1328: Change TokenSource to iteration based r=matklad a=edwin0cheng

This PR change the `TokenSource` trait from random access to be an iteration based trait:
```rust

/// `TokenSource` abstracts the source of the tokens parser operates one.
///
/// Hopefully this will allow us to treat text and token trees in the same way!
pub trait TokenSource {
    fn current(&self) -> Token;

    /// Lookahead n token
    fn lookahead_nth(&self, n: usize) -> Token;

    /// bump cursor to next token
    fn bump(&mut self);

    /// Is the current token a specified keyword?
    fn is_keyword(&self, kw: &str) -> bool;
}

/// `TokenCursor` abstracts the cursor of `TokenSource` operates one.
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
pub struct Token {
    /// What is the current token?
    pub kind: SyntaxKind,

    /// Is the current token joined to the next one (`> >` vs `>>`).
    pub is_jointed_to_next: bool,
}
```

Note that the refactoring based on this new trait will be separated to incoming PRs

Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
This commit is contained in:
bors[bot] 2019-05-27 07:28:13 +00:00
commit ce694ae118
8 changed files with 169 additions and 129 deletions

View file

@ -68,13 +68,13 @@ impl<'a> Parser<'a> {
fn parse<F>(self, f: F) -> Option<tt::TokenTree> fn parse<F>(self, f: F) -> Option<tt::TokenTree>
where where
F: FnOnce(&dyn TokenSource, &mut dyn TreeSink), F: FnOnce(&mut dyn TokenSource, &mut dyn TreeSink),
{ {
let buffer = TokenBuffer::new(&self.subtree.token_trees[*self.cur_pos..]); let buffer = TokenBuffer::new(&self.subtree.token_trees[*self.cur_pos..]);
let mut src = SubtreeTokenSource::new(&buffer); let mut src = SubtreeTokenSource::new(&buffer);
let mut sink = OffsetTokenSink { token_pos: 0, error: false }; let mut sink = OffsetTokenSink { token_pos: 0, error: false };
f(&src, &mut sink); f(&mut src, &mut sink);
let r = self.finish(sink.token_pos, &mut src); let r = self.finish(sink.token_pos, &mut src);
if sink.error { if sink.error {

View file

@ -1,6 +1,7 @@
use ra_parser::{TokenSource}; use ra_parser::{TokenSource, Token};
use ra_syntax::{classify_literal, SmolStr, SyntaxKind, SyntaxKind::*, T}; use ra_syntax::{classify_literal, SmolStr, SyntaxKind, SyntaxKind::*, T};
use std::cell::{RefCell, Cell}; use std::cell::{RefCell, Cell};
use std::sync::Arc;
use tt::buffer::{TokenBuffer, Cursor}; use tt::buffer::{TokenBuffer, Cursor};
pub(crate) trait Querier { pub(crate) trait Querier {
@ -65,7 +66,7 @@ impl<'a> SubtreeWalk<'a> {
return cached[pos].clone(); return cached[pos].clone();
} }
fn collect_token_trees(&mut self, n: usize) -> Vec<tt::TokenTree> { fn collect_token_trees(&self, n: usize) -> Vec<tt::TokenTree> {
let mut res = vec![]; let mut res = vec![];
let mut pos = 0; let mut pos = 0;
@ -117,43 +118,59 @@ impl<'a> Querier for SubtreeWalk<'a> {
} }
pub(crate) struct SubtreeTokenSource<'a> { pub(crate) struct SubtreeTokenSource<'a> {
walker: SubtreeWalk<'a>, walker: Arc<SubtreeWalk<'a>>,
curr: (Token, usize),
} }
impl<'a> SubtreeTokenSource<'a> { impl<'a> SubtreeTokenSource<'a> {
pub fn new(buffer: &'a TokenBuffer) -> SubtreeTokenSource<'a> { pub fn new(buffer: &'a TokenBuffer) -> SubtreeTokenSource<'a> {
SubtreeTokenSource { walker: SubtreeWalk::new(buffer.begin()) } let mut res = SubtreeTokenSource {
walker: Arc::new(SubtreeWalk::new(buffer.begin())),
curr: (Token { kind: EOF, is_jointed_to_next: false }, 0),
};
res.curr = (res.mk_token(0), 0);
res
} }
pub fn querier<'b>(&'a self) -> &'b SubtreeWalk<'a> pub fn querier(&self) -> Arc<SubtreeWalk<'a>> {
where self.walker.clone()
'a: 'b,
{
&self.walker
} }
pub(crate) fn bump_n(&mut self, parsed_tokens: usize) -> Vec<tt::TokenTree> { pub(crate) fn bump_n(&mut self, parsed_tokens: usize) -> Vec<tt::TokenTree> {
let res = self.walker.collect_token_trees(parsed_tokens); let res = self.walker.collect_token_trees(parsed_tokens);
res res
} }
fn mk_token(&self, pos: usize) -> Token {
match self.walker.get(pos) {
Some(tt) => Token { kind: tt.kind, is_jointed_to_next: tt.is_joint_to_next },
None => Token { kind: EOF, is_jointed_to_next: false },
}
}
} }
impl<'a> TokenSource for SubtreeTokenSource<'a> { impl<'a> TokenSource for SubtreeTokenSource<'a> {
fn token_kind(&self, pos: usize) -> SyntaxKind { fn current(&self) -> Token {
if let Some(tok) = self.walker.get(pos) { self.curr.0
tok.kind
} else {
SyntaxKind::EOF
}
} }
fn is_token_joint_to_next(&self, pos: usize) -> bool {
match self.walker.get(pos) { /// Lookahead n token
Some(t) => t.is_joint_to_next, fn lookahead_nth(&self, n: usize) -> Token {
_ => false, self.mk_token(self.curr.1 + n)
}
} }
fn is_keyword(&self, pos: usize, kw: &str) -> bool {
match self.walker.get(pos) { /// bump cursor to next token
fn bump(&mut self) {
if self.current().kind == EOF {
return;
}
self.curr = (self.mk_token(self.curr.1 + 1), self.curr.1 + 1)
}
/// Is the current token a specified keyword?
fn is_keyword(&self, kw: &str) -> bool {
match self.walker.get(self.curr.1) {
Some(t) => t.text == *kw, Some(t) => t.text == *kw,
_ => false, _ => false,
} }

View file

@ -45,17 +45,25 @@ pub fn syntax_node_to_token_tree(node: &SyntaxNode) -> Option<(tt::Subtree, Toke
// //
// //
/// Parses the token tree (result of macro expansion) to an expression fn token_tree_to_syntax_node<F>(tt: &tt::Subtree, f: F) -> Result<TreeArc<SyntaxNode>, ExpandError>
pub fn token_tree_to_expr(tt: &tt::Subtree) -> Result<TreeArc<ast::Expr>, ExpandError> { where
F: Fn(&mut ra_parser::TokenSource, &mut ra_parser::TreeSink),
{
let buffer = tt::buffer::TokenBuffer::new(&[tt.clone().into()]); let buffer = tt::buffer::TokenBuffer::new(&[tt.clone().into()]);
let token_source = SubtreeTokenSource::new(&buffer); let mut token_source = SubtreeTokenSource::new(&buffer);
let mut tree_sink = TtTreeSink::new(token_source.querier()); let querier = token_source.querier();
ra_parser::parse_expr(&token_source, &mut tree_sink); let mut tree_sink = TtTreeSink::new(querier.as_ref());
f(&mut token_source, &mut tree_sink);
if tree_sink.roots.len() != 1 { if tree_sink.roots.len() != 1 {
return Err(ExpandError::ConversionError); return Err(ExpandError::ConversionError);
} }
let syntax = tree_sink.inner.finish(); Ok(tree_sink.inner.finish())
}
/// Parses the token tree (result of macro expansion) to an expression
pub fn token_tree_to_expr(tt: &tt::Subtree) -> Result<TreeArc<ast::Expr>, ExpandError> {
let syntax = token_tree_to_syntax_node(tt, ra_parser::parse_expr)?;
ast::Expr::cast(&syntax) ast::Expr::cast(&syntax)
.map(|m| m.to_owned()) .map(|m| m.to_owned())
.ok_or_else(|| crate::ExpandError::ConversionError) .ok_or_else(|| crate::ExpandError::ConversionError)
@ -63,28 +71,13 @@ pub fn token_tree_to_expr(tt: &tt::Subtree) -> Result<TreeArc<ast::Expr>, Expand
/// Parses the token tree (result of macro expansion) to a Pattern /// Parses the token tree (result of macro expansion) to a Pattern
pub fn token_tree_to_pat(tt: &tt::Subtree) -> Result<TreeArc<ast::Pat>, ExpandError> { pub fn token_tree_to_pat(tt: &tt::Subtree) -> Result<TreeArc<ast::Pat>, ExpandError> {
let buffer = tt::buffer::TokenBuffer::new(&[tt.clone().into()]); let syntax = token_tree_to_syntax_node(tt, ra_parser::parse_pat)?;
let token_source = SubtreeTokenSource::new(&buffer);
let mut tree_sink = TtTreeSink::new(token_source.querier());
ra_parser::parse_pat(&token_source, &mut tree_sink);
if tree_sink.roots.len() != 1 {
return Err(ExpandError::ConversionError);
}
let syntax = tree_sink.inner.finish();
ast::Pat::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError) ast::Pat::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError)
} }
/// Parses the token tree (result of macro expansion) to a Type /// Parses the token tree (result of macro expansion) to a Type
pub fn token_tree_to_ty(tt: &tt::Subtree) -> Result<TreeArc<ast::TypeRef>, ExpandError> { pub fn token_tree_to_ty(tt: &tt::Subtree) -> Result<TreeArc<ast::TypeRef>, ExpandError> {
let buffer = tt::buffer::TokenBuffer::new(&[tt.clone().into()]); let syntax = token_tree_to_syntax_node(tt, ra_parser::parse_ty)?;
let token_source = SubtreeTokenSource::new(&buffer);
let mut tree_sink = TtTreeSink::new(token_source.querier());
ra_parser::parse_ty(&token_source, &mut tree_sink);
if tree_sink.roots.len() != 1 {
return Err(ExpandError::ConversionError);
}
let syntax = tree_sink.inner.finish();
ast::TypeRef::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError) ast::TypeRef::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError)
} }
@ -92,14 +85,7 @@ pub fn token_tree_to_ty(tt: &tt::Subtree) -> Result<TreeArc<ast::TypeRef>, Expan
pub fn token_tree_to_macro_stmts( pub fn token_tree_to_macro_stmts(
tt: &tt::Subtree, tt: &tt::Subtree,
) -> Result<TreeArc<ast::MacroStmts>, ExpandError> { ) -> Result<TreeArc<ast::MacroStmts>, ExpandError> {
let buffer = tt::buffer::TokenBuffer::new(&[tt.clone().into()]); let syntax = token_tree_to_syntax_node(tt, ra_parser::parse_macro_stmts)?;
let token_source = SubtreeTokenSource::new(&buffer);
let mut tree_sink = TtTreeSink::new(token_source.querier());
ra_parser::parse_macro_stmts(&token_source, &mut tree_sink);
if tree_sink.roots.len() != 1 {
return Err(ExpandError::ConversionError);
}
let syntax = tree_sink.inner.finish();
ast::MacroStmts::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError) ast::MacroStmts::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError)
} }
@ -107,24 +93,13 @@ pub fn token_tree_to_macro_stmts(
pub fn token_tree_to_macro_items( pub fn token_tree_to_macro_items(
tt: &tt::Subtree, tt: &tt::Subtree,
) -> Result<TreeArc<ast::MacroItems>, ExpandError> { ) -> Result<TreeArc<ast::MacroItems>, ExpandError> {
let buffer = tt::buffer::TokenBuffer::new(&[tt.clone().into()]); let syntax = token_tree_to_syntax_node(tt, ra_parser::parse_macro_items)?;
let token_source = SubtreeTokenSource::new(&buffer);
let mut tree_sink = TtTreeSink::new(token_source.querier());
ra_parser::parse_macro_items(&token_source, &mut tree_sink);
if tree_sink.roots.len() != 1 {
return Err(ExpandError::ConversionError);
}
let syntax = tree_sink.inner.finish();
ast::MacroItems::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError) ast::MacroItems::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError)
} }
/// Parses the token tree (result of macro expansion) as a sequence of items /// Parses the token tree (result of macro expansion) as a sequence of items
pub fn token_tree_to_ast_item_list(tt: &tt::Subtree) -> TreeArc<ast::SourceFile> { pub fn token_tree_to_ast_item_list(tt: &tt::Subtree) -> TreeArc<ast::SourceFile> {
let buffer = tt::buffer::TokenBuffer::new(&[tt.clone().into()]); let syntax = token_tree_to_syntax_node(tt, ra_parser::parse).unwrap();
let token_source = SubtreeTokenSource::new(&buffer);
let mut tree_sink = TtTreeSink::new(token_source.querier());
ra_parser::parse(&token_source, &mut tree_sink);
let syntax = tree_sink.inner.finish();
ast::SourceFile::cast(&syntax).unwrap().to_owned() ast::SourceFile::cast(&syntax).unwrap().to_owned()
} }

View file

@ -31,12 +31,26 @@ pub struct ParseError(pub String);
/// ///
/// Hopefully this will allow us to treat text and token trees in the same way! /// Hopefully this will allow us to treat text and token trees in the same way!
pub trait TokenSource { pub trait TokenSource {
/// What is the current token? fn current(&self) -> Token;
fn token_kind(&self, pos: usize) -> SyntaxKind;
/// Is the current token joined to the next one (`> >` vs `>>`). /// Lookahead n token
fn is_token_joint_to_next(&self, pos: usize) -> bool; fn lookahead_nth(&self, n: usize) -> Token;
/// bump cursor to next token
fn bump(&mut self);
/// Is the current token a specified keyword? /// Is the current token a specified keyword?
fn is_keyword(&self, pos: usize, kw: &str) -> bool; fn is_keyword(&self, kw: &str) -> bool;
}
/// `TokenCursor` abstracts the cursor of `TokenSource` operates one.
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
pub struct Token {
/// What is the current token?
pub kind: SyntaxKind,
/// Is the current token joined to the next one (`> >` vs `>>`).
pub is_jointed_to_next: bool,
} }
/// `TreeSink` abstracts details of a particular syntax tree implementation. /// `TreeSink` abstracts details of a particular syntax tree implementation.
@ -54,7 +68,7 @@ pub trait TreeSink {
fn error(&mut self, error: ParseError); fn error(&mut self, error: ParseError);
} }
fn parse_from_tokens<F>(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink, f: F) fn parse_from_tokens<F>(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink, f: F)
where where
F: FnOnce(&mut parser::Parser), F: FnOnce(&mut parser::Parser),
{ {
@ -65,61 +79,65 @@ where
} }
/// 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(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, grammar::root); parse_from_tokens(token_source, tree_sink, grammar::root);
} }
/// Parse given tokens into the given sink as a path /// Parse given tokens into the given sink as a path
pub fn parse_path(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse_path(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, grammar::path); parse_from_tokens(token_source, tree_sink, grammar::path);
} }
/// Parse given tokens into the given sink as a expression /// Parse given tokens into the given sink as a expression
pub fn parse_expr(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse_expr(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, grammar::expr); parse_from_tokens(token_source, tree_sink, grammar::expr);
} }
/// Parse given tokens into the given sink as a ty /// Parse given tokens into the given sink as a ty
pub fn parse_ty(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse_ty(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, grammar::type_); parse_from_tokens(token_source, tree_sink, grammar::type_);
} }
/// Parse given tokens into the given sink as a pattern /// Parse given tokens into the given sink as a pattern
pub fn parse_pat(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse_pat(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, grammar::pattern); parse_from_tokens(token_source, tree_sink, grammar::pattern);
} }
/// Parse given tokens into the given sink as a statement /// Parse given tokens into the given sink as a statement
pub fn parse_stmt(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink, with_semi: bool) { pub fn parse_stmt(
token_source: &mut dyn TokenSource,
tree_sink: &mut dyn TreeSink,
with_semi: bool,
) {
parse_from_tokens(token_source, tree_sink, |p| grammar::stmt(p, with_semi)); parse_from_tokens(token_source, tree_sink, |p| grammar::stmt(p, with_semi));
} }
/// Parse given tokens into the given sink as a block /// Parse given tokens into the given sink as a block
pub fn parse_block(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse_block(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, grammar::block); parse_from_tokens(token_source, tree_sink, grammar::block);
} }
pub fn parse_meta(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse_meta(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, grammar::meta_item); parse_from_tokens(token_source, tree_sink, grammar::meta_item);
} }
/// Parse given tokens into the given sink as an item /// Parse given tokens into the given sink as an item
pub fn parse_item(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse_item(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, grammar::item); parse_from_tokens(token_source, tree_sink, grammar::item);
} }
/// Parse given tokens into the given sink as an visibility qualifier /// Parse given tokens into the given sink as an visibility qualifier
pub fn parse_vis(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse_vis(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, |p| { parse_from_tokens(token_source, tree_sink, |p| {
grammar::opt_visibility(p); grammar::opt_visibility(p);
}); });
} }
pub fn parse_macro_items(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse_macro_items(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, grammar::macro_items); parse_from_tokens(token_source, tree_sink, grammar::macro_items);
} }
pub fn parse_macro_stmts(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse_macro_stmts(token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
parse_from_tokens(token_source, tree_sink, grammar::macro_stmts); parse_from_tokens(token_source, tree_sink, grammar::macro_stmts);
} }
@ -140,7 +158,7 @@ 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, token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) { pub fn parse(self, token_source: &mut dyn TokenSource, tree_sink: &mut dyn TreeSink) {
let Reparser(r) = self; let Reparser(r) = self;
let mut p = parser::Parser::new(token_source); let mut p = parser::Parser::new(token_source);
r(&mut p); r(&mut p);

View file

@ -19,15 +19,14 @@ 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> {
token_source: &'t dyn TokenSource, token_source: &'t mut dyn TokenSource,
token_pos: usize,
events: Vec<Event>, events: Vec<Event>,
steps: Cell<u32>, steps: Cell<u32>,
} }
impl<'t> Parser<'t> { impl<'t> Parser<'t> {
pub(super) fn new(token_source: &'t dyn TokenSource) -> Parser<'t> { pub(super) fn new(token_source: &'t mut dyn TokenSource) -> Parser<'t> {
Parser { token_source, token_pos: 0, events: Vec::new(), steps: Cell::new(0) } Parser { token_source, events: Vec::new(), steps: Cell::new(0) }
} }
pub(crate) fn finish(self) -> Vec<Event> { pub(crate) fn finish(self) -> Vec<Event> {
@ -49,7 +48,7 @@ impl<'t> Parser<'t> {
let c1 = self.nth(0); let c1 = self.nth(0);
let c2 = self.nth(1); let c2 = self.nth(1);
if self.token_source.is_token_joint_to_next(self.token_pos) { if self.token_source.current().is_jointed_to_next {
Some((c1, c2)) Some((c1, c2))
} else { } else {
None None
@ -64,8 +63,8 @@ impl<'t> Parser<'t> {
let c1 = self.nth(0); let c1 = self.nth(0);
let c2 = self.nth(1); let c2 = self.nth(1);
let c3 = self.nth(2); let c3 = self.nth(2);
if self.token_source.is_token_joint_to_next(self.token_pos) if self.token_source.current().is_jointed_to_next
&& self.token_source.is_token_joint_to_next(self.token_pos + 1) && self.token_source.lookahead_nth(1).is_jointed_to_next
{ {
Some((c1, c2, c3)) Some((c1, c2, c3))
} else { } else {
@ -76,6 +75,8 @@ impl<'t> Parser<'t> {
/// Lookahead operation: returns the kind of the next nth /// Lookahead operation: returns the kind of the next nth
/// token. /// token.
pub(crate) fn nth(&self, n: usize) -> SyntaxKind { pub(crate) fn nth(&self, n: usize) -> SyntaxKind {
assert!(n <= 3);
let steps = self.steps.get(); let steps = self.steps.get();
assert!(steps <= 10_000_000, "the parser seems stuck"); assert!(steps <= 10_000_000, "the parser seems stuck");
self.steps.set(steps + 1); self.steps.set(steps + 1);
@ -86,7 +87,7 @@ impl<'t> Parser<'t> {
let mut i = 0; let mut i = 0;
loop { loop {
let mut kind = self.token_source.token_kind(self.token_pos + i); let mut kind = self.token_source.lookahead_nth(i).kind;
if let Some((composited, step)) = self.is_composite(kind, i) { if let Some((composited, step)) = self.is_composite(kind, i) {
kind = composited; kind = composited;
i += step; i += step;
@ -115,7 +116,7 @@ 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: &str) -> bool { pub(crate) fn at_contextual_kw(&self, kw: &str) -> bool {
self.token_source.is_keyword(self.token_pos, kw) self.token_source.is_keyword(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
@ -130,12 +131,12 @@ impl<'t> Parser<'t> {
/// Advances the parser by one token unconditionally /// Advances the parser by one token unconditionally
/// Mainly use in `token_tree` parsing /// Mainly use in `token_tree` parsing
pub(crate) fn bump_raw(&mut self) { pub(crate) fn bump_raw(&mut self) {
let mut kind = self.token_source.token_kind(self.token_pos); let mut kind = self.token_source.current().kind;
// Skip dollars, do_bump will eat these later // Skip dollars, do_bump will eat these later
let mut i = 0; let mut i = 0;
while kind == SyntaxKind::L_DOLLAR || kind == SyntaxKind::R_DOLLAR { while kind == SyntaxKind::L_DOLLAR || kind == SyntaxKind::R_DOLLAR {
kind = self.token_source.token_kind(self.token_pos + i); kind = self.token_source.lookahead_nth(i).kind;
i += 1; i += 1;
} }
@ -236,7 +237,11 @@ impl<'t> Parser<'t> {
fn do_bump(&mut self, kind: SyntaxKind, n_raw_tokens: u8) { fn do_bump(&mut self, kind: SyntaxKind, n_raw_tokens: u8) {
self.eat_dollars(); self.eat_dollars();
self.token_pos += usize::from(n_raw_tokens);
for _ in 0..n_raw_tokens {
self.token_source.bump();
}
self.push_event(Event::Token { kind, n_raw_tokens }); self.push_event(Event::Token { kind, n_raw_tokens });
} }
@ -249,10 +254,14 @@ impl<'t> Parser<'t> {
// We assume the dollars will not occuried between // We assume the dollars will not occuried between
// mult-byte tokens // mult-byte tokens
let jn1 = self.token_source.is_token_joint_to_next(self.token_pos + n); let first = self.token_source.lookahead_nth(n);
let la2 = self.token_source.token_kind(self.token_pos + n + 1); let second = self.token_source.lookahead_nth(n + 1);
let jn2 = self.token_source.is_token_joint_to_next(self.token_pos + n + 1); let third = self.token_source.lookahead_nth(n + 2);
let la3 = self.token_source.token_kind(self.token_pos + n + 2);
let jn1 = first.is_jointed_to_next;
let la2 = second.kind;
let jn2 = second.is_jointed_to_next;
let la3 = third.kind;
match kind { match kind {
T![.] if jn1 && la2 == T![.] && jn2 && la3 == T![.] => Some((T![...], 3)), T![.] if jn1 && la2 == T![.] && jn2 && la3 == T![.] => Some((T![...], 3)),
@ -271,9 +280,9 @@ impl<'t> Parser<'t> {
fn eat_dollars(&mut self) { fn eat_dollars(&mut self) {
loop { loop {
match self.token_source.token_kind(self.token_pos) { match self.token_source.current().kind {
k @ SyntaxKind::L_DOLLAR | k @ SyntaxKind::R_DOLLAR => { k @ SyntaxKind::L_DOLLAR | k @ SyntaxKind::R_DOLLAR => {
self.token_pos += 1; self.token_source.bump();
self.push_event(Event::Token { kind: k, n_raw_tokens: 1 }); self.push_event(Event::Token { kind: k, n_raw_tokens: 1 });
} }
_ => { _ => {
@ -286,9 +295,9 @@ impl<'t> Parser<'t> {
pub(crate) fn eat_l_dollars(&mut self) -> usize { pub(crate) fn eat_l_dollars(&mut self) -> usize {
let mut ate_count = 0; let mut ate_count = 0;
loop { loop {
match self.token_source.token_kind(self.token_pos) { match self.token_source.current().kind {
k @ SyntaxKind::L_DOLLAR => { k @ SyntaxKind::L_DOLLAR => {
self.token_pos += 1; self.token_source.bump();
self.push_event(Event::Token { kind: k, n_raw_tokens: 1 }); self.push_event(Event::Token { kind: k, n_raw_tokens: 1 });
ate_count += 1; ate_count += 1;
} }
@ -302,9 +311,9 @@ impl<'t> Parser<'t> {
pub(crate) fn eat_r_dollars(&mut self, max_count: usize) -> usize { pub(crate) fn eat_r_dollars(&mut self, max_count: usize) -> usize {
let mut ate_count = 0; let mut ate_count = 0;
loop { loop {
match self.token_source.token_kind(self.token_pos) { match self.token_source.current().kind {
k @ SyntaxKind::R_DOLLAR => { k @ SyntaxKind::R_DOLLAR => {
self.token_pos += 1; self.token_source.bump();
self.push_event(Event::Token { kind: k, n_raw_tokens: 1 }); self.push_event(Event::Token { kind: k, n_raw_tokens: 1 });
ate_count += 1; ate_count += 1;
@ -320,12 +329,12 @@ impl<'t> Parser<'t> {
} }
pub(crate) fn at_l_dollar(&self) -> bool { pub(crate) fn at_l_dollar(&self) -> bool {
let kind = self.token_source.token_kind(self.token_pos); let kind = self.token_source.current().kind;
(kind == SyntaxKind::L_DOLLAR) (kind == SyntaxKind::L_DOLLAR)
} }
pub(crate) fn at_r_dollar(&self) -> bool { pub(crate) fn at_r_dollar(&self) -> bool {
let kind = self.token_source.token_kind(self.token_pos); let kind = self.token_source.current().kind;
(kind == SyntaxKind::R_DOLLAR) (kind == SyntaxKind::R_DOLLAR)
} }
} }

View file

@ -17,8 +17,8 @@ pub(crate) use self::reparsing::incremental_reparse;
pub(crate) fn parse_text(text: &str) -> (GreenNode, Vec<SyntaxError>) { pub(crate) fn parse_text(text: &str) -> (GreenNode, Vec<SyntaxError>) {
let tokens = tokenize(&text); let tokens = tokenize(&text);
let token_source = text_token_source::TextTokenSource::new(text, &tokens); let mut token_source = text_token_source::TextTokenSource::new(text, &tokens);
let mut tree_sink = text_tree_sink::TextTreeSink::new(text, &tokens); let mut tree_sink = text_tree_sink::TextTreeSink::new(text, &tokens);
ra_parser::parse(&token_source, &mut tree_sink); ra_parser::parse(&mut token_source, &mut tree_sink);
tree_sink.finish() tree_sink.finish()
} }

View file

@ -85,9 +85,9 @@ fn reparse_block<'node>(
if !is_balanced(&tokens) { if !is_balanced(&tokens) {
return None; return None;
} }
let token_source = TextTokenSource::new(&text, &tokens); let mut token_source = TextTokenSource::new(&text, &tokens);
let mut tree_sink = TextTreeSink::new(&text, &tokens); let mut tree_sink = TextTreeSink::new(&text, &tokens);
reparser.parse(&token_source, &mut tree_sink); reparser.parse(&mut token_source, &mut tree_sink);
let (green, new_errors) = tree_sink.finish(); let (green, new_errors) = tree_sink.finish();
Some((node.replace_with(green), new_errors, node.range())) Some((node.replace_with(green), new_errors, node.range()))
} }

View file

@ -1,7 +1,8 @@
use ra_parser::TokenSource; use ra_parser::TokenSource;
use ra_parser::Token as PToken;
use crate::{ use crate::{
SyntaxKind, SyntaxKind::EOF, TextRange, TextUnit, SyntaxKind::EOF, TextRange, TextUnit,
parsing::lexer::Token, parsing::lexer::Token,
}; };
@ -23,31 +24,50 @@ pub(crate) struct TextTokenSource<'t> {
/// ``` /// ```
/// tokens: `[struct, Foo, {, }]` /// tokens: `[struct, Foo, {, }]`
tokens: Vec<Token>, tokens: Vec<Token>,
/// Current token and position
curr: (PToken, usize),
} }
impl<'t> TokenSource for TextTokenSource<'t> { impl<'t> TokenSource for TextTokenSource<'t> {
fn token_kind(&self, pos: usize) -> SyntaxKind { fn current(&self) -> PToken {
if !(pos < self.tokens.len()) { return self.curr.0;
return EOF;
}
self.tokens[pos].kind
} }
fn is_token_joint_to_next(&self, pos: usize) -> bool {
if !(pos + 1 < self.tokens.len()) { fn lookahead_nth(&self, n: usize) -> PToken {
return true; mk_token(self.curr.1 + n, &self.start_offsets, &self.tokens)
}
self.start_offsets[pos] + self.tokens[pos].len == self.start_offsets[pos + 1]
} }
fn is_keyword(&self, pos: usize, kw: &str) -> bool {
fn bump(&mut self) {
if self.curr.0.kind == EOF {
return;
}
let pos = self.curr.1 + 1;
self.curr = (mk_token(pos, &self.start_offsets, &self.tokens), pos);
}
fn is_keyword(&self, kw: &str) -> bool {
let pos = self.curr.1;
if !(pos < self.tokens.len()) { if !(pos < self.tokens.len()) {
return false; return false;
} }
let range = TextRange::offset_len(self.start_offsets[pos], self.tokens[pos].len); let range = TextRange::offset_len(self.start_offsets[pos], self.tokens[pos].len);
self.text[range] == *kw self.text[range] == *kw
} }
} }
fn mk_token(pos: usize, start_offsets: &[TextUnit], tokens: &[Token]) -> PToken {
let kind = tokens.get(pos).map(|t| t.kind).unwrap_or(EOF);
let is_jointed_to_next = if pos + 1 < start_offsets.len() {
start_offsets[pos] + tokens[pos].len == start_offsets[pos + 1]
} else {
false
};
PToken { kind, is_jointed_to_next }
}
impl<'t> TextTokenSource<'t> { impl<'t> TextTokenSource<'t> {
/// Generate input from tokens(expect comment and whitespace). /// Generate input from tokens(expect comment and whitespace).
pub fn new(text: &'t str, raw_tokens: &'t [Token]) -> TextTokenSource<'t> { pub fn new(text: &'t str, raw_tokens: &'t [Token]) -> TextTokenSource<'t> {
@ -62,6 +82,7 @@ impl<'t> TextTokenSource<'t> {
len += token.len; len += token.len;
} }
TextTokenSource { text, start_offsets, tokens } let first = mk_token(0, &start_offsets, &tokens);
TextTokenSource { text, start_offsets, tokens, curr: (first, 0) }
} }
} }