15061769677

Committed 16 May 2025 06:02AM UTC coverage: 75.523% (+0.7%) from 74.809%

Build # 15061769677

Build Type

push

github

Committed by

vcfxb

Commit Message

chore: cargo fmt

Run Details

6 of 13 new or added lines in 4 files covered. (46.15%)

27 existing lines in 2 files now uncovered.

938 of 1242 relevant lines covered (75.52%)

29.83 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

79.27

/wright/src/parser.rs

//! This parser module is responsible for turning the stream of [Token]s from the [Lexer] into a tree of [AST] nodes.
//!
//! [AST]: crate::ast
//! [Token]: crate::lexer::token::Token

use error::{ParserError, ParserErrorKind};

use super::lexer::Lexer;
use crate::{
    lexer::token::{Token, TokenTy},
    source_tracking::fragment::Fragment,
};
use std::collections::VecDeque;

mod decl;
pub mod error;
mod identifier;
mod literal;
mod path;
mod ty;
pub mod whitespace;

/// The [Parser] struct wraps a [Lexer] and adds lookahead and functions that are useful for parsing.
#[derive(Debug)]
pub struct Parser {
    lexer: Lexer,
    lookahead: VecDeque<Token>,
}

impl Parser {
    /// Construct a new parser around a given [Lexer].
    pub fn new(lexer: Lexer) -> Self {
        Parser {
            lexer,
            lookahead: VecDeque::new(),
        }
    }

    /// Get the number of remaining bytes on this parser. This is potentially useful for checking
    /// if a parser has advanced between two calls (or checking if a parser has reached end of input).
    pub fn bytes_remaining(&self) -> usize {
        let bytes_remaining_in_lookahead_buffer = self
            .lookahead
            .iter()
            .map(|t| t.fragment.len())
            .sum::<usize>();

        let bytes_remaining_in_lexer = self.lexer.bytes_remaining();

        bytes_remaining_in_lexer + bytes_remaining_in_lookahead_buffer
    }

    /// Get the next [Token] from this [Parser]. This may be a token that's already been peeked.
    ///
    /// Skips any non-document comments encountered via the lexer implementation.
    ///
    /// Return an error if a [Token] with [TokenTy::Unknown] is encountered.
    pub fn next_token(&mut self) -> Result<Option<Token>, ParserError> {
        let token = self
            .lookahead
            .pop_front()
            .or_else(|| self.lexer.next_token());

        // Check for unknown tokens, which should always convert to an error.
        match token {
            Some(Token {
                variant: TokenTy::Unknown,
                fragment,
            }) => Err(ParserErrorKind::EncounteredUnknownToken.at(fragment)),
            known_token_or_none => Ok(known_token_or_none),
        }
    }

    /// Advance this [Parser] by `n` [Token]s. If this [Parser] runs out of [Token]s, panic.
    ///
    /// Panics
    /// - If `n` is greater than the number of remaining tokens.
    pub fn advance(&mut self, n: usize) {
        // Add tokens to the lookahead buffer until we have enough to split off.
        while self.lookahead.len() < n {
            let token = self
                .lexer
                .next_token()
                .expect("advance: `n` <= number of remaining tokens");

            self.lookahead.push_back(token);
        }

        // Split them off.
        self.lookahead = self.lookahead.split_off(n);
    }

    /// Peek at the next token from the [Lexer] (cached in the lookahead queue if peeked before).
    pub fn peek(&mut self) -> Option<&Token> {
        if self.lookahead.is_empty() {
            self.lookahead.push_back(self.lexer.next_token()?);
        }

        self.lookahead.front()
    }

    /// Peek the [Fragment] of the next [Token].
    pub fn peek_fragment(&mut self) -> Option<&Fragment> {
        self.peek().map(|token| &token.fragment)
    }

    /// Peek the [TokenTy] of the next [Token].
    pub fn peek_variant(&mut self) -> Option<TokenTy> {
        self.peek().map(|token| token.variant)
    }

    /// Peek the [Fragment] of the next [Token] and clone it or return a clone of the
    /// remainder [Fragment] of the internal [Lexer]
    /// (which will be empty, since there wasn't a [Token] to peek).
    ///
    /// This is likely only useful for error reporting -- a clone of a potentially empty fragment is
    /// rarely ever useful otherwise.
    pub fn peek_fragment_or_rest_cloned(&mut self) -> Fragment {
        match self.peek() {
            Some(Token { fragment, .. }) => fragment.clone(),
            None => {
                let rest = self.lexer.remaining.clone();

                // Assert that we're making the right assumptions about the remaining fragment.
                // These are (unidiomatically) done using debug_assert -- perhaps that changes eventually
                // however it should be fine for now, since this can only produce logic bugs (never memory or
                // concurrency bugs).
                debug_assert!(rest.is_valid());
                debug_assert!(rest.is_empty());
                debug_assert!(rest.is_empty_at_end_of_source());

                rest
            }
        }
    }

    /// Get the [Lexer] that's wrapped.
    pub fn lexer(&self) -> &Lexer {
        &self.lexer
    }

    /// Lookahead `k` [Token]s.
    ///
    /// If `k == 0` then this is effectively peeking at the next [Token] from the wrapped [Lexer].
    pub fn lookahead(&mut self, k: usize) -> Option<&Token> {
        while self.lookahead.len() <= k {
            self.lookahead.push_back(self.lexer.next_token()?);
        }

        self.lookahead.get(k)
    }

    /// Similar to [Parser::lookahead] but instead returns a slice of `n` [Token]s, starting with the next [Token].
    ///
    /// Returns [None] if `n` is greater than the number of remaining [Token]s for this [Parser].
    pub fn lookahead_window(&mut self, n: usize) -> Option<&[Token]> {
        while self.lookahead.len() < n {
            self.lookahead.push_back(self.lexer.next_token()?);
        }

        // Use make contiguous here to get a unified/single slice.
        Some(&self.lookahead.make_contiguous()[..n])
    }

    /// Get the next [Token] from this parser if its [Token::variant] is the given `token_ty`.
    pub fn next_if_is(&mut self, token_ty: TokenTy) -> Option<Token> {
        // Peeking successfully first means that the lookahead vec will never be empty here.
        (self.peek()?.variant == token_ty)
            // SAFETY: We just peeked a token to check its variant so this unwrap is always ok.
            .then(|| unsafe { self.lookahead.pop_front().unwrap_unchecked() })
    }

    /// Peek at the next [Token]s of this [Parser] and determine if the [Token::variant]s match this
    /// sequence of [TokenTy]s.
    pub fn matches(&mut self, seq: &[TokenTy]) -> bool {
        // Use the rare let-else to ensure there are at minimum, the given number of tokens remaining.
        let Some(lookahead_window) = self.lookahead_window(seq.len()) else {
            return false;
        };

        // Use a zipped iterator to compare all the token variants.
        lookahead_window
            .iter()
            .zip(seq)
            .all(|(token, matches)| token.variant == *matches)
    }
}

1	//! This parser module is responsible for turning the stream of [Token]s from the [Lexer] into a tree of [AST] nodes.
2	//!
3	//! [AST]: crate::ast
4	//! [Token]: crate::lexer::token::Token
5
6	use error::{ParserError, ParserErrorKind};
7
8	use super::lexer::Lexer;
9	use crate::{
10	lexer::token::{Token, TokenTy},
11	source_tracking::fragment::Fragment,
12	};
13	use std::collections::VecDeque;
14
15	mod decl;
16	pub mod error;
17	mod identifier;
18	mod literal;
19	mod path;
20	mod ty;
21	pub mod whitespace;
22
23	/// The [Parser] struct wraps a [Lexer] and adds lookahead and functions that are useful for parsing.
24	#[derive(Debug)]
25	pub struct Parser {
26	lexer: Lexer,
27	lookahead: VecDeque<Token>,
28	}
29
30	impl Parser {
31	/// Construct a new parser around a given [Lexer].
32	pub fn new(lexer: Lexer) -> Self {	25✔
33	Parser {	25✔
34	lexer,	25✔
35	lookahead: VecDeque::new(),	25✔
36	}	25✔
37	}	25✔
38
39	/// Get the number of remaining bytes on this parser. This is potentially useful for checking
40	/// if a parser has advanced between two calls (or checking if a parser has reached end of input).
41	pub fn bytes_remaining(&self) -> usize {	1✔
42	let bytes_remaining_in_lookahead_buffer = self	1✔
43	.lookahead	1✔
44	.iter()	1✔
45	.map(\|t\| t.fragment.len())	1✔
46	.sum::<usize>();	1✔
47
48	let bytes_remaining_in_lexer = self.lexer.bytes_remaining();	1✔
49
50	bytes_remaining_in_lexer + bytes_remaining_in_lookahead_buffer	1✔
51	}	1✔
52
53	/// Get the next [Token] from this [Parser]. This may be a token that's already been peeked.
54	///
55	/// Skips any non-document comments encountered via the lexer implementation.
56	///
57	/// Return an error if a [Token] with [TokenTy::Unknown] is encountered.
58	pub fn next_token(&mut self) -> Result<Option<Token>, ParserError> {	3✔
59	let token = self	3✔
60	.lookahead	3✔
61	.pop_front()	3✔
62	.or_else(\|\| self.lexer.next_token());	3✔
63
64	// Check for unknown tokens, which should always convert to an error.
65	match token {	3✔
66	Some(Token {
67	variant: TokenTy::Unknown,
68	fragment,	×
69	}) => Err(ParserErrorKind::EncounteredUnknownToken.at(fragment)),	×
70	known_token_or_none => Ok(known_token_or_none),	3✔
71	}
72	}	3✔
73
74	/// Advance this [Parser] by `n` [Token]s. If this [Parser] runs out of [Token]s, panic.
75	///
76	/// Panics
77	/// - If `n` is greater than the number of remaining tokens.
78	pub fn advance(&mut self, n: usize) {	12✔
79	// Add tokens to the lookahead buffer until we have enough to split off.
80	while self.lookahead.len() < n {	12✔
UNCOV 81	let token = self	×
UNCOV 82	.lexer	×
UNCOV 83	.next_token()	×
UNCOV 84	.expect("advance: `n` <= number of remaining tokens");	×
UNCOV 85		×
UNCOV 86	self.lookahead.push_back(token);	×
UNCOV 87	}	×
88
89	// Split them off.
90	self.lookahead = self.lookahead.split_off(n);	12✔
91	}	12✔
92
93	/// Peek at the next token from the [Lexer] (cached in the lookahead queue if peeked before).
94	pub fn peek(&mut self) -> Option<&Token> {	84✔
95	if self.lookahead.is_empty() {	84✔
96	self.lookahead.push_back(self.lexer.next_token()?);	40✔
97	}	44✔
98
99	self.lookahead.front()	82✔
100	}	84✔
101
102	/// Peek the [Fragment] of the next [Token].
UNCOV 103	pub fn peek_fragment(&mut self) -> Option<&Fragment> {	×
UNCOV 104	self.peek().map(\|token\| &token.fragment)	×
UNCOV 105	}	×
106
107	/// Peek the [TokenTy] of the next [Token].
108	pub fn peek_variant(&mut self) -> Option<TokenTy> {	19✔
109	self.peek().map(\|token\| token.variant)	19✔
110	}	19✔
111
112	/// Peek the [Fragment] of the next [Token] and clone it or return a clone of the
113	/// remainder [Fragment] of the internal [Lexer]
114	/// (which will be empty, since there wasn't a [Token] to peek).
115	///
116	/// This is likely only useful for error reporting -- a clone of a potentially empty fragment is
117	/// rarely ever useful otherwise.
118	pub fn peek_fragment_or_rest_cloned(&mut self) -> Fragment {	15✔
119	match self.peek() {	15✔
120	Some(Token { fragment, .. }) => fragment.clone(),	14✔
121	None => {
122	let rest = self.lexer.remaining.clone();	1✔
123
124	// Assert that we're making the right assumptions about the remaining fragment.
125	// These are (unidiomatically) done using debug_assert -- perhaps that changes eventually
126	// however it should be fine for now, since this can only produce logic bugs (never memory or
127	// concurrency bugs).
128	debug_assert!(rest.is_valid());	1✔
129	debug_assert!(rest.is_empty());	1✔
130	debug_assert!(rest.is_empty_at_end_of_source());	1✔
131
132	rest	1✔
133	}
134	}
135	}	15✔
136
137	/// Get the [Lexer] that's wrapped.
138	pub fn lexer(&self) -> &Lexer {	1✔
139	&self.lexer	1✔
140	}	1✔
141
142	/// Lookahead `k` [Token]s.
143	///
144	/// If `k == 0` then this is effectively peeking at the next [Token] from the wrapped [Lexer].
UNCOV 145	pub fn lookahead(&mut self, k: usize) -> Option<&Token> {	×
UNCOV 146	while self.lookahead.len() <= k {	×
UNCOV 147	self.lookahead.push_back(self.lexer.next_token()?);	×
148	}
149
UNCOV 150	self.lookahead.get(k)	×
UNCOV 151	}	×
152
153	/// Similar to [Parser::lookahead] but instead returns a slice of `n` [Token]s, starting with the next [Token].
154	///
155	/// Returns [None] if `n` is greater than the number of remaining [Token]s for this [Parser].
156	pub fn lookahead_window(&mut self, n: usize) -> Option<&[Token]> {	59✔
157	while self.lookahead.len() < n {	92✔
158	self.lookahead.push_back(self.lexer.next_token()?);	63✔
159	}
160
161	// Use make contiguous here to get a unified/single slice.
162	Some(&self.lookahead.make_contiguous()[..n])	29✔
163	}	59✔
164
165	/// Get the next [Token] from this parser if its [Token::variant] is the given `token_ty`.
166	pub fn next_if_is(&mut self, token_ty: TokenTy) -> Option<Token> {	50✔
167	// Peeking successfully first means that the lookahead vec will never be empty here.
168	(self.peek()?.variant == token_ty)	50✔
169	// SAFETY: We just peeked a token to check its variant so this unwrap is always ok.
170	.then(\|\| unsafe { self.lookahead.pop_front().unwrap_unchecked() })	49✔
171	}	50✔
172
173	/// Peek at the next [Token]s of this [Parser] and determine if the [Token::variant]s match this
174	/// sequence of [TokenTy]s.
175	pub fn matches(&mut self, seq: &[TokenTy]) -> bool {	59✔
176	// Use the rare let-else to ensure there are at minimum, the given number of tokens remaining.
177	let Some(lookahead_window) = self.lookahead_window(seq.len()) else {	59✔
178	return false;	30✔
179	};
180
181	// Use a zipped iterator to compare all the token variants.
182	lookahead_window	29✔
183	.iter()	29✔
184	.zip(seq)	29✔
185	.all(\|(token, matches)\| token.variant == *matches)	53✔
186	}	59✔
187	}

vcfxb / wright-lang / 15061769677

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous