tokens.rs

use crate::lexer::Span;
use std::error::Error;
use std::fmt::{format, Debug, Display, Formatter};

#[derive(PartialEq, Clone, Debug)]
pub enum ScriptTokenType<'a> {
	// Structural Tokens
	LeftParen,
	RightParen,
	LeftBrace,
	RightBrace,
	Comma,
	Dot,
	Semicolon,

	// Unary Operators
	Bang,
	Minus,

	// Binary Operators
	Asterisk,
	Slash,
	Plus,
	BangEqual,
	Equal,
	EqualEqual,
	Greater,
	GreaterEqual,
	Less,
	LessEqual,
	DoublePipe,
	DoubleAmpersand,
	Modulo,
	Caret,

	// Literals
	Identifier(&'a str),
	String(&'a str),
	OwnedString(String),
	Integer(i64),
	Float(f64),
	Boolean(bool),

	// Keywords
	Class,
	Else,
	Function,
	For,
	If,
	Null,
	Print,
	Return,
	Super,
	This,
	Let,
	While,
	Export,
	Import,
	Alias,
	From,
	Typeof,
	Finally,

	// Misc
	Eof,
}

impl<'a> ScriptTokenType<'a> {
	pub fn len(&self) -> usize {
		match self {
			ScriptTokenType::LeftParen => 1,
			ScriptTokenType::RightParen => 1,
			ScriptTokenType::LeftBrace => 2,
			ScriptTokenType::RightBrace => 2,
			ScriptTokenType::Comma => 1,
			ScriptTokenType::Dot => 1,
			ScriptTokenType::Minus => 1,
			ScriptTokenType::Plus => 1,
			ScriptTokenType::Semicolon => 1,
			ScriptTokenType::Slash => 1,
			ScriptTokenType::Asterisk => 1,
			ScriptTokenType::Bang => 1,
			ScriptTokenType::BangEqual => 2,
			ScriptTokenType::Equal => 1,
			ScriptTokenType::EqualEqual => 2,
			ScriptTokenType::Greater => 1,
			ScriptTokenType::GreaterEqual => 2,
			ScriptTokenType::Less => 1,
			ScriptTokenType::LessEqual => 2,
			ScriptTokenType::DoublePipe => 2,
			ScriptTokenType::DoubleAmpersand => 2,
			ScriptTokenType::Modulo => 1,
			ScriptTokenType::Caret => 1,
			ScriptTokenType::Identifier(value) => value.len(),
			ScriptTokenType::String(value) => value.len() + 2,
			ScriptTokenType::OwnedString(value) => value.len() + 2,
			ScriptTokenType::Integer(value) => format!("{}", value).len(),
			ScriptTokenType::Float(value) => format!("{}", value).len(),
			ScriptTokenType::Boolean(value) => {
				if *value {
					4
				} else {
					5
				}
			}
			ScriptTokenType::Class => 6,
			ScriptTokenType::Else => 4,
			ScriptTokenType::Function => 2,
			ScriptTokenType::For => 3,
			ScriptTokenType::If => 2,
			ScriptTokenType::Null => 4,
			ScriptTokenType::Print => 5,
			ScriptTokenType::Return => 6,
			ScriptTokenType::Super => 5,
			ScriptTokenType::This => 4,
			ScriptTokenType::Let => 3,
			ScriptTokenType::While => 5,
			ScriptTokenType::Export => 6,
			ScriptTokenType::Import => 6,
			ScriptTokenType::Alias => 2,
			ScriptTokenType::From => 4,
			ScriptTokenType::Eof => 0,
			ScriptTokenType::Typeof => 6,
			ScriptTokenType::Finally => 7,
		}
	}
}

impl<'a> Display for ScriptTokenType<'a> {
	fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
		match self {
			ScriptTokenType::LeftParen => write!(f, "("),
			ScriptTokenType::RightParen => write!(f, ")"),
			ScriptTokenType::LeftBrace => write!(f, "{{"),
			ScriptTokenType::RightBrace => write!(f, "}}"),
			ScriptTokenType::Comma => write!(f, ","),
			ScriptTokenType::Dot => write!(f, "."),
			ScriptTokenType::Minus => write!(f, "-"),
			ScriptTokenType::Plus => write!(f, "+"),
			ScriptTokenType::Semicolon => write!(f, ";"),
			ScriptTokenType::Slash => write!(f, "/"),
			ScriptTokenType::Asterisk => write!(f, "*"),
			ScriptTokenType::Bang => write!(f, "!"),
			ScriptTokenType::BangEqual => write!(f, "!="),
			ScriptTokenType::Equal => write!(f, "="),
			ScriptTokenType::EqualEqual => write!(f, "=="),
			ScriptTokenType::Greater => write!(f, ">"),
			ScriptTokenType::GreaterEqual => write!(f, ">="),
			ScriptTokenType::Less => write!(f, "<"),
			ScriptTokenType::LessEqual => write!(f, "<="),
			ScriptTokenType::DoublePipe => write!(f, "||"),
			ScriptTokenType::DoubleAmpersand => write!(f, "&&"),
			ScriptTokenType::Modulo => write!(f, "%"),
			ScriptTokenType::Caret => write!(f, "^"),
			ScriptTokenType::Identifier(value) => write!(f, "{}", value),
			ScriptTokenType::String(value) => write!(f, "{}", value),
			ScriptTokenType::OwnedString(value) => write!(f, "{}", value),
			ScriptTokenType::Integer(value) => write!(f, "{}", value),
			ScriptTokenType::Float(value) => write!(f, "{}", value),
			ScriptTokenType::Boolean(value) => write!(f, "{}", value),
			ScriptTokenType::Class => write!(f, "struct"),
			ScriptTokenType::Else => write!(f, "else"),
			ScriptTokenType::Function => write!(f, "fn"),
			ScriptTokenType::For => write!(f, "for"),
			ScriptTokenType::If => write!(f, "if"),
			ScriptTokenType::Null => write!(f, "null"),
			ScriptTokenType::Print => write!(f, "print"),
			ScriptTokenType::Return => write!(f, "return"),
			ScriptTokenType::Super => write!(f, "super"),
			ScriptTokenType::This => write!(f, "this"),
			ScriptTokenType::Let => write!(f, "let"),
			ScriptTokenType::While => write!(f, "while"),
			ScriptTokenType::Export => write!(f, "export"),
			ScriptTokenType::Import => write!(f, "import"),
			ScriptTokenType::Alias => write!(f, "as"),
			ScriptTokenType::From => write!(f, "from"),
			ScriptTokenType::Eof => write!(f, ""),
			ScriptTokenType::Typeof => write!(f, "typeof"),
			ScriptTokenType::Finally => write!(f, "finally"),
		}
	}
}

#[derive(Copy, Clone, Debug, PartialEq)]
pub struct TokenFromStringError<'a> {
	source: &'a str,
}
impl<'a> Display for TokenFromStringError<'a> {
	fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
		write!(f, "Failed to parse into token; value {}", self.source)
	}
}
impl<'a> Error for TokenFromStringError<'a> {}

impl<'a> TryFrom<&'a str> for ScriptTokenType<'a> {
	type Error = TokenFromStringError<'a>;

	fn try_from(value: &'a str) -> Result<Self, Self::Error> {
		match value {
			"(" => Ok(ScriptTokenType::LeftParen),
			")" => Ok(ScriptTokenType::RightParen),
			"{" => Ok(ScriptTokenType::LeftBrace),
			"}" => Ok(ScriptTokenType::RightBrace),
			"," => Ok(ScriptTokenType::Comma),
			"." => Ok(ScriptTokenType::Dot),
			"-" => Ok(ScriptTokenType::Minus),
			"+" => Ok(ScriptTokenType::Plus),
			";" => Ok(ScriptTokenType::Semicolon),
			"/" => Ok(ScriptTokenType::Slash),
			"*" => Ok(ScriptTokenType::Asterisk),
			"!" => Ok(ScriptTokenType::Bang),
			"!=" => Ok(ScriptTokenType::BangEqual),
			"=" => Ok(ScriptTokenType::Equal),
			"==" => Ok(ScriptTokenType::EqualEqual),
			">" => Ok(ScriptTokenType::Greater),
			">=" => Ok(ScriptTokenType::GreaterEqual),
			"<" => Ok(ScriptTokenType::Less),
			"<=" => Ok(ScriptTokenType::LessEqual),
			"||" => Ok(ScriptTokenType::DoublePipe),
			"&&" => Ok(ScriptTokenType::DoubleAmpersand),
			"%" => Ok(ScriptTokenType::Modulo),
			"^" => Ok(ScriptTokenType::Caret),
			"struct" => Ok(ScriptTokenType::Class),
			"else" => Ok(ScriptTokenType::Else),
			"fn" => Ok(ScriptTokenType::Function),
			"for" => Ok(ScriptTokenType::For),
			"if" => Ok(ScriptTokenType::If),
			"null" => Ok(ScriptTokenType::Null),
			"print" => Ok(ScriptTokenType::Print),
			"return" => Ok(ScriptTokenType::Return),
			"super" => Ok(ScriptTokenType::Super),
			"this" => Ok(ScriptTokenType::This),
			"let" => Ok(ScriptTokenType::Let),
			"while" => Ok(ScriptTokenType::While),
			"export" => Ok(ScriptTokenType::Export),
			"import" => Ok(ScriptTokenType::Import),
			"as" => Ok(ScriptTokenType::Alias),
			"from" => Ok(ScriptTokenType::From),
			"typeof" => Ok(ScriptTokenType::Typeof),
			"finally" => Ok(ScriptTokenType::Finally),
			"false" => Ok(ScriptTokenType::Boolean(false)),
			"true" => Ok(ScriptTokenType::Boolean(true)),
			_ => Err(TokenFromStringError { source: value }),
		}
	}
}

#[derive(Clone, PartialEq)]
pub struct ScriptToken<'a> {
	pub position: Span<'a>,
	pub token_type: ScriptTokenType<'a>,
}

impl<'a> Display for ScriptToken<'a> {
	fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
		write!(f, "{}", self.token_type)
	}
}

impl<'a> Debug for ScriptToken<'a> {
	fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
		write!(
			f,
			"[{}:{}] {:?}",
			self.position.location_line(),
			self.position.get_column(),
			self.token_type
		)
	}
}

#[cfg(test)]
mod token_tests {
	use super::{ScriptTokenType, TokenFromStringError};
	use test_case::test_case;

	#[test_case("(" => Ok(ScriptTokenType::LeftParen); r#"Parse LeftParen"#)]
	#[test_case(")" => Ok(ScriptTokenType::RightParen); r#"Parse RightParen"#)]
	#[test_case("{" => Ok(ScriptTokenType::LeftBrace); r#"Parse LeftBrace"#)]
	#[test_case("}" => Ok(ScriptTokenType::RightBrace); r#"Parse RightBrace"#)]
	#[test_case("," => Ok(ScriptTokenType::Comma); r#"Parse Comma"#)]
	#[test_case("." => Ok(ScriptTokenType::Dot); r#"Parse Dot"#)]
	#[test_case("-" => Ok(ScriptTokenType::Minus); r#"Parse Minus"#)]
	#[test_case("+" => Ok(ScriptTokenType::Plus); r#"Parse Plus"#)]
	#[test_case(";" => Ok(ScriptTokenType::Semicolon); r#"Parse Semicolon"#)]
	#[test_case("/" => Ok(ScriptTokenType::Slash); r#"Parse Slash"#)]
	#[test_case("*" => Ok(ScriptTokenType::Asterisk); r#"Parse Asterisk"#)]
	#[test_case("!" => Ok(ScriptTokenType::Bang); r#"Parse Bang"#)]
	#[test_case("!=" => Ok(ScriptTokenType::BangEqual); r#"Parse BangEqual"#)]
	#[test_case("=" => Ok(ScriptTokenType::Equal); r#"Parse Equal"#)]
	#[test_case("==" => Ok(ScriptTokenType::EqualEqual); r#"Parse EqualEqual"#)]
	#[test_case(">" => Ok(ScriptTokenType::Greater); r#"Parse Greater"#)]
	#[test_case(">=" => Ok(ScriptTokenType::GreaterEqual); r#"Parse GreaterEqual"#)]
	#[test_case("<" => Ok(ScriptTokenType::Less); r#"Parse Less"#)]
	#[test_case("<=" => Ok(ScriptTokenType::LessEqual); r#"Parse LessEqual"#)]
	#[test_case("||" => Ok(ScriptTokenType::DoublePipe); r#"Parse DoublePipe"#)]
	#[test_case("&&" => Ok(ScriptTokenType::DoubleAmpersand); r#"Parse DoubleAmpersand"#)]
	#[test_case("%" => Ok(ScriptTokenType::Modulo); r#"Parse Modulo"#)]
	#[test_case("^" => Ok(ScriptTokenType::Caret); r#"Parse Caret"#)]
	#[test_case("struct" => Ok(ScriptTokenType::Class); r#"Parse Class"#)]
	#[test_case("else" => Ok(ScriptTokenType::Else); r#"Parse Else"#)]
	#[test_case("fn" => Ok(ScriptTokenType::Function); r#"Parse Function"#)]
	#[test_case("for" => Ok(ScriptTokenType::For); r#"Parse For"#)]
	#[test_case("if" => Ok(ScriptTokenType::If); r#"Parse If"#)]
	#[test_case("null" => Ok(ScriptTokenType::Null); r#"Parse Null"#)]
	#[test_case("print" => Ok(ScriptTokenType::Print); r#"Parse Print"#)]
	#[test_case("return" => Ok(ScriptTokenType::Return); r#"Parse Return"#)]
	#[test_case("super" => Ok(ScriptTokenType::Super); r#"Parse Super"#)]
	#[test_case("this" => Ok(ScriptTokenType::This); r#"Parse This"#)]
	#[test_case("let" => Ok(ScriptTokenType::Let); r#"Parse Let"#)]
	#[test_case("while" => Ok(ScriptTokenType::While); r#"Parse While"#)]
	#[test_case("export" => Ok(ScriptTokenType::Export); r#"Parse Export"#)]
	#[test_case("import" => Ok(ScriptTokenType::Import); r#"Parse Import"#)]
	#[test_case("as" => Ok(ScriptTokenType::Alias); r#"Parse Alias"#)]
	#[test_case("from" => Ok(ScriptTokenType::From); r#"Parse From"#)]
	#[test_case("typeof" => Ok(ScriptTokenType::Typeof); r#"Parse Typeof"#)]
	#[test_case("true" => Ok(ScriptTokenType::Boolean(true)); r#"Parse true"#)]
	#[test_case("false" => Ok(ScriptTokenType::Boolean(false)); r#"Parse false"#)]
	#[test_case("finally" => Ok(ScriptTokenType::Finally); r#"Parse finally"#)]
	fn parse_token_type(value: &'static str) -> Result<ScriptTokenType, TokenFromStringError> {
		ScriptTokenType::try_from(value)
	}
}