mirror of
https://github.com/rust-lang/rust.git
synced 2024-11-26 16:54:01 +00:00
Auto merge of #14084 - Veykril:float-parse, r=Veykril
fix: Fix parsing of nested tuple field accesses in a cursed way This is absolutely terrible but seems to work. Macro fragment parsing comes next.
This commit is contained in:
commit
7f1234492e
@ -111,7 +111,8 @@ impl ItemTree {
|
||||
Some(node) => node,
|
||||
None => return Default::default(),
|
||||
};
|
||||
if never!(syntax.kind() == SyntaxKind::ERROR) {
|
||||
if never!(syntax.kind() == SyntaxKind::ERROR, "{:?} from {:?} {}", file_id, syntax, syntax)
|
||||
{
|
||||
// FIXME: not 100% sure why these crop up, but return an empty tree to avoid a panic
|
||||
return Default::default();
|
||||
}
|
||||
@ -133,7 +134,7 @@ impl ItemTree {
|
||||
ctx.lower_macro_stmts(stmts)
|
||||
},
|
||||
_ => {
|
||||
panic!("cannot create item tree from {syntax:?} {syntax}");
|
||||
panic!("cannot create item tree for file {file_id:?} from {syntax:?} {syntax}");
|
||||
},
|
||||
}
|
||||
};
|
||||
|
@ -97,6 +97,41 @@ fn#19 main#20(#21)#21 {#22
|
||||
"##]],
|
||||
);
|
||||
}
|
||||
#[test]
|
||||
fn float_field_acces_macro_input() {
|
||||
check(
|
||||
r#"
|
||||
macro_rules! foo {
|
||||
($expr:expr) => {
|
||||
fn foo() {
|
||||
$expr;
|
||||
}
|
||||
};
|
||||
}
|
||||
foo!(x .0.1);
|
||||
foo!(x .2. 3);
|
||||
foo!(x .4 .5);
|
||||
"#,
|
||||
expect![[r#"
|
||||
macro_rules! foo {
|
||||
($expr:expr) => {
|
||||
fn foo() {
|
||||
$expr;
|
||||
}
|
||||
};
|
||||
}
|
||||
fn foo() {
|
||||
(x.0.1);
|
||||
}
|
||||
fn foo() {
|
||||
(x.2.3);
|
||||
}
|
||||
fn foo() {
|
||||
(x.4.5);
|
||||
}
|
||||
"#]],
|
||||
);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn mbe_smoke_test() {
|
||||
|
@ -104,7 +104,7 @@ macro_rules! id {
|
||||
$($t)*
|
||||
};
|
||||
}
|
||||
id /*+errors*/! {
|
||||
id! {
|
||||
#[proc_macros::identity]
|
||||
impl Foo for WrapBj {
|
||||
async fn foo(&self) {
|
||||
@ -113,18 +113,17 @@ id /*+errors*/! {
|
||||
}
|
||||
}
|
||||
"#,
|
||||
expect![[r##"
|
||||
expect![[r#"
|
||||
macro_rules! id {
|
||||
($($t:tt)*) => {
|
||||
$($t)*
|
||||
};
|
||||
}
|
||||
/* parse error: expected SEMICOLON */
|
||||
#[proc_macros::identity] impl Foo for WrapBj {
|
||||
async fn foo(&self ) {
|
||||
self .0.id().await ;
|
||||
}
|
||||
}
|
||||
"##]],
|
||||
"#]],
|
||||
);
|
||||
}
|
||||
|
@ -95,6 +95,9 @@ pub fn token_tree_to_syntax_node(
|
||||
parser::Step::Token { kind, n_input_tokens: n_raw_tokens } => {
|
||||
tree_sink.token(kind, n_raw_tokens)
|
||||
}
|
||||
parser::Step::FloatSplit { ends_in_dot: has_pseudo_dot } => {
|
||||
tree_sink.float_split(has_pseudo_dot)
|
||||
}
|
||||
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()),
|
||||
@ -796,6 +799,43 @@ fn delim_to_str(d: tt::DelimiterKind, closing: bool) -> Option<&'static str> {
|
||||
}
|
||||
|
||||
impl<'a> TtTreeSink<'a> {
|
||||
/// Parses a float literal as if it was a one to two name ref nodes with a dot inbetween.
|
||||
/// This occurs when a float literal is used as a field access.
|
||||
fn float_split(&mut self, has_pseudo_dot: bool) {
|
||||
let (text, _span) = match self.cursor.token_tree() {
|
||||
Some(tt::buffer::TokenTreeRef::Leaf(tt::Leaf::Literal(lit), _)) => {
|
||||
(lit.text.as_str(), lit.span)
|
||||
}
|
||||
_ => unreachable!(),
|
||||
};
|
||||
match text.split_once('.') {
|
||||
Some((left, right)) => {
|
||||
assert!(!left.is_empty());
|
||||
self.inner.start_node(SyntaxKind::NAME_REF);
|
||||
self.inner.token(SyntaxKind::INT_NUMBER, left);
|
||||
self.inner.finish_node();
|
||||
|
||||
// here we move the exit up, the original exit has been deleted in process
|
||||
self.inner.finish_node();
|
||||
|
||||
self.inner.token(SyntaxKind::DOT, ".");
|
||||
|
||||
if has_pseudo_dot {
|
||||
assert!(right.is_empty(), "{left}.{right}");
|
||||
} else {
|
||||
self.inner.start_node(SyntaxKind::NAME_REF);
|
||||
self.inner.token(SyntaxKind::INT_NUMBER, right);
|
||||
self.inner.finish_node();
|
||||
|
||||
// the parser creates an unbalanced start node, we are required to close it here
|
||||
self.inner.finish_node();
|
||||
}
|
||||
}
|
||||
None => unreachable!(),
|
||||
}
|
||||
self.cursor = self.cursor.bump();
|
||||
}
|
||||
|
||||
fn token(&mut self, kind: SyntaxKind, mut n_tokens: u8) {
|
||||
if kind == LIFETIME_IDENT {
|
||||
n_tokens = 2;
|
||||
|
@ -45,6 +45,13 @@ pub(crate) fn to_parser_input(buffer: &TokenBuffer<'_>) -> parser::Input {
|
||||
.unwrap_or_else(|| panic!("Fail to convert given literal {:#?}", &lit));
|
||||
|
||||
res.push(kind);
|
||||
|
||||
if kind == FLOAT_NUMBER && !inner_text.ends_with('.') {
|
||||
// Tag the token as joint if it is float with a fractional part
|
||||
// we use this jointness to inform the parser about what token split
|
||||
// event to emit when we encounter a float literal in a field access
|
||||
res.was_joint();
|
||||
}
|
||||
}
|
||||
tt::Leaf::Ident(ident) => match ident.text.as_ref() {
|
||||
"_" => res.push(T![_]),
|
||||
|
@ -150,6 +150,11 @@ impl<'a> TtIter<'a> {
|
||||
cursor = cursor.bump_subtree();
|
||||
}
|
||||
}
|
||||
parser::Step::FloatSplit { .. } => {
|
||||
// FIXME: We need to split the tree properly here, but mutating the token trees
|
||||
// in the buffer is somewhat tricky to pull off.
|
||||
cursor = cursor.bump_subtree();
|
||||
}
|
||||
parser::Step::Enter { .. } | parser::Step::Exit => (),
|
||||
parser::Step::Error { .. } => error = true,
|
||||
}
|
||||
@ -166,19 +171,18 @@ impl<'a> TtIter<'a> {
|
||||
|
||||
if cursor.is_root() {
|
||||
while curr != cursor {
|
||||
if let Some(token) = curr.token_tree() {
|
||||
res.push(token);
|
||||
}
|
||||
let Some(token) = curr.token_tree() else { break };
|
||||
res.push(token.cloned());
|
||||
curr = curr.bump();
|
||||
}
|
||||
}
|
||||
|
||||
self.inner = self.inner.as_slice()[res.len()..].iter();
|
||||
let res = match res.len() {
|
||||
1 => Some(res[0].cloned()),
|
||||
0 => None,
|
||||
0 | 1 => res.pop(),
|
||||
_ => Some(tt::TokenTree::Subtree(tt::Subtree {
|
||||
delimiter: tt::Delimiter::unspecified(),
|
||||
token_trees: res.into_iter().map(|it| it.cloned()).collect(),
|
||||
token_trees: res,
|
||||
})),
|
||||
};
|
||||
ExpandResult { value: res, err }
|
||||
|
@ -74,7 +74,13 @@ pub(crate) enum Event {
|
||||
kind: SyntaxKind,
|
||||
n_raw_tokens: u8,
|
||||
},
|
||||
|
||||
/// When we parse `foo.0.0` or `foo. 0. 0` the lexer will hand us a float literal
|
||||
/// instead of an integer literal followed by a dot as the lexer has no contextual knowledge.
|
||||
/// This event instructs whatever consumes the events to split the float literal into
|
||||
/// the corresponding parts.
|
||||
FloatSplitHack {
|
||||
ends_in_dot: bool,
|
||||
},
|
||||
Error {
|
||||
msg: String,
|
||||
},
|
||||
@ -125,6 +131,11 @@ pub(super) fn process(mut events: Vec<Event>) -> Output {
|
||||
Event::Token { kind, n_raw_tokens } => {
|
||||
res.token(kind, n_raw_tokens);
|
||||
}
|
||||
Event::FloatSplitHack { ends_in_dot } => {
|
||||
res.float_split_hack(ends_in_dot);
|
||||
let ev = mem::replace(&mut events[i + 1], Event::tombstone());
|
||||
assert!(matches!(ev, Event::Finish), "{ev:?}");
|
||||
}
|
||||
Event::Error { msg } => res.error(msg),
|
||||
}
|
||||
}
|
||||
|
@ -379,7 +379,7 @@ fn postfix_expr(
|
||||
// }
|
||||
T!['('] if allow_calls => call_expr(p, lhs),
|
||||
T!['['] if allow_calls => index_expr(p, lhs),
|
||||
T![.] => match postfix_dot_expr(p, lhs) {
|
||||
T![.] => match postfix_dot_expr::<false>(p, lhs) {
|
||||
Ok(it) => it,
|
||||
Err(it) => {
|
||||
lhs = it;
|
||||
@ -393,35 +393,44 @@ fn postfix_expr(
|
||||
block_like = BlockLike::NotBlock;
|
||||
}
|
||||
return (lhs, block_like);
|
||||
}
|
||||
|
||||
fn postfix_dot_expr(
|
||||
p: &mut Parser<'_>,
|
||||
lhs: CompletedMarker,
|
||||
) -> Result<CompletedMarker, CompletedMarker> {
|
||||
fn postfix_dot_expr<const FLOAT_RECOVERY: bool>(
|
||||
p: &mut Parser<'_>,
|
||||
lhs: CompletedMarker,
|
||||
) -> Result<CompletedMarker, CompletedMarker> {
|
||||
if !FLOAT_RECOVERY {
|
||||
assert!(p.at(T![.]));
|
||||
if p.nth(1) == IDENT && (p.nth(2) == T!['('] || p.nth_at(2, T![::])) {
|
||||
return Ok(method_call_expr(p, lhs));
|
||||
}
|
||||
|
||||
// test await_expr
|
||||
// fn foo() {
|
||||
// x.await;
|
||||
// x.0.await;
|
||||
// x.0().await?.hello();
|
||||
// }
|
||||
if p.nth(1) == T![await] {
|
||||
let m = lhs.precede(p);
|
||||
p.bump(T![.]);
|
||||
p.bump(T![await]);
|
||||
return Ok(m.complete(p, AWAIT_EXPR));
|
||||
}
|
||||
|
||||
if p.at(T![..=]) || p.at(T![..]) {
|
||||
return Err(lhs);
|
||||
}
|
||||
|
||||
Ok(field_expr(p, lhs))
|
||||
}
|
||||
let nth1 = if FLOAT_RECOVERY { 0 } else { 1 };
|
||||
let nth2 = if FLOAT_RECOVERY { 1 } else { 2 };
|
||||
|
||||
if p.nth(nth1) == IDENT && (p.nth(nth2) == T!['('] || p.nth_at(nth2, T![::])) {
|
||||
return Ok(method_call_expr::<FLOAT_RECOVERY>(p, lhs));
|
||||
}
|
||||
|
||||
// test await_expr
|
||||
// fn foo() {
|
||||
// x.await;
|
||||
// x.0.await;
|
||||
// x.0().await?.hello();
|
||||
// x.0.0.await;
|
||||
// x.0. await;
|
||||
// }
|
||||
if p.nth(nth1) == T![await] {
|
||||
let m = lhs.precede(p);
|
||||
if !FLOAT_RECOVERY {
|
||||
p.bump(T![.]);
|
||||
}
|
||||
p.bump(T![await]);
|
||||
return Ok(m.complete(p, AWAIT_EXPR));
|
||||
}
|
||||
|
||||
if p.at(T![..=]) || p.at(T![..]) {
|
||||
return Err(lhs);
|
||||
}
|
||||
|
||||
field_expr::<FLOAT_RECOVERY>(p, lhs)
|
||||
}
|
||||
|
||||
// test call_expr
|
||||
@ -455,11 +464,22 @@ fn index_expr(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
|
||||
// fn foo() {
|
||||
// x.foo();
|
||||
// y.bar::<T>(1, 2,);
|
||||
// x.0.0.call();
|
||||
// x.0. call();
|
||||
// }
|
||||
fn method_call_expr(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
|
||||
assert!(p.at(T![.]) && p.nth(1) == IDENT && (p.nth(2) == T!['('] || p.nth_at(2, T![::])));
|
||||
fn method_call_expr<const FLOAT_RECOVERY: bool>(
|
||||
p: &mut Parser<'_>,
|
||||
lhs: CompletedMarker,
|
||||
) -> CompletedMarker {
|
||||
if FLOAT_RECOVERY {
|
||||
assert!(p.nth(0) == IDENT && (p.nth(1) == T!['('] || p.nth_at(1, T![::])));
|
||||
} else {
|
||||
assert!(p.at(T![.]) && p.nth(1) == IDENT && (p.nth(2) == T!['('] || p.nth_at(2, T![::])));
|
||||
}
|
||||
let m = lhs.precede(p);
|
||||
p.bump_any();
|
||||
if !FLOAT_RECOVERY {
|
||||
p.bump(T![.]);
|
||||
}
|
||||
name_ref(p);
|
||||
generic_args::opt_generic_arg_list(p, true);
|
||||
if p.at(T!['(']) {
|
||||
@ -472,21 +492,35 @@ fn method_call_expr(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker
|
||||
// fn foo() {
|
||||
// x.foo;
|
||||
// x.0.bar;
|
||||
// x.0.1;
|
||||
// x.0. bar;
|
||||
// x.0();
|
||||
// }
|
||||
fn field_expr(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
|
||||
assert!(p.at(T![.]));
|
||||
fn field_expr<const FLOAT_RECOVERY: bool>(
|
||||
p: &mut Parser<'_>,
|
||||
lhs: CompletedMarker,
|
||||
) -> Result<CompletedMarker, CompletedMarker> {
|
||||
if !FLOAT_RECOVERY {
|
||||
assert!(p.at(T![.]));
|
||||
}
|
||||
let m = lhs.precede(p);
|
||||
p.bump(T![.]);
|
||||
if !FLOAT_RECOVERY {
|
||||
p.bump(T![.]);
|
||||
}
|
||||
if p.at(IDENT) || p.at(INT_NUMBER) {
|
||||
name_ref_or_index(p);
|
||||
} else if p.at(FLOAT_NUMBER) {
|
||||
// FIXME: How to recover and instead parse INT + T![.]?
|
||||
p.bump_any();
|
||||
return match p.split_float(m) {
|
||||
(true, m) => {
|
||||
let lhs = m.complete(p, FIELD_EXPR);
|
||||
postfix_dot_expr::<true>(p, lhs)
|
||||
}
|
||||
(false, m) => Ok(m.complete(p, FIELD_EXPR)),
|
||||
};
|
||||
} else {
|
||||
p.error("expected field name or number");
|
||||
}
|
||||
m.complete(p, FIELD_EXPR)
|
||||
Ok(m.complete(p, FIELD_EXPR))
|
||||
}
|
||||
|
||||
// test try_expr
|
||||
|
@ -102,10 +102,14 @@ impl TopEntryPoint {
|
||||
match step {
|
||||
Step::Enter { .. } => depth += 1,
|
||||
Step::Exit => depth -= 1,
|
||||
Step::FloatSplit { ends_in_dot: has_pseudo_dot } => {
|
||||
depth -= 1 + !has_pseudo_dot as usize
|
||||
}
|
||||
Step::Token { .. } | Step::Error { .. } => (),
|
||||
}
|
||||
}
|
||||
assert!(!first, "no tree at all");
|
||||
assert_eq!(depth, 0, "unbalanced tree");
|
||||
}
|
||||
|
||||
res
|
||||
|
@ -25,53 +25,88 @@ pub struct Output {
|
||||
#[derive(Debug)]
|
||||
pub enum Step<'a> {
|
||||
Token { kind: SyntaxKind, n_input_tokens: u8 },
|
||||
FloatSplit { ends_in_dot: bool },
|
||||
Enter { kind: SyntaxKind },
|
||||
Exit,
|
||||
Error { msg: &'a str },
|
||||
}
|
||||
|
||||
impl Output {
|
||||
const EVENT_MASK: u32 = 0b1;
|
||||
const TAG_MASK: u32 = 0x0000_00F0;
|
||||
const N_INPUT_TOKEN_MASK: u32 = 0x0000_FF00;
|
||||
const KIND_MASK: u32 = 0xFFFF_0000;
|
||||
|
||||
const ERROR_SHIFT: u32 = Self::EVENT_MASK.trailing_ones();
|
||||
const TAG_SHIFT: u32 = Self::TAG_MASK.trailing_zeros();
|
||||
const N_INPUT_TOKEN_SHIFT: u32 = Self::N_INPUT_TOKEN_MASK.trailing_zeros();
|
||||
const KIND_SHIFT: u32 = Self::KIND_MASK.trailing_zeros();
|
||||
|
||||
const TOKEN_EVENT: u8 = 0;
|
||||
const ENTER_EVENT: u8 = 1;
|
||||
const EXIT_EVENT: u8 = 2;
|
||||
const SPLIT_EVENT: u8 = 3;
|
||||
|
||||
pub fn iter(&self) -> impl Iterator<Item = Step<'_>> {
|
||||
self.event.iter().map(|&event| {
|
||||
if event & 0b1 == 0 {
|
||||
return Step::Error { msg: self.error[(event as usize) >> 1].as_str() };
|
||||
if event & Self::EVENT_MASK == 0 {
|
||||
return Step::Error {
|
||||
msg: self.error[(event as usize) >> Self::ERROR_SHIFT].as_str(),
|
||||
};
|
||||
}
|
||||
let tag = ((event & 0x0000_00F0) >> 4) as u8;
|
||||
let tag = ((event & Self::TAG_MASK) >> Self::TAG_SHIFT) as u8;
|
||||
match tag {
|
||||
0 => {
|
||||
let kind: SyntaxKind = (((event & 0xFFFF_0000) >> 16) as u16).into();
|
||||
let n_input_tokens = ((event & 0x0000_FF00) >> 8) as u8;
|
||||
Self::TOKEN_EVENT => {
|
||||
let kind: SyntaxKind =
|
||||
(((event & Self::KIND_MASK) >> Self::KIND_SHIFT) as u16).into();
|
||||
let n_input_tokens =
|
||||
((event & Self::N_INPUT_TOKEN_MASK) >> Self::N_INPUT_TOKEN_SHIFT) as u8;
|
||||
Step::Token { kind, n_input_tokens }
|
||||
}
|
||||
1 => {
|
||||
let kind: SyntaxKind = (((event & 0xFFFF_0000) >> 16) as u16).into();
|
||||
Self::ENTER_EVENT => {
|
||||
let kind: SyntaxKind =
|
||||
(((event & Self::KIND_MASK) >> Self::KIND_SHIFT) as u16).into();
|
||||
Step::Enter { kind }
|
||||
}
|
||||
2 => Step::Exit,
|
||||
Self::EXIT_EVENT => Step::Exit,
|
||||
Self::SPLIT_EVENT => {
|
||||
Step::FloatSplit { ends_in_dot: event & Self::N_INPUT_TOKEN_MASK != 0 }
|
||||
}
|
||||
_ => unreachable!(),
|
||||
}
|
||||
})
|
||||
}
|
||||
|
||||
pub(crate) fn token(&mut self, kind: SyntaxKind, n_tokens: u8) {
|
||||
let e = ((kind as u16 as u32) << 16) | ((n_tokens as u32) << 8) | 1;
|
||||
let e = ((kind as u16 as u32) << Self::KIND_SHIFT)
|
||||
| ((n_tokens as u32) << Self::N_INPUT_TOKEN_SHIFT)
|
||||
| Self::EVENT_MASK;
|
||||
self.event.push(e)
|
||||
}
|
||||
|
||||
pub(crate) fn float_split_hack(&mut self, ends_in_dot: bool) {
|
||||
let e = (Self::SPLIT_EVENT as u32) << Self::TAG_SHIFT
|
||||
| ((ends_in_dot as u32) << Self::N_INPUT_TOKEN_SHIFT)
|
||||
| Self::EVENT_MASK;
|
||||
self.event.push(e);
|
||||
}
|
||||
|
||||
pub(crate) fn enter_node(&mut self, kind: SyntaxKind) {
|
||||
let e = ((kind as u16 as u32) << 16) | (1 << 4) | 1;
|
||||
let e = ((kind as u16 as u32) << Self::KIND_SHIFT)
|
||||
| ((Self::ENTER_EVENT as u32) << Self::TAG_SHIFT)
|
||||
| Self::EVENT_MASK;
|
||||
self.event.push(e)
|
||||
}
|
||||
|
||||
pub(crate) fn leave_node(&mut self) {
|
||||
let e = 2 << 4 | 1;
|
||||
let e = (Self::EXIT_EVENT as u32) << Self::TAG_SHIFT | Self::EVENT_MASK;
|
||||
self.event.push(e)
|
||||
}
|
||||
|
||||
pub(crate) fn error(&mut self, error: String) {
|
||||
let idx = self.error.len();
|
||||
self.error.push(error);
|
||||
let e = (idx as u32) << 1;
|
||||
let e = (idx as u32) << Self::ERROR_SHIFT;
|
||||
self.event.push(e);
|
||||
}
|
||||
}
|
||||
|
@ -181,6 +181,35 @@ impl<'t> Parser<'t> {
|
||||
self.do_bump(kind, 1);
|
||||
}
|
||||
|
||||
/// Advances the parser by one token
|
||||
pub(crate) fn split_float(&mut self, mut marker: Marker) -> (bool, Marker) {
|
||||
assert!(self.at(SyntaxKind::FLOAT_NUMBER));
|
||||
// we have parse `<something>.`
|
||||
// `<something>`.0.1
|
||||
// here we need to insert an extra event
|
||||
//
|
||||
// `<something>`. 0. 1;
|
||||
// here we need to change the follow up parse, the return value will cause us to emulate a dot
|
||||
// the actual splitting happens later
|
||||
let ends_in_dot = !self.inp.is_joint(self.pos);
|
||||
if !ends_in_dot {
|
||||
let new_marker = self.start();
|
||||
let idx = marker.pos as usize;
|
||||
match &mut self.events[idx] {
|
||||
Event::Start { forward_parent, kind } => {
|
||||
*kind = SyntaxKind::FIELD_EXPR;
|
||||
*forward_parent = Some(new_marker.pos - marker.pos);
|
||||
}
|
||||
_ => unreachable!(),
|
||||
}
|
||||
marker.bomb.defuse();
|
||||
marker = new_marker;
|
||||
};
|
||||
self.pos += 1 as usize;
|
||||
self.push_event(Event::FloatSplitHack { ends_in_dot });
|
||||
(ends_in_dot, marker)
|
||||
}
|
||||
|
||||
/// Advances the parser by one token, remapping its kind.
|
||||
/// This is useful to create contextual keywords from
|
||||
/// identifiers. For example, the lexer creates a `union`
|
||||
|
@ -43,7 +43,16 @@ impl<'a> LexedStr<'a> {
|
||||
res.was_joint();
|
||||
}
|
||||
res.push(kind);
|
||||
// Tag the token as joint if it is float with a fractional part
|
||||
// we use this jointness to inform the parser about what token split
|
||||
// event to emit when we encounter a float literal in a field access
|
||||
if kind == SyntaxKind::FLOAT_NUMBER {
|
||||
if !self.text(i).ends_with('.') {
|
||||
res.was_joint();
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
was_joint = true;
|
||||
}
|
||||
}
|
||||
@ -63,6 +72,9 @@ impl<'a> LexedStr<'a> {
|
||||
Step::Token { kind, n_input_tokens: n_raw_tokens } => {
|
||||
builder.token(kind, n_raw_tokens)
|
||||
}
|
||||
Step::FloatSplit { ends_in_dot: has_pseudo_dot } => {
|
||||
builder.float_split(has_pseudo_dot)
|
||||
}
|
||||
Step::Enter { kind } => builder.enter(kind),
|
||||
Step::Exit => builder.exit(),
|
||||
Step::Error { msg } => {
|
||||
@ -109,6 +121,16 @@ impl Builder<'_, '_> {
|
||||
self.do_token(kind, n_tokens as usize);
|
||||
}
|
||||
|
||||
fn float_split(&mut self, has_pseudo_dot: bool) {
|
||||
match mem::replace(&mut self.state, State::Normal) {
|
||||
State::PendingEnter => unreachable!(),
|
||||
State::PendingExit => (self.sink)(StrStep::Exit),
|
||||
State::Normal => (),
|
||||
}
|
||||
self.eat_trivias();
|
||||
self.do_float_split(has_pseudo_dot);
|
||||
}
|
||||
|
||||
fn enter(&mut self, kind: SyntaxKind) {
|
||||
match mem::replace(&mut self.state, State::Normal) {
|
||||
State::PendingEnter => {
|
||||
@ -164,6 +186,37 @@ impl Builder<'_, '_> {
|
||||
self.pos += n_tokens;
|
||||
(self.sink)(StrStep::Token { kind, text });
|
||||
}
|
||||
|
||||
fn do_float_split(&mut self, has_pseudo_dot: bool) {
|
||||
let text = &self.lexed.range_text(self.pos..self.pos + 1);
|
||||
self.pos += 1;
|
||||
match text.split_once('.') {
|
||||
Some((left, right)) => {
|
||||
assert!(!left.is_empty());
|
||||
(self.sink)(StrStep::Enter { kind: SyntaxKind::NAME_REF });
|
||||
(self.sink)(StrStep::Token { kind: SyntaxKind::INT_NUMBER, text: left });
|
||||
(self.sink)(StrStep::Exit);
|
||||
|
||||
// here we move the exit up, the original exit has been deleted in process
|
||||
(self.sink)(StrStep::Exit);
|
||||
|
||||
(self.sink)(StrStep::Token { kind: SyntaxKind::DOT, text: "." });
|
||||
|
||||
if has_pseudo_dot {
|
||||
assert!(right.is_empty(), "{left}.{right}");
|
||||
self.state = State::Normal;
|
||||
} else {
|
||||
(self.sink)(StrStep::Enter { kind: SyntaxKind::NAME_REF });
|
||||
(self.sink)(StrStep::Token { kind: SyntaxKind::INT_NUMBER, text: right });
|
||||
(self.sink)(StrStep::Exit);
|
||||
|
||||
// the parser creates an unbalanced start node, we are required to close it here
|
||||
self.state = State::PendingExit;
|
||||
}
|
||||
}
|
||||
None => unreachable!(),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn n_attached_trivias<'a>(
|
||||
|
@ -51,6 +51,9 @@ fn expr() {
|
||||
check(PrefixEntryPoint::Expr, "-1", "-1");
|
||||
check(PrefixEntryPoint::Expr, "fn foo() {}", "fn");
|
||||
check(PrefixEntryPoint::Expr, "#[attr] ()", "#[attr] ()");
|
||||
check(PrefixEntryPoint::Expr, "foo.0", "foo.0");
|
||||
check(PrefixEntryPoint::Expr, "foo.0.1", "foo.0.1");
|
||||
check(PrefixEntryPoint::Expr, "foo.0. foo", "foo.0. foo");
|
||||
}
|
||||
|
||||
#[test]
|
||||
@ -88,6 +91,7 @@ fn check(entry: PrefixEntryPoint, input: &str, prefix: &str) {
|
||||
for step in entry.parse(&input).iter() {
|
||||
match step {
|
||||
Step::Token { n_input_tokens, .. } => n_tokens += n_input_tokens as usize,
|
||||
Step::FloatSplit { .. } => n_tokens += 1,
|
||||
Step::Enter { .. } | Step::Exit | Step::Error { .. } => (),
|
||||
}
|
||||
}
|
||||
|
@ -40,6 +40,39 @@ SOURCE_FILE
|
||||
IDENT "bar"
|
||||
SEMICOLON ";"
|
||||
WHITESPACE "\n "
|
||||
EXPR_STMT
|
||||
FIELD_EXPR
|
||||
FIELD_EXPR
|
||||
PATH_EXPR
|
||||
PATH
|
||||
PATH_SEGMENT
|
||||
NAME_REF
|
||||
IDENT "x"
|
||||
DOT "."
|
||||
NAME_REF
|
||||
INT_NUMBER "0"
|
||||
DOT "."
|
||||
NAME_REF
|
||||
INT_NUMBER "1"
|
||||
SEMICOLON ";"
|
||||
WHITESPACE "\n "
|
||||
EXPR_STMT
|
||||
FIELD_EXPR
|
||||
FIELD_EXPR
|
||||
PATH_EXPR
|
||||
PATH
|
||||
PATH_SEGMENT
|
||||
NAME_REF
|
||||
IDENT "x"
|
||||
DOT "."
|
||||
NAME_REF
|
||||
INT_NUMBER "0"
|
||||
DOT "."
|
||||
WHITESPACE " "
|
||||
NAME_REF
|
||||
IDENT "bar"
|
||||
SEMICOLON ";"
|
||||
WHITESPACE "\n "
|
||||
EXPR_STMT
|
||||
CALL_EXPR
|
||||
FIELD_EXPR
|
||||
|
@ -1,5 +1,7 @@
|
||||
fn foo() {
|
||||
x.foo;
|
||||
x.0.bar;
|
||||
x.0.1;
|
||||
x.0. bar;
|
||||
x.0();
|
||||
}
|
||||
|
@ -58,6 +58,49 @@ SOURCE_FILE
|
||||
COMMA ","
|
||||
R_PAREN ")"
|
||||
SEMICOLON ";"
|
||||
WHITESPACE "\n "
|
||||
EXPR_STMT
|
||||
METHOD_CALL_EXPR
|
||||
FIELD_EXPR
|
||||
FIELD_EXPR
|
||||
PATH_EXPR
|
||||
PATH
|
||||
PATH_SEGMENT
|
||||
NAME_REF
|
||||
IDENT "x"
|
||||
DOT "."
|
||||
NAME_REF
|
||||
INT_NUMBER "0"
|
||||
DOT "."
|
||||
NAME_REF
|
||||
INT_NUMBER "0"
|
||||
DOT "."
|
||||
NAME_REF
|
||||
IDENT "call"
|
||||
ARG_LIST
|
||||
L_PAREN "("
|
||||
R_PAREN ")"
|
||||
SEMICOLON ";"
|
||||
WHITESPACE "\n "
|
||||
EXPR_STMT
|
||||
METHOD_CALL_EXPR
|
||||
FIELD_EXPR
|
||||
PATH_EXPR
|
||||
PATH
|
||||
PATH_SEGMENT
|
||||
NAME_REF
|
||||
IDENT "x"
|
||||
DOT "."
|
||||
NAME_REF
|
||||
INT_NUMBER "0"
|
||||
DOT "."
|
||||
WHITESPACE " "
|
||||
NAME_REF
|
||||
IDENT "call"
|
||||
ARG_LIST
|
||||
L_PAREN "("
|
||||
R_PAREN ")"
|
||||
SEMICOLON ";"
|
||||
WHITESPACE "\n"
|
||||
R_CURLY "}"
|
||||
WHITESPACE "\n"
|
||||
|
@ -1,4 +1,6 @@
|
||||
fn foo() {
|
||||
x.foo();
|
||||
y.bar::<T>(1, 2,);
|
||||
x.0.0.call();
|
||||
x.0. call();
|
||||
}
|
||||
|
@ -65,6 +65,41 @@ SOURCE_FILE
|
||||
L_PAREN "("
|
||||
R_PAREN ")"
|
||||
SEMICOLON ";"
|
||||
WHITESPACE "\n "
|
||||
EXPR_STMT
|
||||
AWAIT_EXPR
|
||||
FIELD_EXPR
|
||||
FIELD_EXPR
|
||||
PATH_EXPR
|
||||
PATH
|
||||
PATH_SEGMENT
|
||||
NAME_REF
|
||||
IDENT "x"
|
||||
DOT "."
|
||||
NAME_REF
|
||||
INT_NUMBER "0"
|
||||
DOT "."
|
||||
NAME_REF
|
||||
INT_NUMBER "0"
|
||||
DOT "."
|
||||
AWAIT_KW "await"
|
||||
SEMICOLON ";"
|
||||
WHITESPACE "\n "
|
||||
EXPR_STMT
|
||||
AWAIT_EXPR
|
||||
FIELD_EXPR
|
||||
PATH_EXPR
|
||||
PATH
|
||||
PATH_SEGMENT
|
||||
NAME_REF
|
||||
IDENT "x"
|
||||
DOT "."
|
||||
NAME_REF
|
||||
INT_NUMBER "0"
|
||||
DOT "."
|
||||
WHITESPACE " "
|
||||
AWAIT_KW "await"
|
||||
SEMICOLON ";"
|
||||
WHITESPACE "\n"
|
||||
R_CURLY "}"
|
||||
WHITESPACE "\n"
|
||||
|
@ -2,4 +2,6 @@ fn foo() {
|
||||
x.await;
|
||||
x.0.await;
|
||||
x.0().await?.hello();
|
||||
x.0.0.await;
|
||||
x.0. await;
|
||||
}
|
||||
|
@ -7,7 +7,12 @@ use crate::{Leaf, Subtree, TokenTree};
|
||||
struct EntryId(usize);
|
||||
|
||||
#[derive(Copy, Clone, Debug, Eq, PartialEq)]
|
||||
struct EntryPtr(EntryId, usize);
|
||||
struct EntryPtr(
|
||||
/// The index of the buffer containing the entry.
|
||||
EntryId,
|
||||
/// The index of the entry within the buffer.
|
||||
usize,
|
||||
);
|
||||
|
||||
/// Internal type which is used instead of `TokenTree` to represent a token tree
|
||||
/// within a `TokenBuffer`.
|
||||
@ -16,8 +21,8 @@ enum Entry<'t, Span> {
|
||||
// Mimicking types from proc-macro.
|
||||
Subtree(Option<&'t TokenTree<Span>>, &'t Subtree<Span>, EntryId),
|
||||
Leaf(&'t TokenTree<Span>),
|
||||
// End entries contain a pointer to the entry from the containing
|
||||
// token tree, or None if this is the outermost level.
|
||||
/// End entries contain a pointer to the entry from the containing
|
||||
/// token tree, or [`None`] if this is the outermost level.
|
||||
End(Option<EntryPtr>),
|
||||
}
|
||||
|
||||
@ -226,8 +231,14 @@ impl<'a, Span> Cursor<'a, Span> {
|
||||
/// a cursor into that subtree
|
||||
pub fn bump_subtree(self) -> Cursor<'a, Span> {
|
||||
match self.entry() {
|
||||
Some(Entry::Subtree(_, _, _)) => self.subtree().unwrap(),
|
||||
_ => self.bump(),
|
||||
Some(&Entry::Subtree(_, _, entry_id)) => {
|
||||
Cursor::create(self.buffer, EntryPtr(entry_id, 0))
|
||||
}
|
||||
Some(Entry::End(exit)) => match exit {
|
||||
Some(exit) => Cursor::create(self.buffer, *exit),
|
||||
None => self,
|
||||
},
|
||||
_ => Cursor::create(self.buffer, EntryPtr(self.ptr.0, self.ptr.1 + 1)),
|
||||
}
|
||||
}
|
||||
|
||||
|
Loading…
Reference in New Issue
Block a user