PROGRAM: struct { - statements: []*Node, - }, - VARIABLE_STATEMENT: struct { - is_declaration: bool, - name: []const u8, - expression: *Node, - }, - PRINT_STATEMENT: struct { - expression: *Node, - }, +pub const Node = union(NodeType) { PROGRAM: struct { + statements: []*Node, +}, STATEMENT: struct { + statement: *Node, +}, VARIABLE_STATEMENT: struct { + is_declaration: bool, + name: []const u8, + expression: *Node, +}, PRINT_STATEMENT: struct { + expression: *Node, +}, EXPRESSION: union { NUMBER: struct { value: i64, }, IDENTIFIER: struct { name: []const u8, }, -}; +} }; pub const Parser = struct { tokens: []tokenizer.Token, @@ -50,18 +49,22 @@ pub const Parser = struct { return parser; } + //TODO: We maybe should use an arena alocator or smth pub fn deinit(self: *Parser, ast: ?*Node) void { if (ast != null) { - // std.debug.assert(ast.?.* == .PROGRAM); + std.debug.assert(ast.?.* == .PROGRAM); for (ast.?.PROGRAM.statements) |statement| { switch (statement.*) { - .VARIABLE_STATEMENT => @panic("NOT IMPLEMENTED"), - .PRINT_STATEMENT => |x| { - self.allocator.destroy(x.expression); + .STATEMENT => |x| { + switch (x.statement.*) { + .PRINT_STATEMENT => |p| { + self.allocator.destroy(p.expression); + }, + else => @panic("NOT IMPLEMENTED"), + } + self.allocator.destroy(x.statement); }, - .NUMBER => @panic("NOT IMPLEMENTED"), - .IDENTIFIER => @panic("NOT IMPLEMENTED"), - else => unreachable, + else => @panic("NOT IMPLEMENTED"), } self.allocator.destroy(statement); } @@ -89,55 +92,35 @@ pub const Parser = struct { return node; } - fn parse_identifier(self: *Parser) ParserError!*Node { - const token = self.peek_token() orelse return ParserError.ParsingError; - - if (token != .IDENTIFIER) return ParserError.ParsingError; - - _ = self.consume_token(); - - const node = try self.allocator.create(Node); - node.* = .{ .IDENTIFIER = .{ - .name = token.IDENTIFIER, - } }; - return node; - } - - fn parse_number(self: *Parser) ParserError!*Node { - const token = self.peek_token() orelse return ParserError.ParsingError; - - if (token != .NUMBER) return ParserError.ParsingError; + fn parse_statement(self: *Parser) ParserError!*Node { + //TODO: Add support for parsing variable declaration and assignment. Also here we shouldnt parse numbers/identifiers directly + const token = self.peek_token(); + std.debug.print("PARSING: {any}\n", .{token}); - _ = self.consume_token(); + const print_statement = try self.parse_print_statement(); + _ = try self.accept_token(tokenizer.TokenType.SEMICOLON); const node = try self.allocator.create(Node); - node.* = .{ .NUMBER = .{ - .value = token.NUMBER, - } }; + node.* = .{ + .STATEMENT = .{ + .statement = print_statement, + }, + }; return node; } fn parse_print_statement(self: *Parser) ParserError!*Node { // print + ( + statement + ) + ; - var token = self.consume_token() orelse return ParserError.ParsingError; - if (token != .PRINT) return ParserError.ParsingError; + _ = try self.accept_token(tokenizer.TokenType.PRINT); - token = self.consume_token() orelse return ParserError.ParsingError; + _ = try self.accept_token(tokenizer.TokenType.LPAREN); - if (token != .LPAREN) return ParserError.ParsingError; - - const expression = try self.parse_statement(); + const expression = try self.parse_expression(); std.debug.print("PARSED expression: {any}\n", .{expression}); - token = self.consume_token() orelse return ParserError.ParsingError; - - if (token != .RPAREN) return ParserError.ParsingError; - - token = self.consume_token() orelse return ParserError.ParsingError; - - if (token != .SEMICOLON) return ParserError.ParsingError; //TODO: This should not be handled at this level + _ = try self.accept_token(tokenizer.TokenType.RPAREN); const node = try self.allocator.create(Node); node.* = .{ @@ -148,21 +131,42 @@ pub const Parser = struct { return node; } - fn parse_statement(self: *Parser) ParserError!*Node { + // Expression :== NUMBER | IDENTIFIER + fn parse_expression(self: *Parser) ParserError!*Node { const token = self.peek_token() orelse return ParserError.ParsingError; - std.debug.print("TOKEN: {any}\n", .{token}); - - //TODO: Add support for parsing variable declaration and assignment. Also here we shouldnt parse numbers/identifiers directly + var node: *Node = undefined; if (token == .NUMBER) { - return self.parse_number(); - } else if (token == .IDENTIFIER) { - return self.parse_identifier(); - } else if (token == .PRINT) { - return self.parse_print_statement(); + const a = try self.accept_token(tokenizer.TokenType.NUMBER); + node = try self.allocator.create(Node); + node.* = .{ + .EXPRESSION = .{ + .NUMBER = .{ + .value = a.NUMBER, + }, + }, + }; } else { - return ParserError.ParsingError; + const a = try self.accept_token(tokenizer.TokenType.IDENTIFIER); + node = try self.allocator.create(Node); + node.* = .{ + .EXPRESSION = .{ + .IDENTIFIER = .{ + .name = a.IDENTIFIER, + }, + }, + }; } + + return node; + } + + fn accept_token(self: *Parser, expected_token: tokenizer.TokenType) ParserError!tokenizer.Token { + const token = self.peek_token() orelse return ParserError.ParsingError; + + if (token != expected_token) return ParserError.ParsingError; + + return self.consume_token() orelse unreachable; } fn consume_token(self: *Parser) ?tokenizer.Token { diff --git a/src/tokenizer.zig b/src/tokenizer.zig index e7683e9..688a32f 100644 --- a/src/tokenizer.zig +++ b/src/tokenizer.zig @@ -53,7 +53,7 @@ pub const Tokenizer = struct { } }; -const TokenType = enum { +pub const TokenType = enum { // Keywords LET, PRINT, |