mirror of
https://github.com/rust-lang/rust.git
synced 2025-06-05 03:38:29 +00:00
commit
aa36ad008e
@ -91,6 +91,8 @@ Grammar(
|
||||
"USE_ITEM",
|
||||
"STATIC_ITEM",
|
||||
"CONST_ITEM",
|
||||
"TRAIT_ITEM",
|
||||
"IMPL_ITEM",
|
||||
|
||||
"EXTERN_BLOCK",
|
||||
"ENUM_VARIANT",
|
||||
|
140
src/parser/event.rs
Normal file
140
src/parser/event.rs
Normal file
@ -0,0 +1,140 @@
|
||||
use {File, FileBuilder, Sink, SyntaxKind, Token};
|
||||
use syntax_kinds::TOMBSTONE;
|
||||
use super::is_insignificant;
|
||||
|
||||
/// `Parser` produces a flat list of `Event`s.
|
||||
/// They are converted to a tree-structure in
|
||||
/// a separate pass, via `TreeBuilder`.
|
||||
#[derive(Debug)]
|
||||
pub(crate) enum Event {
|
||||
/// This event signifies the start of the node.
|
||||
/// It should be either abandoned (in which case the
|
||||
/// `kind` is `TOMBSTONE`, and the event is ignored),
|
||||
/// or completed via a `Finish` event.
|
||||
///
|
||||
/// All tokens between a `Start` and a `Finish` would
|
||||
/// become the children of the respective node.
|
||||
///
|
||||
/// For left-recursive syntactic constructs, the parser produces
|
||||
/// a child node before it sees a parent. `forward_parent`
|
||||
/// exists to allow to tweak parent-child relationships.
|
||||
///
|
||||
/// Consider this path
|
||||
///
|
||||
/// foo::bar
|
||||
///
|
||||
/// The events for it would look like this:
|
||||
///
|
||||
///
|
||||
/// START(PATH) IDENT('foo') FINISH START(PATH) COLONCOLON IDENT('bar') FINISH
|
||||
/// | /\
|
||||
/// | |
|
||||
/// +------forward-parent------+
|
||||
///
|
||||
/// And the tree would look like this
|
||||
///
|
||||
/// +--PATH---------+
|
||||
/// | | |
|
||||
/// | | |
|
||||
/// | '::' 'bar'
|
||||
/// |
|
||||
/// PATH
|
||||
/// |
|
||||
/// 'foo'
|
||||
///
|
||||
/// See also `CompleteMarker::precede`.
|
||||
Start {
|
||||
kind: SyntaxKind,
|
||||
forward_parent: Option<u32>,
|
||||
},
|
||||
|
||||
/// Complete the previous `Start` event
|
||||
Finish,
|
||||
|
||||
/// Produce a single leaf-element.
|
||||
/// `n_raw_tokens` is used to glue complex contextual tokens.
|
||||
/// For example, lexer tokenizes `>>` as `>`, `>`, and
|
||||
/// `n_raw_tokens = 2` is used to produced a single `>>`.
|
||||
Token {
|
||||
kind: SyntaxKind,
|
||||
n_raw_tokens: u8,
|
||||
},
|
||||
|
||||
Error {
|
||||
message: String,
|
||||
},
|
||||
}
|
||||
|
||||
pub(super) fn to_file(text: String, tokens: &[Token], events: Vec<Event>) -> File {
|
||||
let mut builder = FileBuilder::new(text);
|
||||
let mut idx = 0;
|
||||
|
||||
let mut holes = Vec::new();
|
||||
let mut forward_parents = Vec::new();
|
||||
|
||||
for (i, event) in events.iter().enumerate() {
|
||||
if holes.last() == Some(&i) {
|
||||
holes.pop();
|
||||
continue;
|
||||
}
|
||||
|
||||
match event {
|
||||
&Event::Start {
|
||||
kind: TOMBSTONE, ..
|
||||
} => (),
|
||||
|
||||
&Event::Start { .. } => {
|
||||
forward_parents.clear();
|
||||
let mut idx = i;
|
||||
loop {
|
||||
let (kind, fwd) = match events[idx] {
|
||||
Event::Start {
|
||||
kind,
|
||||
forward_parent,
|
||||
} => (kind, forward_parent),
|
||||
_ => unreachable!(),
|
||||
};
|
||||
forward_parents.push((idx, kind));
|
||||
if let Some(fwd) = fwd {
|
||||
idx += fwd as usize;
|
||||
} else {
|
||||
break;
|
||||
}
|
||||
}
|
||||
for &(idx, kind) in forward_parents.iter().into_iter().rev() {
|
||||
builder.start_internal(kind);
|
||||
holes.push(idx);
|
||||
}
|
||||
holes.pop();
|
||||
}
|
||||
&Event::Finish => {
|
||||
while idx < tokens.len() {
|
||||
let token = tokens[idx];
|
||||
if is_insignificant(token.kind) {
|
||||
idx += 1;
|
||||
builder.leaf(token.kind, token.len);
|
||||
} else {
|
||||
break;
|
||||
}
|
||||
}
|
||||
builder.finish_internal()
|
||||
}
|
||||
&Event::Token {
|
||||
kind: _,
|
||||
mut n_raw_tokens,
|
||||
} => loop {
|
||||
let token = tokens[idx];
|
||||
if !is_insignificant(token.kind) {
|
||||
n_raw_tokens -= 1;
|
||||
}
|
||||
idx += 1;
|
||||
builder.leaf(token.kind, token.len);
|
||||
if n_raw_tokens == 0 {
|
||||
break;
|
||||
}
|
||||
},
|
||||
&Event::Error { ref message } => builder.error().message(message.clone()).emit(),
|
||||
}
|
||||
}
|
||||
builder.finish()
|
||||
}
|
@ -1,74 +0,0 @@
|
||||
use {SyntaxKind, Token};
|
||||
|
||||
#[macro_use]
|
||||
mod parser;
|
||||
mod grammar;
|
||||
|
||||
/// `Parser` produces a flat list of `Event`s.
|
||||
/// They are converted to a tree-structure in
|
||||
/// a separate pass, via `TreeBuilder`.
|
||||
#[derive(Debug)]
|
||||
pub(crate) enum Event {
|
||||
/// This event signifies the start of the node.
|
||||
/// It should be either abandoned (in which case the
|
||||
/// `kind` is `TOMBSTONE`, and the event is ignored),
|
||||
/// or completed via a `Finish` event.
|
||||
///
|
||||
/// All tokens between a `Start` and a `Finish` would
|
||||
/// become the children of the respective node.
|
||||
///
|
||||
/// For left-recursive syntactic constructs, the parser produces
|
||||
/// a child node before it sees a parent. `forward_parent`
|
||||
/// exists to allow to tweak parent-child relationships.
|
||||
///
|
||||
/// Consider this path
|
||||
///
|
||||
/// foo::bar
|
||||
///
|
||||
/// The events for it would look like this:
|
||||
///
|
||||
///
|
||||
/// START(PATH) IDENT('foo') FINISH START(PATH) COLONCOLON IDENT('bar') FINISH
|
||||
/// | /\
|
||||
/// | |
|
||||
/// +------forward-parent------+
|
||||
///
|
||||
/// And the tree would look like this
|
||||
///
|
||||
/// +--PATH---------+
|
||||
/// | | |
|
||||
/// | | |
|
||||
/// | '::' 'bar'
|
||||
/// |
|
||||
/// PATH
|
||||
/// |
|
||||
/// 'foo'
|
||||
///
|
||||
/// See also `CompleteMarker::precede`.
|
||||
Start {
|
||||
kind: SyntaxKind,
|
||||
forward_parent: Option<u32>,
|
||||
},
|
||||
|
||||
/// Complete the previous `Start` event
|
||||
Finish,
|
||||
|
||||
/// Produce a single leaf-element.
|
||||
/// `n_raw_tokens` is used to glue complex contextual tokens.
|
||||
/// For example, lexer tokenizes `>>` as `>`, `>`, and
|
||||
/// `n_raw_tokens = 2` is used to produced a single `>>`.
|
||||
Token {
|
||||
kind: SyntaxKind,
|
||||
n_raw_tokens: u8,
|
||||
},
|
||||
|
||||
Error {
|
||||
message: String,
|
||||
},
|
||||
}
|
||||
|
||||
pub(crate) fn parse<'t>(text: &'t str, raw_tokens: &'t [Token]) -> Vec<Event> {
|
||||
let mut parser = parser::Parser::new(text, raw_tokens);
|
||||
grammar::file(&mut parser);
|
||||
parser.into_events()
|
||||
}
|
@ -3,6 +3,7 @@ use super::*;
|
||||
mod structs;
|
||||
mod use_item;
|
||||
mod consts;
|
||||
mod traits;
|
||||
|
||||
pub(super) fn mod_contents(p: &mut Parser, stop_on_r_curly: bool) {
|
||||
attributes::inner_attributes(p);
|
||||
@ -80,6 +81,22 @@ fn item(p: &mut Parser) {
|
||||
CONST_ITEM
|
||||
}
|
||||
},
|
||||
// TODO: auto trait
|
||||
// test unsafe_trait
|
||||
// unsafe trait T {}
|
||||
UNSAFE_KW if la == TRAIT_KW => {
|
||||
p.bump();
|
||||
traits::trait_item(p);
|
||||
TRAIT_ITEM
|
||||
}
|
||||
// TODO: default impl
|
||||
// test unsafe_impl
|
||||
// unsafe impl Foo {}
|
||||
UNSAFE_KW if la == IMPL_KW => {
|
||||
p.bump();
|
||||
traits::impl_item(p);
|
||||
IMPL_ITEM
|
||||
}
|
||||
MOD_KW => {
|
||||
mod_item(p);
|
||||
MOD_ITEM
|
||||
@ -131,6 +148,7 @@ fn extern_block(p: &mut Parser) {
|
||||
p.bump();
|
||||
p.expect(R_CURLY);
|
||||
}
|
||||
|
||||
fn mod_item(p: &mut Parser) {
|
||||
assert!(p.at(MOD_KW));
|
||||
p.bump();
|
17
src/parser/grammar/items/traits.rs
Normal file
17
src/parser/grammar/items/traits.rs
Normal file
@ -0,0 +1,17 @@
|
||||
use super::*;
|
||||
|
||||
pub(super) fn trait_item(p: &mut Parser) {
|
||||
assert!(p.at(TRAIT_KW));
|
||||
p.bump();
|
||||
p.expect(IDENT);
|
||||
p.expect(L_CURLY);
|
||||
p.expect(R_CURLY);
|
||||
}
|
||||
|
||||
pub(super) fn impl_item(p: &mut Parser) {
|
||||
assert!(p.at(IMPL_KW));
|
||||
p.bump();
|
||||
p.expect(IDENT);
|
||||
p.expect(L_CURLY);
|
||||
p.expect(R_CURLY);
|
||||
}
|
@ -1,88 +1,21 @@
|
||||
use {File, FileBuilder, Sink, SyntaxKind, Token};
|
||||
use {File, SyntaxKind, Token};
|
||||
|
||||
use syntax_kinds::*;
|
||||
|
||||
mod event_parser;
|
||||
use self::event_parser::Event;
|
||||
#[macro_use]
|
||||
mod parser;
|
||||
mod event;
|
||||
mod grammar;
|
||||
use self::event::Event;
|
||||
|
||||
/// Parse a sequence of tokens into the representative node tree
|
||||
pub fn parse(text: String, tokens: &[Token]) -> File {
|
||||
let events = event_parser::parse(&text, tokens);
|
||||
from_events_to_file(text, tokens, events)
|
||||
}
|
||||
|
||||
fn from_events_to_file(text: String, tokens: &[Token], events: Vec<Event>) -> File {
|
||||
let mut builder = FileBuilder::new(text);
|
||||
let mut idx = 0;
|
||||
|
||||
let mut holes = Vec::new();
|
||||
let mut forward_parents = Vec::new();
|
||||
|
||||
for (i, event) in events.iter().enumerate() {
|
||||
if holes.last() == Some(&i) {
|
||||
holes.pop();
|
||||
continue;
|
||||
}
|
||||
|
||||
match event {
|
||||
&Event::Start {
|
||||
kind: TOMBSTONE, ..
|
||||
} => (),
|
||||
|
||||
&Event::Start { .. } => {
|
||||
forward_parents.clear();
|
||||
let mut idx = i;
|
||||
loop {
|
||||
let (kind, fwd) = match events[idx] {
|
||||
Event::Start {
|
||||
kind,
|
||||
forward_parent,
|
||||
} => (kind, forward_parent),
|
||||
_ => unreachable!(),
|
||||
};
|
||||
forward_parents.push((idx, kind));
|
||||
if let Some(fwd) = fwd {
|
||||
idx += fwd as usize;
|
||||
} else {
|
||||
break;
|
||||
}
|
||||
}
|
||||
for &(idx, kind) in forward_parents.iter().into_iter().rev() {
|
||||
builder.start_internal(kind);
|
||||
holes.push(idx);
|
||||
}
|
||||
holes.pop();
|
||||
}
|
||||
&Event::Finish => {
|
||||
while idx < tokens.len() {
|
||||
let token = tokens[idx];
|
||||
if is_insignificant(token.kind) {
|
||||
idx += 1;
|
||||
builder.leaf(token.kind, token.len);
|
||||
} else {
|
||||
break;
|
||||
}
|
||||
}
|
||||
builder.finish_internal()
|
||||
}
|
||||
&Event::Token {
|
||||
kind: _,
|
||||
mut n_raw_tokens,
|
||||
} => loop {
|
||||
let token = tokens[idx];
|
||||
if !is_insignificant(token.kind) {
|
||||
n_raw_tokens -= 1;
|
||||
}
|
||||
idx += 1;
|
||||
builder.leaf(token.kind, token.len);
|
||||
if n_raw_tokens == 0 {
|
||||
break;
|
||||
}
|
||||
},
|
||||
&Event::Error { ref message } => builder.error().message(message.clone()).emit(),
|
||||
}
|
||||
}
|
||||
builder.finish()
|
||||
let events = {
|
||||
let mut parser = parser::Parser::new(&text, tokens);
|
||||
grammar::file(&mut parser);
|
||||
parser.into_events()
|
||||
};
|
||||
event::to_file(text, tokens, events)
|
||||
}
|
||||
|
||||
fn is_insignificant(kind: SyntaxKind) -> bool {
|
||||
|
@ -1,6 +1,6 @@
|
||||
use {SyntaxKind, TextUnit, Token};
|
||||
use super::Event;
|
||||
use super::super::is_insignificant;
|
||||
use super::is_insignificant;
|
||||
use SyntaxKind::{EOF, TOMBSTONE};
|
||||
|
||||
pub(crate) struct Marker {
|
@ -92,6 +92,8 @@ pub enum SyntaxKind {
|
||||
USE_ITEM,
|
||||
STATIC_ITEM,
|
||||
CONST_ITEM,
|
||||
TRAIT_ITEM,
|
||||
IMPL_ITEM,
|
||||
EXTERN_BLOCK,
|
||||
ENUM_VARIANT,
|
||||
NAMED_FIELD,
|
||||
@ -207,6 +209,8 @@ impl SyntaxKind {
|
||||
USE_ITEM => &SyntaxInfo { name: "USE_ITEM" },
|
||||
STATIC_ITEM => &SyntaxInfo { name: "STATIC_ITEM" },
|
||||
CONST_ITEM => &SyntaxInfo { name: "CONST_ITEM" },
|
||||
TRAIT_ITEM => &SyntaxInfo { name: "TRAIT_ITEM" },
|
||||
IMPL_ITEM => &SyntaxInfo { name: "IMPL_ITEM" },
|
||||
EXTERN_BLOCK => &SyntaxInfo { name: "EXTERN_BLOCK" },
|
||||
ENUM_VARIANT => &SyntaxInfo { name: "ENUM_VARIANT" },
|
||||
NAMED_FIELD => &SyntaxInfo { name: "NAMED_FIELD" },
|
||||
|
1
tests/data/parser/inline/0007_unsafe_trait.rs
Normal file
1
tests/data/parser/inline/0007_unsafe_trait.rs
Normal file
@ -0,0 +1 @@
|
||||
unsafe trait T {}
|
11
tests/data/parser/inline/0007_unsafe_trait.txt
Normal file
11
tests/data/parser/inline/0007_unsafe_trait.txt
Normal file
@ -0,0 +1,11 @@
|
||||
FILE@[0; 18)
|
||||
TRAIT_ITEM@[0; 18)
|
||||
UNSAFE_KW@[0; 6)
|
||||
WHITESPACE@[6; 7)
|
||||
TRAIT_KW@[7; 12)
|
||||
WHITESPACE@[12; 13)
|
||||
IDENT@[13; 14) "T"
|
||||
WHITESPACE@[14; 15)
|
||||
L_CURLY@[15; 16)
|
||||
R_CURLY@[16; 17)
|
||||
WHITESPACE@[17; 18)
|
1
tests/data/parser/inline/0008_unsafe_impl.rs
Normal file
1
tests/data/parser/inline/0008_unsafe_impl.rs
Normal file
@ -0,0 +1 @@
|
||||
unsafe impl Foo {}
|
11
tests/data/parser/inline/0008_unsafe_impl.txt
Normal file
11
tests/data/parser/inline/0008_unsafe_impl.txt
Normal file
@ -0,0 +1,11 @@
|
||||
FILE@[0; 19)
|
||||
IMPL_ITEM@[0; 19)
|
||||
UNSAFE_KW@[0; 6)
|
||||
WHITESPACE@[6; 7)
|
||||
IMPL_KW@[7; 11)
|
||||
WHITESPACE@[11; 12)
|
||||
IDENT@[12; 15) "Foo"
|
||||
WHITESPACE@[15; 16)
|
||||
L_CURLY@[16; 17)
|
||||
R_CURLY@[17; 18)
|
||||
WHITESPACE@[18; 19)
|
@ -79,16 +79,19 @@ fn collect_tests(s: &str) -> Vec<Test> {
|
||||
.map(str::trim_left)
|
||||
.group_by(|line| line.starts_with(prefix));
|
||||
|
||||
for (is_comment, block) in comment_blocks.into_iter() {
|
||||
'outer: for (is_comment, block) in comment_blocks.into_iter() {
|
||||
if !is_comment {
|
||||
continue;
|
||||
}
|
||||
let mut block = block.map(|line| &line[prefix.len()..]);
|
||||
let first = block.next().unwrap();
|
||||
if !first.starts_with("test ") {
|
||||
continue;
|
||||
}
|
||||
let name = first["test ".len()..].to_string();
|
||||
|
||||
let name = loop {
|
||||
match block.next() {
|
||||
Some(line) if line.starts_with("test ") => break line["test ".len()..].to_string(),
|
||||
Some(_) => (),
|
||||
None => continue 'outer,
|
||||
}
|
||||
};
|
||||
let text: String = itertools::join(block.chain(::std::iter::once("")), "\n");
|
||||
assert!(!text.trim().is_empty() && text.ends_with("\n"));
|
||||
res.push(Test { name, text })
|
||||
@ -121,7 +124,7 @@ fn inline_tests_dir() -> PathBuf {
|
||||
}
|
||||
|
||||
fn grammar_dir() -> PathBuf {
|
||||
base_dir().join("src/parser/event_parser/grammar")
|
||||
base_dir().join("src/parser/grammar")
|
||||
}
|
||||
|
||||
fn base_dir() -> PathBuf {
|
||||
|
Loading…
Reference in New Issue
Block a user