SeaLang/src/lexer.rs
2024-08-07 00:35:10 +02:00

640 lines
18 KiB
Rust

use crate::tokens::Token;
use crate::tokens::TokenPos;
use itertools::Itertools;
use crate::common::FallibleParse;
use crate::common::NextIf;
#[derive(Debug, thiserror::Error)]
pub enum Error {
#[error("{0}")]
StringError(String),
#[error("Exp part of floating constant had no digits.")]
FloatingConstantExpPartNoDigit,
#[error("Dummy Message.")]
NumericalConstantDigitLeadingUnderscore,
#[error("Dummy Message.")]
NumericalConstantDigitNoDigit,
#[error("Dummy Message.")]
IntegralTypeExpectedDigit,
#[error("Dummy Message.")]
FloatingConstantInvalidTrailingType,
#[error("Dummy Message.")]
InvalidToken,
#[error("Dummy Message.")]
ExpectedIdStartForIdentifier,
}
pub type Result<T> = core::result::Result<T, Error>;
#[derive(Debug, Clone)]
pub struct Chars<'a> {
bytes: &'a [u8],
offset: usize,
}
impl<'a> Chars<'a> {
pub fn as_str(&self) -> &str {
unsafe { core::str::from_utf8_unchecked(&self.bytes[self.offset..]) }
}
pub fn is_eof(&self) -> bool {
self.offset >= self.bytes.len()
}
pub fn peek(&self) -> Option<char> {
self.clone().next()
}
pub fn position(&self) -> u32 {
self.offset() as u32
}
pub fn offset(&self) -> usize {
self.offset
}
pub fn get_source_span(
&self,
start: u32,
end: u32,
) -> std::ops::RangeInclusive<SourceLocation> {
let (start_l, start_c) = {
let range = self.get_range(0, start);
range.chars().fold((1u32, 0u32), |(line, col), c| {
if c == '\n' {
(line + 1, 0)
} else {
(line, col + 1)
}
})
};
let (end_l, end_c) = {
let range = self.get_range(start, end);
range.chars().fold((start_l, start_c), |(line, col), c| {
if c == '\n' {
(line + 1, 0)
} else {
(line, col + 1)
}
})
};
core::ops::RangeInclusive::new(
SourceLocation::new(start_l, start_c),
SourceLocation::new(end_l, end_c),
)
}
pub fn get_range(&self, start: u32, end: u32) -> &str {
unsafe { core::str::from_utf8_unchecked(&self.bytes[start as usize..end as usize]) }
}
fn next_char(&mut self) -> Option<char> {
let ch = self.as_str().chars().next()?;
self.offset += ch.len_utf8();
Some(ch)
}
}
impl<'a> Iterator for Chars<'a> {
type Item = char;
fn next(&mut self) -> Option<Self::Item> {
self.next_char()
}
}
#[derive(Debug, Clone)]
pub struct Tokenizer<'a> {
source: Chars<'a>,
tokens: Vec<TokenPos>,
}
#[derive(Debug, Clone)]
pub struct TokenIterator<'a> {
tokenizer: &'a Tokenizer<'a>,
offset: usize,
}
impl<'a> TokenIterator<'a> {
pub fn expect_token(&mut self, token: Token) -> crate::parser::Result<TokenItem<'a>> {
self.next_if(|item| item.token() == token)
.ok_or(crate::parser::Error::ExpectedTokenNotFound(token))
}
pub fn eat_token(&mut self, token: Token) -> Option<TokenItem<'a>> {
self.next_if(|item| item.token() == token)
}
pub fn peek_token(&mut self) -> Option<TokenItem<'a>> {
self.clone().next()
}
pub fn peek_token_or_err(&mut self) -> crate::parser::Result<TokenItem<'a>> {
self.clone()
.next()
.ok_or(crate::parser::Error::UnexpectedEndOfTokens)
}
pub fn peek_expect_token(&mut self, token: Token) -> crate::parser::Result<TokenItem<'a>> {
self.clone()
.next()
.ok_or(crate::parser::Error::ExpectedTokenNotFound(token))
}
pub fn is_next_token(&mut self, token: Token) -> bool {
self.clone().next_if(|item| item.token() == token).is_some()
}
}
pub struct TokenItem<'a> {
tokenizer: &'a Tokenizer<'a>,
inner: TokenPos,
}
#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Clone, Copy, Hash)]
pub struct SourceLocation {
pub line: u32,
pub column: u32,
}
impl SourceLocation {
pub fn new(line: u32, column: u32) -> Self {
Self { line, column }
}
}
impl<'a> TokenItem<'a> {
pub fn token(&self) -> Token {
self.inner.token
}
pub fn lexeme(&self) -> &str {
self.tokenizer
.source
.get_range(self.inner.start, self.inner.end)
}
pub fn source_location(&self) -> std::ops::RangeInclusive<SourceLocation> {
self.tokenizer
.source
.get_source_span(self.inner.start, self.inner.end)
}
}
impl<'a> Iterator for TokenIterator<'a> {
type Item = TokenItem<'a>;
fn next(&mut self) -> Option<Self::Item> {
if self.tokenizer.tokens.len() >= self.offset {
None
} else {
let index = self.offset;
self.offset += 1;
match self.tokenizer.tokens[index].token {
Token::SlashSlash
| Token::SlashSlashSlash
| Token::SlashStar
| Token::SlashStarStar
| Token::Comment
| Token::DocComment => self.next(),
_ => Some(Self::Item {
tokenizer: self.tokenizer,
inner: self.tokenizer.tokens[index],
}),
}
}
}
}
macro_rules! next_or_eof {
($expr:expr) => {
match $expr.next() {
Some(c) => c,
None => {
return Ok(Token::Eof);
}
}
};
(?$expr:expr) => {
match $expr.peek() {
Some(c) => c,
None => {
return Ok(Token::Eof);
}
}
};
}
macro_rules! residual {
(ok: $expr:expr) => {
match $expr {
Ok(t) => t,
Err(e) => {
return Err(e);
}
}
};
(none: $expr:expr) => {
match $expr {
Ok(Some(t)) => {
return Ok(Some(t));
}
Ok(val) => val,
Err(e) => {
return Err(e);
}
}
};
(flatten: none: $expr:expr) => {
match $expr {
Ok(Some(t)) => {
return Ok(t);
}
Ok(val) => val,
Err(e) => {
return Err(e);
}
}
};
(some: $expr:expr) => {
match $expr {
Ok(Some(t)) => t,
Ok(None) => {
return Ok(None);
}
Err(e) => {
return Err(e);
}
}
};
}
impl<'a> Tokenizer<'a> {
pub fn iter(&self) -> TokenIterator {
TokenIterator {
tokenizer: self,
offset: 0,
}
}
fn push_token(&mut self, token: Token, start: u32, end: u32) -> Result<()> {
self.tokens.push(TokenPos::new(token, start, end));
Ok(())
}
pub fn next_token(&mut self) -> Result<()> {
self.source
.take_while_ref(|&c| crate::common::is_whitespace(c))
.count();
let start = self.source.position();
let token = self.source.try_parse_result(|source| {
let a = try_parse_integral_type(source).map(|o| o.map(|_| Token::IntegralType));
residual!(none: a);
let mut peeking = source.clone();
match peeking.next() {
Some('0'..='9') => {
return Ok(Some(parse_constant(source)?));
}
Some('.') if peeking.next().map(|c| ['b', 'x', 'o'].contains(&c)) == Some(true) => {
return Ok(Some(parse_constant(source)?));
}
_ => {}
}
Ok(None)
});
if let Some(token) = token? {
return self.push_token(token, start, self.source.position());
}
// lexical tokens
let token = crate::tokens::LexemeParser::parse(self.source.clone());
if let Some(token) = token {
_ = self.source.advance_by(token.lexeme_len());
match token {
Token::SlashSlash | Token::SlashSlashSlash => {
_ = self.push_token(token, start, self.source.position());
let start = self.source.position();
loop {
// advance until either EOF or newline
let Some(ch) = self.source.next() else {
break;
};
if ch == '\n' {
break;
}
}
let end = self.source.position() - 1;
return self.push_token(
if token == Token::SlashSlash {
Token::Comment
} else {
Token::DocComment
},
start,
end,
);
}
Token::SlashStar | Token::SlashStarStar => {
let start = self.source.position();
let mut end = self.source.position();
let mut last = self.source.next();
loop {
// break out of loop if EOF
let Some(l) = last.replace(match self.source.next() {
Some(ch) => ch,
None => {
break;
}
}) else {
break;
};
// break out of loop if end of comment
if (l, last.unwrap()) == ('*', '/') {
break;
}
end = self.source.position() - 1;
}
return self.push_token(
if token == Token::SlashStar {
Token::Comment
} else {
Token::DocComment
},
start,
end,
);
}
_ => {}
}
if token.maybe_ident() {
if self
.source
.take_while_ref(|&c| crate::common::is_id_continue(c))
.count()
.gt(&0)
{
return self.push_token(Token::Ident, start, self.source.position());
}
}
return self.push_token(token, start, self.source.position());
}
self.source
.next_if(|&c| crate::common::is_id_start(c))
.ok_or(Error::ExpectedIdStartForIdentifier)?;
self.source
.take_while_ref(|&c| crate::common::is_id_continue(c))
.count();
return self.push_token(Token::Ident, start, self.source.position());
}
}
/// IntegralType <-
/// ( 'u' | 'i' ) DIGITS+
fn try_parse_integral_type(source: &mut Chars) -> Result<Option<()>> {
if !source.next_if(|&c| c == 'u' || c == 'i').is_some() {
return Ok(None);
}
if source
.take_while_ref(|&c| crate::common::is_digit(c))
.count()
<= 0
{
return Err(Error::IntegralTypeExpectedDigit);
};
Ok(Some(()))
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum Radix {
Hex,
Bin,
Dec,
Oct,
}
impl Radix {
/// must be called with one of `['b','x','d','o']`
unsafe fn from_char_unchecked(c: char) -> Self {
match c.to_ascii_lowercase() {
'o' => Self::Oct,
'b' => Self::Bin,
'x' => Self::Hex,
'd' => Self::Dec,
_ => unreachable!(),
}
}
fn from_char(c: char) -> Option<Self> {
match c.to_ascii_lowercase() {
'o' => Some(Self::Oct),
'b' => Some(Self::Bin),
'x' => Some(Self::Hex),
'd' => Some(Self::Dec),
_ => None,
}
}
fn radix(self) -> u8 {
match self {
Radix::Hex => 16,
Radix::Bin => 2,
Radix::Oct => 8,
Radix::Dec => 10,
}
}
fn to_token(self) -> Token {
match self {
Radix::Hex => Token::IntegerHexConstant,
Radix::Bin => Token::IntegerBinConstant,
Radix::Oct => Token::IntegerOctConstant,
Radix::Dec => Token::IntegerConstant,
}
}
pub fn from_token(token: Token) -> Option<Self> {
match token {
Token::IntegerHexConstant => Some(Radix::Hex),
Token::IntegerBinConstant => Some(Radix::Bin),
Token::IntegerOctConstant => Some(Radix::Oct),
Token::IntegerConstant => Some(Radix::Dec),
_ => None,
}
}
pub fn folding_method(self) -> fn(u64, char) -> u64 {
match self {
Radix::Hex => {
fn fold(acc: u64, c: char) -> u64 {
let digit = match c {
'0'..='9' => c as u8 - b'0',
'a'..='f' => c as u8 - b'a',
'A'..='F' => c as u8 - b'A',
_ => unreachable!(),
};
acc + digit as u64 * 16
}
fold
}
Radix::Bin => {
fn fold(acc: u64, c: char) -> u64 {
let digit = match c {
'0'..='1' => c as u8 - b'0',
_ => unreachable!(),
};
acc + digit as u64 * 2
}
fold
}
Radix::Dec => {
fn fold(acc: u64, c: char) -> u64 {
let digit = match c {
'0'..='9' => c as u8 - b'0',
_ => unreachable!(),
};
acc + digit as u64 * 10
}
fold
}
Radix::Oct => {
fn fold(acc: u64, c: char) -> u64 {
let digit = match c {
'0'..='7' => c as u8 - b'0',
_ => unreachable!(),
};
acc + digit as u64 * 8
}
fold
}
}
}
pub fn is_digit(self) -> fn(char) -> bool {
match self {
Radix::Hex => crate::common::is_hex_digit,
Radix::Bin => crate::common::is_bin_digit,
Radix::Oct => crate::common::is_oct_digit,
Radix::Dec => crate::common::is_digit,
}
}
}
/// where DIGIT is defined by radix:
/// DIGITS <-
/// if allow_leading_underscore: `_`* DIGIT (DIGIT|`_`)*
/// else: DIGIT (DIGIT|`_`)*
fn parse_digit_part(
source: &mut Chars,
allow_leading_underscore: bool,
radix: Radix,
) -> Result<()> {
let radix = radix.is_digit();
if allow_leading_underscore {
let _underscore = source.take_while_ref(|&c| c == '_').count();
}
let _need_digit = source.next_if(|&c| radix(c)).ok_or_else(|| {
if source.peek() == Some('_') {
Error::NumericalConstantDigitLeadingUnderscore
} else {
Error::NumericalConstantDigitNoDigit
}
})?;
let _rest = source.take_while_ref(|&c| radix(c) || c == '_').count();
Ok(())
}
/// returns `Err(E)` if it failed to parse.
/// returns `Ok(None)` if no exp part was found.
/// returns `Ok(Some(()))` if an exp part was found and parsed.
///
/// EXP_PART <-
/// (`e`|`E`) (`-`|`+`)? DEC_DIGITS
fn try_parse_exp_part(source: &mut Chars) -> Result<Option<()>> {
if source.next_if(|&c| c.to_ascii_lowercase() == 'e').is_some() {
let _sign = source.next_if(|&c| c == '-' || c == '+');
if source
.take_while_ref(|&c| crate::common::is_digit(c))
.count()
.lt(&1)
{
// need digits following exp notation
Err(Error::FloatingConstantExpPartNoDigit)
} else {
Ok(Some(()))
}
} else {
Ok(None)
}
}
/// CONSTANT <-
/// DEC_DIGITS IntegralType?
/// `0x` HEX_DIGITS IntegralType?
/// `0b` BIN_DIGITS IntegralType?
/// `0o` OCT_DIGITS IntegralType?
/// DEC_DIGITS FloatingType?
/// `.` DEC_DIGITS EXP_PART? FloatingType?
/// DEC_DIGITS `.` DEC_DIGITS? EXP_PART? FloatingType?
fn parse_constant(source: &mut Chars) -> Result<Token> {
let zero = source.next_if(|&c| c == '0').is_some();
let radix = zero
.then(|| source.next_if_map(|c| Radix::from_char(c)))
.flatten();
if let Some(radix) = radix {
parse_digit_part(source, false, radix)?;
if source.peek().map(|c| c == 'u' || c == 'i') == Some(true) {
try_parse_integral_type(source)?;
}
return Ok(radix.to_token());
}
// if zero: `_`* DIGIT (DIGIT|`_`)*
// else: DIGIT (DIGIT|`_`)*
let _digits = parse_digit_part(source, false, Radix::Dec)?;
if let Ok(_) = source.try_parse_result(|source| try_parse_integral_type(source)) {
return Ok(Token::IntegerConstant);
}
let dot = source.next_if(|&c| c == '.').is_some();
if dot {
parse_digit_part(source, false, Radix::Dec)?;
}
// parse exp notation
let exp = try_parse_exp_part(source)?.is_some();
// trailing FloatingType?
let floating = if source.next_if(|&c| c == 'f').is_some() {
let digits = source.next_tuple::<(char, char)>();
if !(digits == Some(('6', '4')) || digits == Some(('3', '2'))) {
// need either f64 or f32 here!
return Err(Error::FloatingConstantInvalidTrailingType);
}
true
} else {
false
};
let token = match (dot, exp, floating) {
(false, false, false) => Token::IntegerConstant,
(true, false, _) => Token::DotFloatingConstant,
(true, true, _) => Token::DotFloatingExpConstant,
(false, true, _) => Token::FloatingExpConstant,
(false, _, _) => Token::FloatingConstant,
};
Ok(token)
}