From 74de79b1daeefb4868ce34e03e84949d33d3dd1e Mon Sep 17 00:00:00 2001 From: Aleksey Kladov Date: Sat, 25 Dec 2021 21:59:02 +0300 Subject: [PATCH] internal: rename --- crates/mbe/src/lib.rs | 2 +- crates/mbe/src/syntax_bridge.rs | 16 ++++---- ...to_parser_tokens.rs => to_parser_input.rs} | 4 +- crates/mbe/src/tt_iter.rs | 16 ++++---- crates/parser/src/event.rs | 6 +-- crates/parser/src/{tokens.rs => input.rs} | 22 +++++------ crates/parser/src/lexed_str.rs | 4 +- crates/parser/src/lib.rs | 26 ++++++++----- .../src/{tree_traversal.rs => output.rs} | 39 ++++++++++++------- crates/parser/src/parser.rs | 30 +++++++------- crates/syntax/src/parsing.rs | 12 +++--- crates/syntax/src/parsing/reparsing.rs | 2 +- crates/syntax/src/parsing/text_tree_sink.rs | 16 ++++---- 13 files changed, 106 insertions(+), 89 deletions(-) rename crates/mbe/src/{to_parser_tokens.rs => to_parser_input.rs} (97%) rename crates/parser/src/{tokens.rs => input.rs} (84%) rename crates/parser/src/{tree_traversal.rs => output.rs} (57%) diff --git a/crates/mbe/src/lib.rs b/crates/mbe/src/lib.rs index 1a56878fdb5..5e14a3fb590 100644 --- a/crates/mbe/src/lib.rs +++ b/crates/mbe/src/lib.rs @@ -10,7 +10,7 @@ mod parser; mod expander; mod syntax_bridge; mod tt_iter; -mod to_parser_tokens; +mod to_parser_input; #[cfg(test)] mod benchmark; diff --git a/crates/mbe/src/syntax_bridge.rs b/crates/mbe/src/syntax_bridge.rs index 39129b03056..f0c1f806ffa 100644 --- a/crates/mbe/src/syntax_bridge.rs +++ b/crates/mbe/src/syntax_bridge.rs @@ -10,7 +10,7 @@ use syntax::{ use tt::buffer::{Cursor, TokenBuffer}; use crate::{ - to_parser_tokens::to_parser_tokens, tt_iter::TtIter, ExpandError, ParserEntryPoint, TokenMap, + to_parser_input::to_parser_input, tt_iter::TtIter, ExpandError, ParserEntryPoint, TokenMap, }; /// Convert the syntax node to a `TokenTree` (what macro @@ -54,17 +54,17 @@ pub fn token_tree_to_syntax_node( } _ => TokenBuffer::from_subtree(tt), }; - let parser_tokens = to_parser_tokens(&buffer); - let tree_traversal = parser::parse(&parser_tokens, entry_point); + let parser_input = to_parser_input(&buffer); + let parser_output = parser::parse(&parser_input, entry_point); let mut tree_sink = TtTreeSink::new(buffer.begin()); - for event in tree_traversal.iter() { + for event in parser_output.iter() { match event { - parser::TraversalStep::Token { kind, n_raw_tokens } => { + parser::Step::Token { kind, n_input_tokens: n_raw_tokens } => { tree_sink.token(kind, n_raw_tokens) } - parser::TraversalStep::EnterNode { kind } => tree_sink.start_node(kind), - parser::TraversalStep::LeaveNode => tree_sink.finish_node(), - parser::TraversalStep::Error { msg } => tree_sink.error(msg.to_string()), + parser::Step::Enter { kind } => tree_sink.start_node(kind), + parser::Step::Exit => tree_sink.finish_node(), + parser::Step::Error { msg } => tree_sink.error(msg.to_string()), } } if tree_sink.roots.len() != 1 { diff --git a/crates/mbe/src/to_parser_tokens.rs b/crates/mbe/src/to_parser_input.rs similarity index 97% rename from crates/mbe/src/to_parser_tokens.rs rename to crates/mbe/src/to_parser_input.rs index f419c78d46c..6faa147218e 100644 --- a/crates/mbe/src/to_parser_tokens.rs +++ b/crates/mbe/src/to_parser_input.rs @@ -4,8 +4,8 @@ use syntax::{SyntaxKind, SyntaxKind::*, T}; use tt::buffer::TokenBuffer; -pub(crate) fn to_parser_tokens(buffer: &TokenBuffer) -> parser::Tokens { - let mut res = parser::Tokens::default(); +pub(crate) fn to_parser_input(buffer: &TokenBuffer) -> parser::Input { + let mut res = parser::Input::default(); let mut current = buffer.begin(); diff --git a/crates/mbe/src/tt_iter.rs b/crates/mbe/src/tt_iter.rs index 632b591f65e..2d2dbd8994f 100644 --- a/crates/mbe/src/tt_iter.rs +++ b/crates/mbe/src/tt_iter.rs @@ -1,7 +1,7 @@ //! A "Parser" structure for token trees. We use this when parsing a declarative //! macro definition into a list of patterns and templates. -use crate::{to_parser_tokens::to_parser_tokens, ExpandError, ExpandResult, ParserEntryPoint}; +use crate::{to_parser_input::to_parser_input, ExpandError, ExpandResult, ParserEntryPoint}; use syntax::SyntaxKind; use tt::buffer::TokenBuffer; @@ -94,23 +94,23 @@ impl<'a> TtIter<'a> { entry_point: ParserEntryPoint, ) -> ExpandResult> { let buffer = TokenBuffer::from_tokens(self.inner.as_slice()); - let parser_tokens = to_parser_tokens(&buffer); - let tree_traversal = parser::parse(&parser_tokens, entry_point); + let parser_input = to_parser_input(&buffer); + let tree_traversal = parser::parse(&parser_input, entry_point); let mut cursor = buffer.begin(); let mut error = false; for step in tree_traversal.iter() { match step { - parser::TraversalStep::Token { kind, mut n_raw_tokens } => { + parser::Step::Token { kind, mut n_input_tokens } => { if kind == SyntaxKind::LIFETIME_IDENT { - n_raw_tokens = 2; + n_input_tokens = 2; } - for _ in 0..n_raw_tokens { + for _ in 0..n_input_tokens { cursor = cursor.bump_subtree(); } } - parser::TraversalStep::EnterNode { .. } | parser::TraversalStep::LeaveNode => (), - parser::TraversalStep::Error { .. } => error = true, + parser::Step::Enter { .. } | parser::Step::Exit => (), + parser::Step::Error { .. } => error = true, } } diff --git a/crates/parser/src/event.rs b/crates/parser/src/event.rs index ca4c38f2e6c..b0e70e79430 100644 --- a/crates/parser/src/event.rs +++ b/crates/parser/src/event.rs @@ -10,7 +10,7 @@ use std::mem; use crate::{ - tree_traversal::TreeTraversal, + output::Output, SyntaxKind::{self, *}, }; @@ -87,8 +87,8 @@ impl Event { } /// Generate the syntax tree with the control of events. -pub(super) fn process(mut events: Vec) -> TreeTraversal { - let mut res = TreeTraversal::default(); +pub(super) fn process(mut events: Vec) -> Output { + let mut res = Output::default(); let mut forward_parents = Vec::new(); for i in 0..events.len() { diff --git a/crates/parser/src/tokens.rs b/crates/parser/src/input.rs similarity index 84% rename from crates/parser/src/tokens.rs rename to crates/parser/src/input.rs index 4fc2361add2..9504bd4d9ec 100644 --- a/crates/parser/src/tokens.rs +++ b/crates/parser/src/input.rs @@ -1,26 +1,26 @@ -//! Input for the parser -- a sequence of tokens. -//! -//! As of now, parser doesn't have access to the *text* of the tokens, and makes -//! decisions based solely on their classification. Unlike `LexerToken`, the -//! `Tokens` doesn't include whitespace and comments. +//! See [`Input`]. use crate::SyntaxKind; #[allow(non_camel_case_types)] type bits = u64; -/// Main input to the parser. +/// Input for the parser -- a sequence of tokens. /// -/// A sequence of tokens represented internally as a struct of arrays. +/// As of now, parser doesn't have access to the *text* of the tokens, and makes +/// decisions based solely on their classification. Unlike `LexerToken`, the +/// `Tokens` doesn't include whitespace and comments. Main input to the parser. +/// +/// Struct of arrays internally, but this shouldn't really matter. #[derive(Default)] -pub struct Tokens { +pub struct Input { kind: Vec, joint: Vec, contextual_kind: Vec, } /// `pub` impl used by callers to create `Tokens`. -impl Tokens { +impl Input { #[inline] pub fn push(&mut self, kind: SyntaxKind) { self.push_impl(kind, SyntaxKind::EOF) @@ -63,7 +63,7 @@ impl Tokens { } /// pub(crate) impl used by the parser to consume `Tokens`. -impl Tokens { +impl Input { pub(crate) fn kind(&self, idx: usize) -> SyntaxKind { self.kind.get(idx).copied().unwrap_or(SyntaxKind::EOF) } @@ -76,7 +76,7 @@ impl Tokens { } } -impl Tokens { +impl Input { fn bit_index(&self, n: usize) -> (usize, usize) { let idx = n / (bits::BITS as usize); let b_idx = n % (bits::BITS as usize); diff --git a/crates/parser/src/lexed_str.rs b/crates/parser/src/lexed_str.rs index b8936c34403..f17aae1d314 100644 --- a/crates/parser/src/lexed_str.rs +++ b/crates/parser/src/lexed_str.rs @@ -122,8 +122,8 @@ impl<'a> LexedStr<'a> { self.error.iter().map(|it| (it.token as usize, it.msg.as_str())) } - pub fn to_tokens(&self) -> crate::Tokens { - let mut res = crate::Tokens::default(); + pub fn to_input(&self) -> crate::Input { + let mut res = crate::Input::default(); let mut was_joint = false; for i in 0..self.len() { let kind = self.kind(i); diff --git a/crates/parser/src/lib.rs b/crates/parser/src/lib.rs index 67bc7d99065..da78889f350 100644 --- a/crates/parser/src/lib.rs +++ b/crates/parser/src/lib.rs @@ -24,8 +24,8 @@ mod syntax_kind; mod event; mod parser; mod grammar; -mod tokens; -mod tree_traversal; +mod input; +mod output; #[cfg(test)] mod tests; @@ -33,10 +33,10 @@ mod tests; pub(crate) use token_set::TokenSet; pub use crate::{ + input::Input, lexed_str::LexedStr, + output::{Output, Step}, syntax_kind::SyntaxKind, - tokens::Tokens, - tree_traversal::{TraversalStep, TreeTraversal}, }; /// rust-analyzer parser allows you to choose one of the possible entry points. @@ -62,11 +62,19 @@ pub enum ParserEntryPoint { } /// Parse given tokens into the given sink as a rust file. -pub fn parse_source_file(tokens: &Tokens) -> TreeTraversal { - parse(tokens, ParserEntryPoint::SourceFile) +pub fn parse_source_file(inp: &Input) -> Output { + parse(inp, ParserEntryPoint::SourceFile) } -pub fn parse(tokens: &Tokens, entry_point: ParserEntryPoint) -> TreeTraversal { +/// Parses the given [`Input`] into [`Output`] assuming that the top-level +/// syntactic construct is the given [`ParserEntryPoint`]. +/// +/// Both input and output here are fairly abstract. The overall flow is that the +/// caller has some "real" tokens, converts them to [`Input`], parses them to +/// [`Output`], and then converts that into a "real" tree. The "real" tree is +/// made of "real" tokens, so this all hinges on rather tight coordination of +/// indices between the four stages. +pub fn parse(inp: &Input, entry_point: ParserEntryPoint) -> Output { let entry_point: fn(&'_ mut parser::Parser) = match entry_point { ParserEntryPoint::SourceFile => grammar::entry_points::source_file, ParserEntryPoint::Path => grammar::entry_points::path, @@ -84,7 +92,7 @@ pub fn parse(tokens: &Tokens, entry_point: ParserEntryPoint) -> TreeTraversal { ParserEntryPoint::Attr => grammar::entry_points::attr, }; - let mut p = parser::Parser::new(tokens); + let mut p = parser::Parser::new(inp); entry_point(&mut p); let events = p.finish(); event::process(events) @@ -107,7 +115,7 @@ impl Reparser { /// /// Tokens must start with `{`, end with `}` and form a valid brace /// sequence. - pub fn parse(self, tokens: &Tokens) -> TreeTraversal { + pub fn parse(self, tokens: &Input) -> Output { let Reparser(r) = self; let mut p = parser::Parser::new(tokens); r(&mut p); diff --git a/crates/parser/src/tree_traversal.rs b/crates/parser/src/output.rs similarity index 57% rename from crates/parser/src/tree_traversal.rs rename to crates/parser/src/output.rs index 4b3a64c85c1..b613df029f8 100644 --- a/crates/parser/src/tree_traversal.rs +++ b/crates/parser/src/output.rs @@ -1,43 +1,52 @@ -//! TODO +//! See [`Output`] + use crate::SyntaxKind; -/// Output of the parser. +/// Output of the parser -- a DFS traversal of a concrete syntax tree. +/// +/// Use the [`Output::iter`] method to iterate over traversal steps and consume +/// a syntax tree. +/// +/// In a sense, this is just a sequence of [`SyntaxKind`]-colored parenthesis +/// interspersed into the original [`crate::Input`]. The output is fundamentally +/// coordinated with the input and `n_input_tokens` refers to the number of +/// times [`crate::Input::push`] was called. #[derive(Default)] -pub struct TreeTraversal { +pub struct Output { /// 32-bit encoding of events. If LSB is zero, then that's an index into the /// error vector. Otherwise, it's one of the thee other variants, with data encoded as /// - /// |16 bit kind|8 bit n_raw_tokens|4 bit tag|4 bit leftover| + /// |16 bit kind|8 bit n_input_tokens|4 bit tag|4 bit leftover| /// event: Vec, error: Vec, } -pub enum TraversalStep<'a> { - Token { kind: SyntaxKind, n_raw_tokens: u8 }, - EnterNode { kind: SyntaxKind }, - LeaveNode, +pub enum Step<'a> { + Token { kind: SyntaxKind, n_input_tokens: u8 }, + Enter { kind: SyntaxKind }, + Exit, Error { msg: &'a str }, } -impl TreeTraversal { - pub fn iter(&self) -> impl Iterator> { +impl Output { + pub fn iter(&self) -> impl Iterator> { self.event.iter().map(|&event| { if event & 0b1 == 0 { - return TraversalStep::Error { msg: self.error[(event as usize) >> 1].as_str() }; + return Step::Error { msg: self.error[(event as usize) >> 1].as_str() }; } let tag = ((event & 0x0000_00F0) >> 4) as u8; match tag { 0 => { let kind: SyntaxKind = (((event & 0xFFFF_0000) >> 16) as u16).into(); - let n_raw_tokens = ((event & 0x0000_FF00) >> 8) as u8; - TraversalStep::Token { kind, n_raw_tokens } + let n_input_tokens = ((event & 0x0000_FF00) >> 8) as u8; + Step::Token { kind, n_input_tokens } } 1 => { let kind: SyntaxKind = (((event & 0xFFFF_0000) >> 16) as u16).into(); - TraversalStep::EnterNode { kind } + Step::Enter { kind } } - 2 => TraversalStep::LeaveNode, + 2 => Step::Exit, _ => unreachable!(), } }) diff --git a/crates/parser/src/parser.rs b/crates/parser/src/parser.rs index 4fc734f9c6d..d4aecf9b446 100644 --- a/crates/parser/src/parser.rs +++ b/crates/parser/src/parser.rs @@ -7,7 +7,7 @@ use limit::Limit; use crate::{ event::Event, - tokens::Tokens, + input::Input, SyntaxKind::{self, EOF, ERROR, TOMBSTONE}, TokenSet, T, }; @@ -22,7 +22,7 @@ use crate::{ /// "start expression, consume number literal, /// finish expression". See `Event` docs for more. pub(crate) struct Parser<'t> { - tokens: &'t Tokens, + inp: &'t Input, pos: usize, events: Vec, steps: Cell, @@ -31,8 +31,8 @@ pub(crate) struct Parser<'t> { static PARSER_STEP_LIMIT: Limit = Limit::new(15_000_000); impl<'t> Parser<'t> { - pub(super) fn new(tokens: &'t Tokens) -> Parser<'t> { - Parser { tokens, pos: 0, events: Vec::new(), steps: Cell::new(0) } + pub(super) fn new(inp: &'t Input) -> Parser<'t> { + Parser { inp, pos: 0, events: Vec::new(), steps: Cell::new(0) } } pub(crate) fn finish(self) -> Vec { @@ -55,7 +55,7 @@ impl<'t> Parser<'t> { assert!(PARSER_STEP_LIMIT.check(steps as usize).is_ok(), "the parser seems stuck"); self.steps.set(steps + 1); - self.tokens.kind(self.pos + n) + self.inp.kind(self.pos + n) } /// Checks if the current token is `kind`. @@ -91,7 +91,7 @@ impl<'t> Parser<'t> { T![<<=] => self.at_composite3(n, T![<], T![<], T![=]), T![>>=] => self.at_composite3(n, T![>], T![>], T![=]), - _ => self.tokens.kind(self.pos + n) == kind, + _ => self.inp.kind(self.pos + n) == kind, } } @@ -130,17 +130,17 @@ impl<'t> Parser<'t> { } fn at_composite2(&self, n: usize, k1: SyntaxKind, k2: SyntaxKind) -> bool { - self.tokens.kind(self.pos + n) == k1 - && self.tokens.kind(self.pos + n + 1) == k2 - && self.tokens.is_joint(self.pos + n) + self.inp.kind(self.pos + n) == k1 + && self.inp.kind(self.pos + n + 1) == k2 + && self.inp.is_joint(self.pos + n) } fn at_composite3(&self, n: usize, k1: SyntaxKind, k2: SyntaxKind, k3: SyntaxKind) -> bool { - self.tokens.kind(self.pos + n) == k1 - && self.tokens.kind(self.pos + n + 1) == k2 - && self.tokens.kind(self.pos + n + 2) == k3 - && self.tokens.is_joint(self.pos + n) - && self.tokens.is_joint(self.pos + n + 1) + self.inp.kind(self.pos + n) == k1 + && self.inp.kind(self.pos + n + 1) == k2 + && self.inp.kind(self.pos + n + 2) == k3 + && self.inp.is_joint(self.pos + n) + && self.inp.is_joint(self.pos + n + 1) } /// Checks if the current token is in `kinds`. @@ -150,7 +150,7 @@ impl<'t> Parser<'t> { /// Checks if the current token is contextual keyword with text `t`. pub(crate) fn at_contextual_kw(&self, kw: SyntaxKind) -> bool { - self.tokens.contextual_kind(self.pos) == kw + self.inp.contextual_kind(self.pos) == kw } /// Starts a new node in the syntax tree. All nodes and tokens diff --git a/crates/syntax/src/parsing.rs b/crates/syntax/src/parsing.rs index 6721e5aa81a..20c7101a032 100644 --- a/crates/syntax/src/parsing.rs +++ b/crates/syntax/src/parsing.rs @@ -12,9 +12,9 @@ pub(crate) use crate::parsing::reparsing::incremental_reparse; pub(crate) fn parse_text(text: &str) -> (GreenNode, Vec) { let lexed = parser::LexedStr::new(text); - let parser_tokens = lexed.to_tokens(); - let tree_traversal = parser::parse_source_file(&parser_tokens); - let (node, errors, _eof) = build_tree(lexed, tree_traversal, false); + let parser_input = lexed.to_input(); + let parser_output = parser::parse_source_file(&parser_input); + let (node, errors, _eof) = build_tree(lexed, parser_output, false); (node, errors) } @@ -27,9 +27,9 @@ pub(crate) fn parse_text_as( if lexed.errors().next().is_some() { return Err(()); } - let parser_tokens = lexed.to_tokens(); - let tree_traversal = parser::parse(&parser_tokens, entry_point); - let (node, errors, eof) = build_tree(lexed, tree_traversal, true); + let parser_input = lexed.to_input(); + let parser_output = parser::parse(&parser_input, entry_point); + let (node, errors, eof) = build_tree(lexed, parser_output, true); if !errors.is_empty() || !eof { return Err(()); diff --git a/crates/syntax/src/parsing/reparsing.rs b/crates/syntax/src/parsing/reparsing.rs index dca97a7f583..bbe7209e726 100644 --- a/crates/syntax/src/parsing/reparsing.rs +++ b/crates/syntax/src/parsing/reparsing.rs @@ -89,7 +89,7 @@ fn reparse_block( let text = get_text_after_edit(node.clone().into(), edit); let lexed = parser::LexedStr::new(text.as_str()); - let parser_tokens = lexed.to_tokens(); + let parser_input = lexed.to_input(); if !is_balanced(&lexed) { return None; } diff --git a/crates/syntax/src/parsing/text_tree_sink.rs b/crates/syntax/src/parsing/text_tree_sink.rs index c435791746b..a5bc3f58dc6 100644 --- a/crates/syntax/src/parsing/text_tree_sink.rs +++ b/crates/syntax/src/parsing/text_tree_sink.rs @@ -2,7 +2,7 @@ use std::mem; -use parser::{LexedStr, TreeTraversal}; +use parser::LexedStr; use crate::{ ast, @@ -14,7 +14,7 @@ use crate::{ pub(crate) fn build_tree( lexed: LexedStr<'_>, - tree_traversal: TreeTraversal, + parser_output: parser::Output, synthetic_root: bool, ) -> (GreenNode, Vec, bool) { let mut builder = TextTreeSink::new(lexed); @@ -23,14 +23,14 @@ pub(crate) fn build_tree( builder.start_node(SyntaxKind::SOURCE_FILE); } - for event in tree_traversal.iter() { + for event in parser_output.iter() { match event { - parser::TraversalStep::Token { kind, n_raw_tokens } => { + parser::Step::Token { kind, n_input_tokens: n_raw_tokens } => { builder.token(kind, n_raw_tokens) } - parser::TraversalStep::EnterNode { kind } => builder.start_node(kind), - parser::TraversalStep::LeaveNode => builder.finish_node(), - parser::TraversalStep::Error { msg } => { + parser::Step::Enter { kind } => builder.start_node(kind), + parser::Step::Exit => builder.finish_node(), + parser::Step::Error { msg } => { let text_pos = builder.lexed.text_start(builder.pos).try_into().unwrap(); builder.inner.error(msg.to_string(), text_pos); } @@ -45,7 +45,7 @@ pub(crate) fn build_tree( /// Bridges the parser with our specific syntax tree representation. /// /// `TextTreeSink` also handles attachment of trivia (whitespace) to nodes. -pub(crate) struct TextTreeSink<'a> { +struct TextTreeSink<'a> { lexed: LexedStr<'a>, pos: usize, state: State,