wgsl_parser/

parser.rs

use gramatika::{ParseStreamer, Spanned, SpannedError, Substr, Token as _, TokenCtor};

use crate::{
	comment::Comment,
	token::{Lexer, Token},
	TokenKind,
};

#[cfg(test)]
use crate::utils::{WithComments, WithErrors, WithTokens};

// TODO: Upstream to Gramatika
pub(crate) trait ErrorRecoveringParseStream: ParseStreamer {
	fn push_error(&mut self, error: SpannedError);

	fn parse_seq_separated<P>(
		&mut self,
		separator: Self::Token,
		loop_condition: impl Fn(&mut Self) -> bool,
	) -> gramatika::Result<Vec<P>>
	where
		P: gramatika::Parse<Stream = Self>,
	{
		self.__parse_seq_impl(&loop_condition, |input, parsed| {
			if loop_condition(input) {
				input.consume(separator.clone())?;
			}
			Ok(parsed)
		})
	}

	fn parse_seq_with_finisher<P, R>(
		&mut self,
		loop_condition: impl Fn(&mut Self) -> bool,
		finish: impl Fn(&mut Self, P) -> gramatika::Result<R>,
	) -> gramatika::Result<Vec<R>>
	where
		P: gramatika::Parse<Stream = Self>,
	{
		self.__parse_seq_impl(&loop_condition, finish)
	}

	fn parse_seq<P>(&mut self, loop_condition: impl Fn(&mut Self) -> bool) -> Vec<P>
	where P: gramatika::Parse<Stream = Self> {
		self.__parse_seq_impl(&loop_condition, |_, parsed| Ok(parsed))
			.unwrap()
	}

	#[inline]
	#[doc(hidden)]
	fn __parse_seq_impl<P, R>(
		&mut self,
		loop_condition: &impl Fn(&mut Self) -> bool,
		finish: impl Fn(&mut Self, P) -> gramatika::Result<R>,
	) -> gramatika::Result<Vec<R>>
	where
		P: gramatika::Parse<Stream = Self>,
	{
		let mut results = vec![];
		let mut recovering = false;

		while !self.is_empty() && loop_condition(self) {
			match self.parse::<P>() {
				Ok(parsed) => {
					if recovering {
						recovering = false;
					}

					results.push(finish(self, parsed)?);
				}
				Err(err) => {
					if recovering {
						let needs_discard = self
							.peek()
							.and_then(|next_token| Some(err.span?.contains(next_token.span())));

						if matches!(needs_discard, Some(true)) {
							self.discard();
						}
					} else {
						self.push_error(err);
						recovering = true;
					}
				}
			}
		}

		Ok(results)
	}
}

pub struct ParseStream {
	inner: gramatika::ParseStream<Token, Lexer>,
	comments: Vec<Comment>,
	parsing_comment: bool,
	pub(crate) errors: Vec<SpannedError>,
}

pub struct ParseResult {
	pub source: Substr,
	pub tokens: Vec<Token>,
	pub comments: Vec<Comment>,
	pub errors: Vec<SpannedError>,
}

impl ParseStream {
	pub fn new(lexer: Lexer) -> Self {
		Self {
			inner: gramatika::ParseStream::new(lexer),
			comments: vec![],
			parsing_comment: false,
			errors: vec![],
		}
	}

	pub fn source(&self) -> Substr {
		self.inner.source()
	}

	pub fn into_inner(self) -> ParseResult {
		let (source, tokens) = self.inner.into_inner();
		ParseResult {
			source,
			tokens,
			comments: self.comments,
			errors: self.errors,
		}
	}

	pub fn split_next(
		&mut self,
		split_at: usize,
		ctors: (TokenCtor<Token>, TokenCtor<Token>),
	) -> gramatika::Result<Token> {
		self.inner.split_next(split_at, ctors)
	}

	fn did_parse_comment(&mut self) -> bool {
		let Some(token) = self.inner.peek() else {
			return false;
		};

		match token.kind() {
			TokenKind::CommentStart => {
				self.parsing_comment = true;

				match self.parse::<Comment>() {
					Ok(comment) => {
						self.comments.push(comment);
						self.parsing_comment = false;
						true
					}
					Err(error) => {
						self.errors.push(error);
						self.parsing_comment = false;
						true
					}
				}
			}
			_ => false,
		}
	}
}

impl ErrorRecoveringParseStream for ParseStream {
	fn push_error(&mut self, error: SpannedError) {
		self.errors.push(error);
	}
}

impl<S> From<S> for ParseStream
where S: Into<Substr>
{
	fn from(value: S) -> Self {
		Self {
			inner: gramatika::ParseStream::from(value),
			comments: vec![],
			parsing_comment: false,
			errors: vec![],
		}
	}
}

impl ParseStreamer for ParseStream {
	type Token = Token;

	fn is_empty(&mut self) -> bool {
		self.peek().is_none()
	}

	fn peek(&mut self) -> Option<&Token> {
		if !self.parsing_comment && self.did_parse_comment() {
			self.peek()
		} else {
			self.inner.peek()
		}
	}

	fn prev(&mut self) -> Option<&Token> {
		self.inner.prev()
	}

	fn check_kind(&mut self, kind: TokenKind) -> bool {
		if !self.parsing_comment && self.did_parse_comment() {
			self.check_kind(kind)
		} else {
			self.inner.check_kind(kind)
		}
	}

	fn check(&mut self, compare: Token) -> bool {
		if !self.parsing_comment && self.did_parse_comment() {
			self.check(compare)
		} else {
			self.inner.check(compare)
		}
	}

	fn consume(&mut self, compare: Token) -> gramatika::Result<Token> {
		if !self.parsing_comment && self.did_parse_comment() {
			self.consume(compare)
		} else {
			self.inner.consume(compare)
		}
	}

	fn consume_kind(&mut self, kind: TokenKind) -> gramatika::Result<Token> {
		if !self.parsing_comment && self.did_parse_comment() {
			self.consume_kind(kind)
		} else {
			self.inner.consume_kind(kind)
		}
	}

	fn consume_as(
		&mut self,
		kind: TokenKind,
		convert: TokenCtor<Token>,
	) -> gramatika::Result<Token> {
		if !self.parsing_comment && self.did_parse_comment() {
			self.consume_as(kind, convert)
		} else {
			self.inner.consume_as(kind, convert)
		}
	}

	fn upgrade_last(
		&mut self,
		kind: TokenKind,
		convert: TokenCtor<Token>,
	) -> gramatika::Result<Token> {
		self.inner.upgrade_last(kind, convert)
	}

	fn upgrade(
		&mut self,
		token: Self::Token,
		convert: TokenCtor<Token>,
	) -> gramatika::Result<Token> {
		self.inner.upgrade(token, convert)
	}

	fn discard(&mut self) {
		self.inner.discard();
	}
}

impl Iterator for ParseStream {
	type Item = Token;

	fn next(&mut self) -> Option<Token> {
		self.inner.next()
	}
}

#[cfg(test)]
impl WithErrors for ParseStream {
	type Output = SpannedError;

	fn get(&self) -> &[Self::Output] {
		&self.errors
	}
}

#[cfg(test)]
impl WithComments for ParseStream {
	type Output = Comment;

	fn get(&self) -> &[Self::Output] {
		&self.comments
	}
}

#[cfg(test)]
impl WithTokens for ParseStream {
	type Output = Token;

	fn into(self) -> Vec<Self::Output> {
		let ParseResult { tokens, .. } = self.into_inner();
		tokens
	}
}