Parser started: basic literal parsing

Mostly this commit implements structure to implement parsing, not actual
parsing itself

Makefile

@@ -4,6 +4,7 @@ CPPFLAGS:=$(CPPFLAGS) -Ilib/expected/ -Ilib/ut/ -Isrc/
 
 Obj=bin/errors.o \
 		bin/lexer.o \
+		bin/parser.o \
 		bin/unicode.o \
 		bin/unicode_tables.o
 

src/errors.cc

@@ -1,49 +1,112 @@
 #include <musique.hh>
 
+#include <iostream>
+#include <sstream>
+
 bool Error::operator==(errors::Type type)
 {
 	return this->type == type;
 }
 
-std::ostream& operator<<(std::ostream& os, Error const& err)
+Error Error::with(Location loc) &&
 {
-	if (err.location) {
-		os << *err.location;
+	location = loc;
+	return *this;
+}
+
+enum class Error_Level
+{
+	Error,
+	Notice,
+	Bug
+};
+
+static void error_heading(std::ostream &os, std::optional<Location> location, Error_Level lvl)
+{
+	if (location) {
+		os << *location;
 	} else {
 		os << "musique";
 	}
 
-	os << ": error: ";
+	switch (lvl) {
+	case Error_Level::Bug:    os << ": implementation bug: "; return;
+	case Error_Level::Error:  os << ": error: ";              return;
+	case Error_Level::Notice: os << ": notice: ";             return;
+	}
+}
+
+std::ostream& operator<<(std::ostream& os, Error const& err)
+{
+	error_heading(os, err.location, Error_Level::Error);
 
 	switch (err.type) {
 	case errors::End_Of_File:
 		return os << "end of file\n";
 
 	case errors::Unrecognized_Character:
-		if (err.invalid_character) {
-			return os << "unrecognized charater U+" << std::hex << err.invalid_character
-				<< " (char: '" << utf8::Print{err.invalid_character} << "')\n";
-		} else {
-			return os << "unrecognized character\n";
-		}
+		return err.message.empty() ? os << "unrecognized character\n" : os << err.message;
+
+	case errors::Unexpected_Token_Type:
+		return os << err.message;
+
+	case errors::Unexpected_Empty_Source:
+		return os << "unexpected end of input\n";
 	}
 
 	return os << "unrecognized error type\n";
 }
 
+static std::string format(auto const& ...args)
+{
+	std::stringstream ss;
+	(void) (ss << ... << args);
+	return ss.str();
+}
+
 Error errors::unrecognized_character(u32 invalid_character)
 {
 	Error err;
 	err.type = errors::Unrecognized_Character;
-	err.invalid_character = invalid_character;
+	err.message = format(
+		"unrecognized charater U+",
+		std::hex, invalid_character,
+		" (char: '", utf8::Print{invalid_character}, "')");
+
 	return err;
 }
 
 Error errors::unrecognized_character(u32 invalid_character, Location location)
+{
+	return unrecognized_character(invalid_character).with(std::move(location));
+}
+
+Error errors::unexpected_token(Token::Type expected, Token const& unexpected)
 {
 	Error err;
-	err.type = errors::Unrecognized_Character;
-	err.invalid_character = invalid_character;
-	err.location = std::move(location);
+	err.type = errors::Unexpected_Token_Type;
+	err.location = unexpected.location;
+	err.message = format("expected ", expected, ", but got ", unexpected.type);
 	return err;
 }
+
+Error errors::unexpected_end_of_source(Location location)
+{
+	Error err;
+	err.type = errors::Unexpected_Empty_Source;
+	err.location = location;
+	return err;
+}
+
+void errors::all_tokens_were_not_parsed(std::span<Token> tokens)
+{
+	error_heading(std::cerr, std::nullopt, Error_Level::Bug);
+	std::cerr << "remaining tokens after parsing. Listing remaining tokens:\n";
+
+	for (auto const& token : tokens) {
+		error_heading(std::cerr, token.location, Error_Level::Notice);
+		std::cerr << token << '\n';
+	}
+
+	std::exit(1);
+}

src/lexer.cc

@@ -1,5 +1,7 @@
 #include <musique.hh>
 
+#include <iomanip>
+
 constexpr std::string_view Notes_Symbols = "abcedefgh";
 constexpr std::string_view Valid_Operator_Chars =
 	"+-*/:%" // arithmetic
@@ -241,10 +243,9 @@ std::string_view Lexer::finish()
 	return result;
 }
 
-std::ostream& operator<<(std::ostream& os, Token const&)
+std::ostream& operator<<(std::ostream& os, Token const& token)
 {
-	os << "Token";
-	return os;
+	return os << '{' << token.type << ", " << std::quoted(token.source) << ", " << token.location << '}';
 }
 
 std::ostream& operator<<(std::ostream& os, Token::Type type)

src/main.cc

@@ -1,25 +1,42 @@
 #include <iostream>
 #include <musique.hh>
 
-std::string_view Source = R"musique(
-	nums = [ 1 2 3 ]
-	say ( min nums + max nums )
-)musique";
+#include <span>
 
-tl::expected<void, Error> Main()
+static std::string_view pop(std::span<char const*> &span)
 {
-	Lexer lexer{Source};
-	lexer.location.filename = "example.mq";
-
-	for (;;) {
-		auto token = Try(lexer.next_token());
-		std::cout << token << '\n';
-	}
+	auto element = span.front();
+	span = span.subspan(1);
+	return element;
 }
 
-int main()
+Result<Unit> Main(std::span<char const*> args)
 {
-	auto result = Main();
+	while (not args.empty()) {
+		std::string_view arg = pop(args);
+
+		if (arg == "-c" || arg == "--run") {
+			if (args.empty()) {
+				std::cerr << "musique: error: option " << arg << " requires an argument" << std::endl;
+				std::exit(1);
+			}
+
+			auto const source = pop(args);
+			Try(Parser::parse(source, "arguments"));
+			std::cout << "successfully parsed: " << source << " \n";
+			continue;
+		}
+
+		std::cerr << "musique: error: unrecognized command line option: " << arg << std::endl;
+		std::exit(1);
+	}
+
+	return {};
+}
+
+int main(int argc, char const** argv)
+{
+	auto result = Main(std::span{ argv+1, usize(argc-1) });
 	if (not result.has_value()) {
 		std::cerr << result.error() << std::flush;
 		return 1;

src/musique.hh

@@ -6,6 +6,7 @@
 #include <cstring>
 #include <optional>
 #include <ostream>
+#include <span>
 #include <string_view>
 #include <tl/expected.hpp>
 #include <variant>
@@ -23,14 +24,21 @@ using i64 = std::int64_t;
 using usize = std::size_t;
 using isize = std::ptrdiff_t;
 
+struct Unit {};
+
 #define Fun(Function) ([]<typename ...T>(T&& ...args) { return (Function)(std::forward<T>(args)...); })
 
+// Error handling mechanism inspired by Andrew Kelly approach, that was implemented
+// as first class feature in Zig programming language.
 namespace errors
 {
 	enum Type
 	{
 		End_Of_File,
-		Unrecognized_Character
+		Unrecognized_Character,
+
+		Unexpected_Token_Type,
+		Unexpected_Empty_Source,
 	};
 }
 
@@ -58,36 +66,40 @@ struct Error
 {
 	errors::Type type;
 	std::optional<Location> location = std::nullopt;
-	u32 invalid_character = 0;
+	std::string message{};
 
 	bool operator==(errors::Type);
+	Error with(Location) &&;
 };
 
-namespace errors
-{
-	Error unrecognized_character(u32 invalid_character);
-	Error unrecognized_character(u32 invalid_character, Location location);
-}
-
-
 template<typename T>
 struct Result : tl::expected<T, Error>
 {
 	using Storage = tl::expected<T, Error>;
 
-	constexpr Result() = default;
+	constexpr Result()                         = default;
+	constexpr Result(Result const&)            = default;
+	constexpr Result(Result&&)                 = default;
+	constexpr Result& operator=(Result const&) = default;
+	constexpr Result& operator=(Result&&)      = default;
 
 	constexpr Result(errors::Type error)
 		: Storage(tl::unexpected(Error { error }))
 	{
 	}
 
-	constexpr Result(Error error)
+	inline Result(Error error)
 		: Storage(tl::unexpected(std::move(error)))
 	{
 	}
 
+	inline Result(tl::unexpected<Error> error)
+		: Storage(std::move(error))
+	{
+	}
+
 	template<typename ...Args>
+	requires std::is_constructible_v<T, Args...>
 	constexpr Result(Args&& ...args)
 		: Storage( T{ std::forward<Args>(args)... } )
 	{
@@ -230,3 +242,55 @@ struct Lexer
 	// Marks end of token and returns it's matching source
 	std::string_view finish();
 };
+
+struct Ast
+{
+	// Named constructors of AST structure
+	static Ast literal(Token);
+
+	enum class Type
+	{
+		Literal
+	};
+
+	Type type;
+	Token token;
+};
+
+template<typename Expected, typename ...T>
+concept Var_Args = (std::is_same_v<Expected, T> && ...) && (sizeof...(T) >= 1);
+
+struct Parser
+{
+	std::vector<Token> tokens;
+	unsigned token_id = 0;
+
+	// Parses whole source code producing Ast or Error
+	// using Parser structure internally
+	static Result<Ast> parse(std::string_view source, std::string_view filename);
+
+	Result<Ast> parse_expression();
+	Result<Ast> parse_binary_operator();
+	Result<Ast> parse_literal();
+
+	Token consume();
+
+	// Tests if current token has given type
+	bool expect(Token::Type type) const;
+
+	// Ensures that current token has one of types given.
+	// Otherwise returns error
+	Result<Unit> ensure(Token::Type type) const;
+};
+
+namespace errors
+{
+	Error unrecognized_character(u32 invalid_character);
+	Error unrecognized_character(u32 invalid_character, Location location);
+
+	Error unexpected_token(Token::Type expected, Token const& unexpected);
+	Error unexpected_end_of_source(Location location);
+
+	[[noreturn]]
+	void all_tokens_were_not_parsed(std::span<Token>);
+}

src/parser.cc

@@ -0,0 +1,67 @@
+#include <musique.hh>
+
+Result<Ast> Parser::parse(std::string_view source, std::string_view filename)
+{
+	Lexer lexer{source};
+	lexer.location.filename = filename;
+	Parser parser;
+
+	for (;;) if (auto maybe_token = lexer.next_token(); maybe_token.has_value()) {
+		parser.tokens.emplace_back(*std::move(maybe_token));
+	} else if (maybe_token.error().type == errors::End_Of_File) {
+		break;
+	} else {
+		return std::move(maybe_token).error();
+	}
+
+	auto const result = parser.parse_expression();
+
+	if (parser.token_id < parser.tokens.size()) {
+		errors::all_tokens_were_not_parsed(std::span(parser.tokens).subspan(parser.token_id));
+	}
+
+	return result;
+}
+
+Result<Ast> Parser::parse_expression()
+{
+	return parse_binary_operator();
+}
+
+Result<Ast> Parser::parse_binary_operator()
+{
+	return parse_literal();
+}
+
+Result<Ast> Parser::parse_literal()
+{
+	Try(ensure(Token::Type::Numeric));
+	return Ast::literal(consume());
+}
+
+Token Parser::consume()
+{
+	return std::move(tokens[token_id++]);
+}
+
+bool Parser::expect(Token::Type type) const
+{
+	return token_id < tokens.size() && tokens[token_id].type == type;
+}
+
+Result<Unit> Parser::ensure(Token::Type type) const
+{
+	return token_id >= tokens.size()
+		? errors::unexpected_end_of_source(tokens.back().location)
+		: tokens[token_id].type != type
+		? errors::unexpected_token(type, tokens[token_id])
+		: Result<Unit>{};
+}
+
+Ast Ast::literal(Token token)
+{
+	Ast ast;
+	ast.type = Ast::Type::Literal;
+	ast.token = std::move(token);
+	return ast;
+}

src/tests/parser.cc

@@ -0,0 +1,11 @@
+#include <boost/ut.hpp>
+#include <musique.hh>
+
+using namespace boost::ut;
+
+suite parser_test = [] {
+	"Literal parsing"_test = [] {
+		auto result = Parser::parse("1", "test");
+		expect(result.has_value()) << "code was expected to parse, but had not";
+	};
+};