internal: rename

This commit is contained in:
Aleksey Kladov 2021-12-25 21:59:02 +03:00
parent d0d05075ed
commit 74de79b1da
13 changed files with 106 additions and 89 deletions

View file

@ -10,7 +10,7 @@ mod parser;
mod expander; mod expander;
mod syntax_bridge; mod syntax_bridge;
mod tt_iter; mod tt_iter;
mod to_parser_tokens; mod to_parser_input;
#[cfg(test)] #[cfg(test)]
mod benchmark; mod benchmark;

View file

@ -10,7 +10,7 @@ use syntax::{
use tt::buffer::{Cursor, TokenBuffer}; use tt::buffer::{Cursor, TokenBuffer};
use crate::{ 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 /// Convert the syntax node to a `TokenTree` (what macro
@ -54,17 +54,17 @@ pub fn token_tree_to_syntax_node(
} }
_ => TokenBuffer::from_subtree(tt), _ => TokenBuffer::from_subtree(tt),
}; };
let parser_tokens = to_parser_tokens(&buffer); let parser_input = to_parser_input(&buffer);
let tree_traversal = parser::parse(&parser_tokens, entry_point); let parser_output = parser::parse(&parser_input, entry_point);
let mut tree_sink = TtTreeSink::new(buffer.begin()); let mut tree_sink = TtTreeSink::new(buffer.begin());
for event in tree_traversal.iter() { for event in parser_output.iter() {
match event { 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) tree_sink.token(kind, n_raw_tokens)
} }
parser::TraversalStep::EnterNode { kind } => tree_sink.start_node(kind), parser::Step::Enter { kind } => tree_sink.start_node(kind),
parser::TraversalStep::LeaveNode => tree_sink.finish_node(), parser::Step::Exit => tree_sink.finish_node(),
parser::TraversalStep::Error { msg } => tree_sink.error(msg.to_string()), parser::Step::Error { msg } => tree_sink.error(msg.to_string()),
} }
} }
if tree_sink.roots.len() != 1 { if tree_sink.roots.len() != 1 {

View file

@ -4,8 +4,8 @@
use syntax::{SyntaxKind, SyntaxKind::*, T}; use syntax::{SyntaxKind, SyntaxKind::*, T};
use tt::buffer::TokenBuffer; use tt::buffer::TokenBuffer;
pub(crate) fn to_parser_tokens(buffer: &TokenBuffer) -> parser::Tokens { pub(crate) fn to_parser_input(buffer: &TokenBuffer) -> parser::Input {
let mut res = parser::Tokens::default(); let mut res = parser::Input::default();
let mut current = buffer.begin(); let mut current = buffer.begin();

View file

@ -1,7 +1,7 @@
//! A "Parser" structure for token trees. We use this when parsing a declarative //! A "Parser" structure for token trees. We use this when parsing a declarative
//! macro definition into a list of patterns and templates. //! 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 syntax::SyntaxKind;
use tt::buffer::TokenBuffer; use tt::buffer::TokenBuffer;
@ -94,23 +94,23 @@ impl<'a> TtIter<'a> {
entry_point: ParserEntryPoint, entry_point: ParserEntryPoint,
) -> ExpandResult<Option<tt::TokenTree>> { ) -> ExpandResult<Option<tt::TokenTree>> {
let buffer = TokenBuffer::from_tokens(self.inner.as_slice()); let buffer = TokenBuffer::from_tokens(self.inner.as_slice());
let parser_tokens = to_parser_tokens(&buffer); let parser_input = to_parser_input(&buffer);
let tree_traversal = parser::parse(&parser_tokens, entry_point); let tree_traversal = parser::parse(&parser_input, entry_point);
let mut cursor = buffer.begin(); let mut cursor = buffer.begin();
let mut error = false; let mut error = false;
for step in tree_traversal.iter() { for step in tree_traversal.iter() {
match step { match step {
parser::TraversalStep::Token { kind, mut n_raw_tokens } => { parser::Step::Token { kind, mut n_input_tokens } => {
if kind == SyntaxKind::LIFETIME_IDENT { 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(); cursor = cursor.bump_subtree();
} }
} }
parser::TraversalStep::EnterNode { .. } | parser::TraversalStep::LeaveNode => (), parser::Step::Enter { .. } | parser::Step::Exit => (),
parser::TraversalStep::Error { .. } => error = true, parser::Step::Error { .. } => error = true,
} }
} }

View file

@ -10,7 +10,7 @@
use std::mem; use std::mem;
use crate::{ use crate::{
tree_traversal::TreeTraversal, output::Output,
SyntaxKind::{self, *}, SyntaxKind::{self, *},
}; };
@ -87,8 +87,8 @@ impl Event {
} }
/// Generate the syntax tree with the control of events. /// Generate the syntax tree with the control of events.
pub(super) fn process(mut events: Vec<Event>) -> TreeTraversal { pub(super) fn process(mut events: Vec<Event>) -> Output {
let mut res = TreeTraversal::default(); let mut res = Output::default();
let mut forward_parents = Vec::new(); let mut forward_parents = Vec::new();
for i in 0..events.len() { for i in 0..events.len() {

View file

@ -1,26 +1,26 @@
//! Input for the parser -- a sequence of tokens. //! See [`Input`].
//!
//! 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.
use crate::SyntaxKind; use crate::SyntaxKind;
#[allow(non_camel_case_types)] #[allow(non_camel_case_types)]
type bits = u64; 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)] #[derive(Default)]
pub struct Tokens { pub struct Input {
kind: Vec<SyntaxKind>, kind: Vec<SyntaxKind>,
joint: Vec<bits>, joint: Vec<bits>,
contextual_kind: Vec<SyntaxKind>, contextual_kind: Vec<SyntaxKind>,
} }
/// `pub` impl used by callers to create `Tokens`. /// `pub` impl used by callers to create `Tokens`.
impl Tokens { impl Input {
#[inline] #[inline]
pub fn push(&mut self, kind: SyntaxKind) { pub fn push(&mut self, kind: SyntaxKind) {
self.push_impl(kind, SyntaxKind::EOF) self.push_impl(kind, SyntaxKind::EOF)
@ -63,7 +63,7 @@ impl Tokens {
} }
/// pub(crate) impl used by the parser to consume `Tokens`. /// pub(crate) impl used by the parser to consume `Tokens`.
impl Tokens { impl Input {
pub(crate) fn kind(&self, idx: usize) -> SyntaxKind { pub(crate) fn kind(&self, idx: usize) -> SyntaxKind {
self.kind.get(idx).copied().unwrap_or(SyntaxKind::EOF) 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) { fn bit_index(&self, n: usize) -> (usize, usize) {
let idx = n / (bits::BITS as usize); let idx = n / (bits::BITS as usize);
let b_idx = n % (bits::BITS as usize); let b_idx = n % (bits::BITS as usize);

View file

@ -122,8 +122,8 @@ impl<'a> LexedStr<'a> {
self.error.iter().map(|it| (it.token as usize, it.msg.as_str())) self.error.iter().map(|it| (it.token as usize, it.msg.as_str()))
} }
pub fn to_tokens(&self) -> crate::Tokens { pub fn to_input(&self) -> crate::Input {
let mut res = crate::Tokens::default(); let mut res = crate::Input::default();
let mut was_joint = false; let mut was_joint = false;
for i in 0..self.len() { for i in 0..self.len() {
let kind = self.kind(i); let kind = self.kind(i);

View file

@ -24,8 +24,8 @@ mod syntax_kind;
mod event; mod event;
mod parser; mod parser;
mod grammar; mod grammar;
mod tokens; mod input;
mod tree_traversal; mod output;
#[cfg(test)] #[cfg(test)]
mod tests; mod tests;
@ -33,10 +33,10 @@ mod tests;
pub(crate) use token_set::TokenSet; pub(crate) use token_set::TokenSet;
pub use crate::{ pub use crate::{
input::Input,
lexed_str::LexedStr, lexed_str::LexedStr,
output::{Output, Step},
syntax_kind::SyntaxKind, syntax_kind::SyntaxKind,
tokens::Tokens,
tree_traversal::{TraversalStep, TreeTraversal},
}; };
/// rust-analyzer parser allows you to choose one of the possible entry points. /// 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. /// Parse given tokens into the given sink as a rust file.
pub fn parse_source_file(tokens: &Tokens) -> TreeTraversal { pub fn parse_source_file(inp: &Input) -> Output {
parse(tokens, ParserEntryPoint::SourceFile) 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 { let entry_point: fn(&'_ mut parser::Parser) = match entry_point {
ParserEntryPoint::SourceFile => grammar::entry_points::source_file, ParserEntryPoint::SourceFile => grammar::entry_points::source_file,
ParserEntryPoint::Path => grammar::entry_points::path, 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, ParserEntryPoint::Attr => grammar::entry_points::attr,
}; };
let mut p = parser::Parser::new(tokens); let mut p = parser::Parser::new(inp);
entry_point(&mut p); entry_point(&mut p);
let events = p.finish(); let events = p.finish();
event::process(events) event::process(events)
@ -107,7 +115,7 @@ impl Reparser {
/// ///
/// Tokens must start with `{`, end with `}` and form a valid brace /// Tokens must start with `{`, end with `}` and form a valid brace
/// sequence. /// sequence.
pub fn parse(self, tokens: &Tokens) -> TreeTraversal { pub fn parse(self, tokens: &Input) -> Output {
let Reparser(r) = self; let Reparser(r) = self;
let mut p = parser::Parser::new(tokens); let mut p = parser::Parser::new(tokens);
r(&mut p); r(&mut p);

View file

@ -1,43 +1,52 @@
//! TODO //! See [`Output`]
use crate::SyntaxKind; 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)] #[derive(Default)]
pub struct TreeTraversal { pub struct Output {
/// 32-bit encoding of events. If LSB is zero, then that's an index into the /// 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 /// 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<u32>, event: Vec<u32>,
error: Vec<String>, error: Vec<String>,
} }
pub enum TraversalStep<'a> { pub enum Step<'a> {
Token { kind: SyntaxKind, n_raw_tokens: u8 }, Token { kind: SyntaxKind, n_input_tokens: u8 },
EnterNode { kind: SyntaxKind }, Enter { kind: SyntaxKind },
LeaveNode, Exit,
Error { msg: &'a str }, Error { msg: &'a str },
} }
impl TreeTraversal { impl Output {
pub fn iter(&self) -> impl Iterator<Item = TraversalStep<'_>> { pub fn iter(&self) -> impl Iterator<Item = Step<'_>> {
self.event.iter().map(|&event| { self.event.iter().map(|&event| {
if event & 0b1 == 0 { 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; let tag = ((event & 0x0000_00F0) >> 4) as u8;
match tag { match tag {
0 => { 0 => {
let kind: SyntaxKind = (((event & 0xFFFF_0000) >> 16) as u16).into(); let kind: SyntaxKind = (((event & 0xFFFF_0000) >> 16) as u16).into();
let n_raw_tokens = ((event & 0x0000_FF00) >> 8) as u8; let n_input_tokens = ((event & 0x0000_FF00) >> 8) as u8;
TraversalStep::Token { kind, n_raw_tokens } Step::Token { kind, n_input_tokens }
} }
1 => { 1 => {
let kind: SyntaxKind = (((event & 0xFFFF_0000) >> 16) as u16).into(); let kind: SyntaxKind = (((event & 0xFFFF_0000) >> 16) as u16).into();
TraversalStep::EnterNode { kind } Step::Enter { kind }
} }
2 => TraversalStep::LeaveNode, 2 => Step::Exit,
_ => unreachable!(), _ => unreachable!(),
} }
}) })

View file

@ -7,7 +7,7 @@ use limit::Limit;
use crate::{ use crate::{
event::Event, event::Event,
tokens::Tokens, input::Input,
SyntaxKind::{self, EOF, ERROR, TOMBSTONE}, SyntaxKind::{self, EOF, ERROR, TOMBSTONE},
TokenSet, T, TokenSet, T,
}; };
@ -22,7 +22,7 @@ use crate::{
/// "start expression, consume number literal, /// "start expression, consume number literal,
/// finish expression". See `Event` docs for more. /// finish expression". See `Event` docs for more.
pub(crate) struct Parser<'t> { pub(crate) struct Parser<'t> {
tokens: &'t Tokens, inp: &'t Input,
pos: usize, pos: usize,
events: Vec<Event>, events: Vec<Event>,
steps: Cell<u32>, steps: Cell<u32>,
@ -31,8 +31,8 @@ pub(crate) struct Parser<'t> {
static PARSER_STEP_LIMIT: Limit = Limit::new(15_000_000); static PARSER_STEP_LIMIT: Limit = Limit::new(15_000_000);
impl<'t> Parser<'t> { impl<'t> Parser<'t> {
pub(super) fn new(tokens: &'t Tokens) -> Parser<'t> { pub(super) fn new(inp: &'t Input) -> Parser<'t> {
Parser { tokens, pos: 0, events: Vec::new(), steps: Cell::new(0) } Parser { inp, pos: 0, events: Vec::new(), steps: Cell::new(0) }
} }
pub(crate) fn finish(self) -> Vec<Event> { pub(crate) fn finish(self) -> Vec<Event> {
@ -55,7 +55,7 @@ impl<'t> Parser<'t> {
assert!(PARSER_STEP_LIMIT.check(steps as usize).is_ok(), "the parser seems stuck"); assert!(PARSER_STEP_LIMIT.check(steps as usize).is_ok(), "the parser seems stuck");
self.steps.set(steps + 1); self.steps.set(steps + 1);
self.tokens.kind(self.pos + n) self.inp.kind(self.pos + n)
} }
/// Checks if the current token is `kind`. /// 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![=]),
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 { fn at_composite2(&self, n: usize, k1: SyntaxKind, k2: SyntaxKind) -> bool {
self.tokens.kind(self.pos + n) == k1 self.inp.kind(self.pos + n) == k1
&& self.tokens.kind(self.pos + n + 1) == k2 && self.inp.kind(self.pos + n + 1) == k2
&& self.tokens.is_joint(self.pos + n) && self.inp.is_joint(self.pos + n)
} }
fn at_composite3(&self, n: usize, k1: SyntaxKind, k2: SyntaxKind, k3: SyntaxKind) -> bool { fn at_composite3(&self, n: usize, k1: SyntaxKind, k2: SyntaxKind, k3: SyntaxKind) -> bool {
self.tokens.kind(self.pos + n) == k1 self.inp.kind(self.pos + n) == k1
&& self.tokens.kind(self.pos + n + 1) == k2 && self.inp.kind(self.pos + n + 1) == k2
&& self.tokens.kind(self.pos + n + 2) == k3 && self.inp.kind(self.pos + n + 2) == k3
&& self.tokens.is_joint(self.pos + n) && self.inp.is_joint(self.pos + n)
&& self.tokens.is_joint(self.pos + n + 1) && self.inp.is_joint(self.pos + n + 1)
} }
/// Checks if the current token is in `kinds`. /// 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`. /// Checks if the current token is contextual keyword with text `t`.
pub(crate) fn at_contextual_kw(&self, kw: SyntaxKind) -> bool { 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 /// Starts a new node in the syntax tree. All nodes and tokens

View file

@ -12,9 +12,9 @@ pub(crate) use crate::parsing::reparsing::incremental_reparse;
pub(crate) fn parse_text(text: &str) -> (GreenNode, Vec<SyntaxError>) { pub(crate) fn parse_text(text: &str) -> (GreenNode, Vec<SyntaxError>) {
let lexed = parser::LexedStr::new(text); let lexed = parser::LexedStr::new(text);
let parser_tokens = lexed.to_tokens(); let parser_input = lexed.to_input();
let tree_traversal = parser::parse_source_file(&parser_tokens); let parser_output = parser::parse_source_file(&parser_input);
let (node, errors, _eof) = build_tree(lexed, tree_traversal, false); let (node, errors, _eof) = build_tree(lexed, parser_output, false);
(node, errors) (node, errors)
} }
@ -27,9 +27,9 @@ pub(crate) fn parse_text_as<T: AstNode>(
if lexed.errors().next().is_some() { if lexed.errors().next().is_some() {
return Err(()); return Err(());
} }
let parser_tokens = lexed.to_tokens(); let parser_input = lexed.to_input();
let tree_traversal = parser::parse(&parser_tokens, entry_point); let parser_output = parser::parse(&parser_input, entry_point);
let (node, errors, eof) = build_tree(lexed, tree_traversal, true); let (node, errors, eof) = build_tree(lexed, parser_output, true);
if !errors.is_empty() || !eof { if !errors.is_empty() || !eof {
return Err(()); return Err(());

View file

@ -89,7 +89,7 @@ fn reparse_block(
let text = get_text_after_edit(node.clone().into(), edit); let text = get_text_after_edit(node.clone().into(), edit);
let lexed = parser::LexedStr::new(text.as_str()); let lexed = parser::LexedStr::new(text.as_str());
let parser_tokens = lexed.to_tokens(); let parser_input = lexed.to_input();
if !is_balanced(&lexed) { if !is_balanced(&lexed) {
return None; return None;
} }

View file

@ -2,7 +2,7 @@
use std::mem; use std::mem;
use parser::{LexedStr, TreeTraversal}; use parser::LexedStr;
use crate::{ use crate::{
ast, ast,
@ -14,7 +14,7 @@ use crate::{
pub(crate) fn build_tree( pub(crate) fn build_tree(
lexed: LexedStr<'_>, lexed: LexedStr<'_>,
tree_traversal: TreeTraversal, parser_output: parser::Output,
synthetic_root: bool, synthetic_root: bool,
) -> (GreenNode, Vec<SyntaxError>, bool) { ) -> (GreenNode, Vec<SyntaxError>, bool) {
let mut builder = TextTreeSink::new(lexed); let mut builder = TextTreeSink::new(lexed);
@ -23,14 +23,14 @@ pub(crate) fn build_tree(
builder.start_node(SyntaxKind::SOURCE_FILE); builder.start_node(SyntaxKind::SOURCE_FILE);
} }
for event in tree_traversal.iter() { for event in parser_output.iter() {
match event { 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) builder.token(kind, n_raw_tokens)
} }
parser::TraversalStep::EnterNode { kind } => builder.start_node(kind), parser::Step::Enter { kind } => builder.start_node(kind),
parser::TraversalStep::LeaveNode => builder.finish_node(), parser::Step::Exit => builder.finish_node(),
parser::TraversalStep::Error { msg } => { parser::Step::Error { msg } => {
let text_pos = builder.lexed.text_start(builder.pos).try_into().unwrap(); let text_pos = builder.lexed.text_start(builder.pos).try_into().unwrap();
builder.inner.error(msg.to_string(), text_pos); 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. /// Bridges the parser with our specific syntax tree representation.
/// ///
/// `TextTreeSink` also handles attachment of trivia (whitespace) to nodes. /// `TextTreeSink` also handles attachment of trivia (whitespace) to nodes.
pub(crate) struct TextTreeSink<'a> { struct TextTreeSink<'a> {
lexed: LexedStr<'a>, lexed: LexedStr<'a>,
pos: usize, pos: usize,
state: State, state: State,