rust-analyzer/crates/mbe/src/syntax_bridge.rs

874 lines
28 KiB
Rust
Raw Normal View History

//! FIXME: write short doc here
2020-08-12 15:06:49 +00:00
use parser::{FragmentKind, ParseError, TreeSink};
2020-08-12 16:26:51 +00:00
use rustc_hash::FxHashMap;
use syntax::{
2020-03-08 16:13:04 +00:00
ast::{self, make::tokens::doc_comment},
2020-03-20 19:04:11 +00:00
tokenize, AstToken, Parse, SmolStr, SyntaxKind,
2020-03-08 16:13:04 +00:00
SyntaxKind::*,
2020-04-24 21:40:41 +00:00
SyntaxNode, SyntaxToken, SyntaxTreeBuilder, TextRange, TextSize, Token as RawToken, T,
2019-02-11 18:12:06 +00:00
};
use tt::buffer::{Cursor, TokenBuffer};
2019-04-07 13:42:53 +00:00
2019-07-20 17:04:34 +00:00
use crate::ExpandError;
use crate::{subtree_source::SubtreeTokenSource, tt_iter::TtIter};
2019-07-20 17:04:34 +00:00
2019-12-13 15:55:51 +00:00
#[derive(Debug, PartialEq, Eq, Clone, Copy)]
pub enum TokenTextRange {
Token(TextRange),
2021-01-19 11:26:53 +00:00
Delimiter(TextRange),
2019-12-13 15:55:51 +00:00
}
impl TokenTextRange {
2019-12-18 03:36:10 +00:00
pub fn by_kind(self, kind: SyntaxKind) -> Option<TextRange> {
2019-12-13 15:55:51 +00:00
match self {
TokenTextRange::Token(it) => Some(it),
2021-01-19 11:26:53 +00:00
TokenTextRange::Delimiter(it) => match kind {
T!['{'] | T!['('] | T!['['] => Some(TextRange::at(it.start(), 1.into())),
T!['}'] | T![')'] | T![']'] => {
Some(TextRange::at(it.end() - TextSize::of('}'), 1.into()))
}
2019-12-13 15:55:51 +00:00
_ => None,
},
}
}
}
2019-02-11 18:31:54 +00:00
/// Maps `tt::TokenId` to the relative range of the original token.
2020-03-18 09:47:59 +00:00
#[derive(Debug, PartialEq, Eq, Clone, Default)]
2019-02-11 18:12:06 +00:00
pub struct TokenMap {
/// Maps `tt::TokenId` to the *relative* source range.
2019-12-13 15:55:51 +00:00
entries: Vec<(tt::TokenId, TokenTextRange)>,
2019-11-03 14:46:12 +00:00
}
2019-02-11 18:31:54 +00:00
/// Convert the syntax tree (what user has written) to a `TokenTree` (what macro
/// will consume).
pub fn ast_to_token_tree(ast: &impl ast::AstNode) -> (tt::Subtree, TokenMap) {
2019-11-18 13:08:41 +00:00
syntax_node_to_token_tree(ast.syntax())
2019-02-11 18:12:06 +00:00
}
/// Convert the syntax node to a `TokenTree` (what macro
/// will consume).
pub fn syntax_node_to_token_tree(node: &SyntaxNode) -> (tt::Subtree, TokenMap) {
2019-11-18 13:08:41 +00:00
let global_offset = node.text_range().start();
2020-03-20 19:04:11 +00:00
let mut c = Convertor::new(node, global_offset);
let subtree = c.go();
c.id_alloc.map.entries.shrink_to_fit();
(subtree, c.id_alloc.map)
}
// The following items are what `rustc` macro can be parsed into :
// link: https://github.com/rust-lang/rust/blob/9ebf47851a357faa4cd97f4b1dc7835f6376e639/src/libsyntax/ext/expand.rs#L141
2019-04-18 19:49:56 +00:00
// * Expr(P<ast::Expr>) -> token_tree_to_expr
// * Pat(P<ast::Pat>) -> token_tree_to_pat
// * Ty(P<ast::Ty>) -> token_tree_to_ty
// * Stmts(SmallVec<[ast::Stmt; 1]>) -> token_tree_to_stmts
// * Items(SmallVec<[P<ast::Item>; 1]>) -> token_tree_to_items
//
// * TraitItems(SmallVec<[ast::TraitItem; 1]>)
2020-05-05 15:56:10 +00:00
// * AssocItems(SmallVec<[ast::AssocItem; 1]>)
// * ForeignItems(SmallVec<[ast::ForeignItem; 1]>
2019-05-25 13:55:46 +00:00
pub fn token_tree_to_syntax_node(
2019-09-02 15:51:03 +00:00
tt: &tt::Subtree,
fragment_kind: FragmentKind,
2019-11-18 13:08:41 +00:00
) -> Result<(Parse<SyntaxNode>, TokenMap), ExpandError> {
2021-01-04 16:22:42 +00:00
let buffer = match tt {
tt::Subtree { delimiter: None, token_trees } => {
TokenBuffer::from_tokens(token_trees.as_slice())
}
2021-01-04 16:22:42 +00:00
_ => TokenBuffer::from_subtree(tt),
};
2019-09-02 15:51:03 +00:00
let mut token_source = SubtreeTokenSource::new(&buffer);
2019-11-03 19:12:19 +00:00
let mut tree_sink = TtTreeSink::new(buffer.begin());
2020-08-12 15:06:49 +00:00
parser::parse_fragment(&mut token_source, &mut tree_sink, fragment_kind);
2019-09-02 15:51:03 +00:00
if tree_sink.roots.len() != 1 {
return Err(ExpandError::ConversionError);
}
//FIXME: would be cool to report errors
2019-11-03 19:12:19 +00:00
let (parse, range_map) = tree_sink.finish();
2019-11-03 14:46:12 +00:00
Ok((parse, range_map))
2019-09-02 15:51:03 +00:00
}
2020-03-08 16:13:04 +00:00
/// Convert a string to a `TokenTree`
pub fn parse_to_token_tree(text: &str) -> Option<(tt::Subtree, TokenMap)> {
let (tokens, errors) = tokenize(text);
if !errors.is_empty() {
return None;
}
let mut conv = RawConvertor {
text,
2020-04-24 21:40:41 +00:00
offset: TextSize::default(),
2020-03-08 16:13:04 +00:00
inner: tokens.iter(),
id_alloc: TokenIdAlloc {
map: Default::default(),
2020-04-24 21:40:41 +00:00
global_offset: TextSize::default(),
2020-03-08 16:13:04 +00:00
next_id: 0,
},
};
let subtree = conv.go();
2020-03-08 16:13:04 +00:00
Some((subtree, conv.id_alloc.map))
}
/// Split token tree with seperate expr: $($e:expr)SEP*
pub fn parse_exprs_with_sep(tt: &tt::Subtree, sep: char) -> Vec<tt::Subtree> {
if tt.token_trees.is_empty() {
return Vec::new();
}
let mut iter = TtIter::new(tt);
let mut res = Vec::new();
while iter.peek_n(0).is_some() {
let expanded = iter.expect_fragment(FragmentKind::Expr);
if expanded.err.is_some() {
break;
}
res.push(match expanded.value {
None => break,
Some(tt @ tt::TokenTree::Leaf(_)) => {
tt::Subtree { delimiter: None, token_trees: vec![tt] }
}
Some(tt::TokenTree::Subtree(tt)) => tt,
});
let mut fork = iter.clone();
if fork.expect_char(sep).is_err() {
break;
}
iter = fork;
}
if iter.peek_n(0).is_some() {
res.push(tt::Subtree { delimiter: None, token_trees: iter.into_iter().cloned().collect() });
}
res
}
2019-02-11 18:12:06 +00:00
impl TokenMap {
pub fn token_by_range(&self, relative_range: TextRange) -> Option<tt::TokenId> {
2019-12-13 15:55:51 +00:00
let &(token_id, _) = self.entries.iter().find(|(_, range)| match range {
TokenTextRange::Token(it) => *it == relative_range,
2021-01-19 11:26:53 +00:00
TokenTextRange::Delimiter(it) => {
let open = TextRange::at(it.start(), 1.into());
let close = TextRange::at(it.end() - TextSize::of('}'), 1.into());
open == relative_range || close == relative_range
2019-12-13 15:55:51 +00:00
}
})?;
2019-11-18 13:08:41 +00:00
Some(token_id)
2019-11-18 12:08:39 +00:00
}
2019-12-13 15:55:51 +00:00
pub fn range_by_token(&self, token_id: tt::TokenId) -> Option<TokenTextRange> {
2019-11-18 13:08:41 +00:00
let &(_, range) = self.entries.iter().find(|(tid, _)| *tid == token_id)?;
Some(range)
}
2019-11-18 13:08:41 +00:00
fn insert(&mut self, token_id: tt::TokenId, relative_range: TextRange) {
2019-12-13 15:55:51 +00:00
self.entries.push((token_id, TokenTextRange::Token(relative_range)));
2019-11-03 14:46:12 +00:00
}
2019-12-12 17:41:44 +00:00
fn insert_delim(
&mut self,
2019-12-13 15:55:51 +00:00
token_id: tt::TokenId,
open_relative_range: TextRange,
close_relative_range: TextRange,
) -> usize {
let res = self.entries.len();
2021-01-19 11:26:53 +00:00
let cover = open_relative_range.cover(close_relative_range);
self.entries.push((token_id, TokenTextRange::Delimiter(cover)));
res
2019-12-12 17:41:44 +00:00
}
2020-03-08 16:13:04 +00:00
fn update_close_delim(&mut self, idx: usize, close_relative_range: TextRange) {
let (_, token_text_range) = &mut self.entries[idx];
2021-01-19 11:26:53 +00:00
if let TokenTextRange::Delimiter(dim) = token_text_range {
let cover = dim.cover(close_relative_range);
*token_text_range = TokenTextRange::Delimiter(cover);
2020-03-08 16:13:04 +00:00
}
}
2020-03-20 21:20:28 +00:00
fn remove_delim(&mut self, idx: usize) {
2021-01-08 14:46:48 +00:00
// FIXME: This could be accidentally quadratic
self.entries.remove(idx);
2020-03-20 21:20:28 +00:00
}
2019-11-03 14:46:12 +00:00
}
2019-05-04 07:00:16 +00:00
/// Returns the textual content of a doc comment block as a quoted string
/// That is, strips leading `///` (or `/**`, etc)
/// and strips the ending `*/`
/// And then quote the string, which is needed to convert to `tt::Literal`
fn doc_comment_text(comment: &ast::Comment) -> SmolStr {
let prefix_len = comment.prefix().len();
let mut text = &comment.text()[prefix_len..];
// Remove ending "*/"
if comment.kind().shape == ast::CommentShape::Block {
text = &text[0..text.len() - 2];
}
// Quote the string
// Note that `tt::Literal` expect an escaped string
let text = format!("\"{}\"", text.escape_debug());
2019-05-04 07:00:16 +00:00
text.into()
}
2020-08-12 16:26:51 +00:00
fn convert_doc_comment(token: &syntax::SyntaxToken) -> Option<Vec<tt::TokenTree>> {
2019-07-18 17:09:50 +00:00
let comment = ast::Comment::cast(token.clone())?;
2019-05-04 07:00:16 +00:00
let doc = comment.kind().doc?;
// Make `doc="\" Comments\""
let meta_tkns = vec![mk_ident("doc"), mk_punct('='), mk_doc_literal(&comment)];
2019-05-04 07:00:16 +00:00
// Make `#![]`
let mut token_trees = vec![mk_punct('#')];
2019-05-04 07:00:16 +00:00
if let ast::CommentPlacement::Inner = doc {
token_trees.push(mk_punct('!'));
}
2019-06-03 14:21:08 +00:00
token_trees.push(tt::TokenTree::from(tt::Subtree {
2019-12-12 17:41:44 +00:00
delimiter: Some(tt::Delimiter {
kind: tt::DelimiterKind::Bracket,
id: tt::TokenId::unspecified(),
}),
2019-06-03 14:21:08 +00:00
token_trees: meta_tkns,
}));
2019-05-04 07:00:16 +00:00
return Some(token_trees);
// Helper functions
fn mk_ident(s: &str) -> tt::TokenTree {
tt::TokenTree::from(tt::Leaf::from(tt::Ident {
text: s.into(),
id: tt::TokenId::unspecified(),
}))
}
fn mk_punct(c: char) -> tt::TokenTree {
2019-12-12 13:47:54 +00:00
tt::TokenTree::from(tt::Leaf::from(tt::Punct {
char: c,
spacing: tt::Spacing::Alone,
id: tt::TokenId::unspecified(),
}))
2019-05-04 07:00:16 +00:00
}
fn mk_doc_literal(comment: &ast::Comment) -> tt::TokenTree {
2019-12-12 13:47:54 +00:00
let lit = tt::Literal { text: doc_comment_text(comment), id: tt::TokenId::unspecified() };
2019-05-04 07:00:16 +00:00
tt::TokenTree::from(tt::Leaf::from(lit))
}
}
2020-03-08 16:13:04 +00:00
struct TokenIdAlloc {
2019-11-18 13:08:41 +00:00
map: TokenMap,
2020-04-24 21:40:41 +00:00
global_offset: TextSize,
2019-11-18 13:08:41 +00:00
next_id: u32,
}
2019-04-25 15:12:57 +00:00
2020-03-08 16:13:04 +00:00
impl TokenIdAlloc {
fn alloc(&mut self, absolute_range: TextRange) -> tt::TokenId {
let relative_range = absolute_range - self.global_offset;
let token_id = tt::TokenId(self.next_id);
self.next_id += 1;
self.map.insert(token_id, relative_range);
token_id
}
fn open_delim(&mut self, open_abs_range: TextRange) -> (tt::TokenId, usize) {
2020-03-08 16:13:04 +00:00
let token_id = tt::TokenId(self.next_id);
self.next_id += 1;
let idx = self.map.insert_delim(
2020-03-20 20:22:21 +00:00
token_id,
open_abs_range - self.global_offset,
open_abs_range - self.global_offset,
);
(token_id, idx)
2020-03-08 16:13:04 +00:00
}
fn close_delim(&mut self, idx: usize, close_abs_range: Option<TextRange>) {
2020-03-20 21:20:28 +00:00
match close_abs_range {
None => {
self.map.remove_delim(idx);
2020-03-20 21:20:28 +00:00
}
Some(close) => {
self.map.update_close_delim(idx, close - self.global_offset);
2020-03-20 21:20:28 +00:00
}
}
2020-03-08 16:13:04 +00:00
}
}
/// A Raw Token (straightly from lexer) convertor
struct RawConvertor<'a> {
text: &'a str,
2020-04-24 21:40:41 +00:00
offset: TextSize,
2020-03-08 16:13:04 +00:00
id_alloc: TokenIdAlloc,
2020-03-20 19:04:11 +00:00
inner: std::slice::Iter<'a, RawToken>,
2020-03-08 16:13:04 +00:00
}
2020-04-18 11:28:07 +00:00
trait SrcToken: std::fmt::Debug {
2020-03-20 19:04:11 +00:00
fn kind(&self) -> SyntaxKind;
fn to_char(&self) -> Option<char>;
fn to_text(&self) -> SmolStr;
2020-03-19 12:40:25 +00:00
}
2020-03-20 19:04:11 +00:00
trait TokenConvertor {
type Token: SrcToken;
2020-03-19 12:40:25 +00:00
fn go(&mut self) -> tt::Subtree {
2020-03-19 12:40:25 +00:00
let mut subtree = tt::Subtree::default();
subtree.delimiter = None;
while self.peek().is_some() {
self.collect_leaf(&mut subtree.token_trees);
}
if subtree.token_trees.len() == 1 {
if let tt::TokenTree::Subtree(first) = &subtree.token_trees[0] {
return first.clone();
2020-03-19 12:40:25 +00:00
}
}
subtree
2020-03-19 12:40:25 +00:00
}
fn collect_leaf(&mut self, result: &mut Vec<tt::TokenTree>) {
let (token, range) = match self.bump() {
None => return,
Some(it) => it,
};
let k: SyntaxKind = token.kind();
if k == COMMENT {
2020-03-20 19:04:11 +00:00
if let Some(tokens) = self.convert_doc_comment(&token) {
2020-03-19 12:40:25 +00:00
result.extend(tokens);
}
return;
}
result.push(if k.is_punct() && k != UNDERSCORE {
2020-04-24 21:40:41 +00:00
assert_eq!(range.len(), TextSize::of('.'));
2020-03-19 12:40:25 +00:00
let delim = match k {
T!['('] => Some((tt::DelimiterKind::Parenthesis, T![')'])),
T!['{'] => Some((tt::DelimiterKind::Brace, T!['}'])),
T!['['] => Some((tt::DelimiterKind::Bracket, T![']'])),
_ => None,
};
if let Some((kind, closed)) = delim {
let mut subtree = tt::Subtree::default();
let (id, idx) = self.id_alloc().open_delim(range);
2021-03-21 14:33:18 +00:00
subtree.delimiter = Some(tt::Delimiter { id, kind });
2020-03-19 12:40:25 +00:00
2020-03-20 19:04:11 +00:00
while self.peek().map(|it| it.kind() != closed).unwrap_or(false) {
2020-03-19 12:40:25 +00:00
self.collect_leaf(&mut subtree.token_trees);
}
let last_range = match self.bump() {
2020-03-20 21:20:28 +00:00
None => {
// For error resilience, we insert an char punct for the opening delim here
self.id_alloc().close_delim(idx, None);
2020-03-20 21:20:28 +00:00
let leaf: tt::Leaf = tt::Punct {
id: self.id_alloc().alloc(range),
char: token.to_char().unwrap(),
spacing: tt::Spacing::Alone,
}
.into();
result.push(leaf.into());
result.extend(subtree.token_trees);
return;
}
2020-03-19 12:40:25 +00:00
Some(it) => it.1,
};
self.id_alloc().close_delim(idx, Some(last_range));
2020-03-19 12:40:25 +00:00
subtree.into()
} else {
let spacing = match self.peek() {
Some(next)
2020-03-20 19:04:11 +00:00
if next.kind().is_trivia()
|| next.kind() == T!['[']
|| next.kind() == T!['{']
|| next.kind() == T!['('] =>
2020-03-19 12:40:25 +00:00
{
tt::Spacing::Alone
}
Some(next) if next.kind().is_punct() && next.kind() != UNDERSCORE => {
tt::Spacing::Joint
}
2020-03-19 12:40:25 +00:00
_ => tt::Spacing::Alone,
};
2020-04-18 11:28:07 +00:00
let char = match token.to_char() {
Some(c) => c,
None => {
panic!("Token from lexer must be single char: token = {:#?}", token);
}
};
2020-03-20 19:04:11 +00:00
tt::Leaf::from(tt::Punct { char, spacing, id: self.id_alloc().alloc(range) }).into()
2020-03-19 12:40:25 +00:00
}
} else {
macro_rules! make_leaf {
($i:ident) => {
2020-03-20 19:04:11 +00:00
tt::$i { id: self.id_alloc().alloc(range), text: token.to_text() }.into()
2020-03-19 12:40:25 +00:00
};
}
let leaf: tt::Leaf = match k {
T![true] | T![false] => make_leaf!(Ident),
2020-04-18 11:28:07 +00:00
IDENT => make_leaf!(Ident),
UNDERSCORE => make_leaf!(Ident),
2020-03-19 12:40:25 +00:00
k if k.is_keyword() => make_leaf!(Ident),
k if k.is_literal() => make_leaf!(Literal),
2020-12-15 18:23:51 +00:00
LIFETIME_IDENT => {
2020-04-24 21:40:41 +00:00
let char_unit = TextSize::of('\'');
let r = TextRange::at(range.start(), char_unit);
2020-04-18 11:28:07 +00:00
let apostrophe = tt::Leaf::from(tt::Punct {
char: '\'',
spacing: tt::Spacing::Joint,
id: self.id_alloc().alloc(r),
});
result.push(apostrophe.into());
2020-04-24 21:40:41 +00:00
let r = TextRange::at(range.start() + char_unit, range.len() - char_unit);
2020-04-18 11:28:07 +00:00
let ident = tt::Leaf::from(tt::Ident {
text: SmolStr::new(&token.to_text()[1..]),
id: self.id_alloc().alloc(r),
});
result.push(ident.into());
return;
}
2020-03-19 12:40:25 +00:00
_ => return,
};
leaf.into()
});
}
2020-03-20 19:04:11 +00:00
fn convert_doc_comment(&self, token: &Self::Token) -> Option<Vec<tt::TokenTree>>;
fn bump(&mut self) -> Option<(Self::Token, TextRange)>;
fn peek(&self) -> Option<Self::Token>;
fn id_alloc(&mut self) -> &mut TokenIdAlloc;
2020-03-19 12:40:25 +00:00
}
2021-01-04 16:11:56 +00:00
impl<'a> SrcToken for (&'a RawToken, &'a str) {
2020-03-20 19:04:11 +00:00
fn kind(&self) -> SyntaxKind {
self.0.kind
2020-03-08 16:13:04 +00:00
}
2020-03-20 19:04:11 +00:00
fn to_char(&self) -> Option<char> {
self.1.chars().next()
2020-03-08 16:13:04 +00:00
}
2020-03-20 19:04:11 +00:00
fn to_text(&self) -> SmolStr {
self.1.into()
2020-03-08 16:13:04 +00:00
}
2020-03-20 19:04:11 +00:00
}
2020-03-08 16:13:04 +00:00
2020-03-20 19:04:11 +00:00
impl RawConvertor<'_> {}
2020-03-08 16:13:04 +00:00
2020-03-20 19:04:11 +00:00
impl<'a> TokenConvertor for RawConvertor<'a> {
2021-01-04 16:11:56 +00:00
type Token = (&'a RawToken, &'a str);
2020-03-08 16:13:04 +00:00
2020-03-20 19:04:11 +00:00
fn convert_doc_comment(&self, token: &Self::Token) -> Option<Vec<tt::TokenTree>> {
convert_doc_comment(&doc_comment(token.1))
}
2020-03-08 16:13:04 +00:00
2020-03-20 19:04:11 +00:00
fn bump(&mut self) -> Option<(Self::Token, TextRange)> {
let token = self.inner.next()?;
2020-04-24 21:40:41 +00:00
let range = TextRange::at(self.offset, token.len);
2020-03-20 19:04:11 +00:00
self.offset += token.len;
2020-03-08 16:13:04 +00:00
2021-01-04 16:11:56 +00:00
Some(((token, &self.text[range]), range))
2020-03-20 19:04:11 +00:00
}
2020-03-08 16:13:04 +00:00
2020-03-20 19:04:11 +00:00
fn peek(&self) -> Option<Self::Token> {
2021-01-04 16:11:56 +00:00
let token = self.inner.as_slice().get(0);
2020-03-08 16:13:04 +00:00
2020-03-20 19:04:11 +00:00
token.map(|it| {
2020-04-24 21:40:41 +00:00
let range = TextRange::at(self.offset, it.len);
2020-03-20 19:04:11 +00:00
(it, &self.text[range])
})
}
fn id_alloc(&mut self) -> &mut TokenIdAlloc {
&mut self.id_alloc
2020-03-08 16:13:04 +00:00
}
}
struct Convertor {
id_alloc: TokenIdAlloc,
2020-03-20 19:04:11 +00:00
current: Option<SyntaxToken>,
range: TextRange,
2020-04-24 21:40:41 +00:00
punct_offset: Option<(SyntaxToken, TextSize)>,
2020-03-08 16:13:04 +00:00
}
2019-11-18 13:08:41 +00:00
impl Convertor {
2020-04-24 21:40:41 +00:00
fn new(node: &SyntaxNode, global_offset: TextSize) -> Convertor {
2020-03-20 19:04:11 +00:00
Convertor {
id_alloc: { TokenIdAlloc { map: TokenMap::default(), global_offset, next_id: 0 } },
current: node.first_token(),
range: node.text_range(),
punct_offset: None,
2019-11-18 13:08:41 +00:00
}
2020-03-20 19:04:11 +00:00
}
}
2020-04-18 11:28:07 +00:00
#[derive(Debug)]
2020-03-20 19:04:11 +00:00
enum SynToken {
2021-01-08 14:41:46 +00:00
Ordinary(SyntaxToken),
2020-04-24 21:40:41 +00:00
Punch(SyntaxToken, TextSize),
2020-03-20 19:04:11 +00:00
}
2019-11-21 15:56:01 +00:00
2020-03-20 19:04:11 +00:00
impl SynToken {
fn token(&self) -> &SyntaxToken {
match self {
2021-01-08 14:41:46 +00:00
SynToken::Ordinary(it) => it,
2020-03-20 19:04:11 +00:00
SynToken::Punch(it, _) => it,
2019-11-21 15:56:01 +00:00
}
2020-03-20 19:04:11 +00:00
}
}
2019-11-21 15:56:01 +00:00
2020-03-20 19:04:11 +00:00
impl SrcToken for SynToken {
fn kind(&self) -> SyntaxKind {
self.token().kind()
}
fn to_char(&self) -> Option<char> {
match self {
2021-01-08 14:41:46 +00:00
SynToken::Ordinary(_) => None,
2020-04-24 21:40:41 +00:00
SynToken::Punch(it, i) => it.text().chars().nth((*i).into()),
2020-03-20 19:04:11 +00:00
}
}
fn to_text(&self) -> SmolStr {
2021-01-19 22:56:11 +00:00
self.token().text().into()
2020-03-20 19:04:11 +00:00
}
}
impl TokenConvertor for Convertor {
type Token = SynToken;
fn convert_doc_comment(&self, token: &Self::Token) -> Option<Vec<tt::TokenTree>> {
convert_doc_comment(token.token())
}
fn bump(&mut self) -> Option<(Self::Token, TextRange)> {
if let Some((punct, offset)) = self.punct_offset.clone() {
2020-04-24 21:40:41 +00:00
if usize::from(offset) + 1 < punct.text().len() {
2020-04-24 22:57:47 +00:00
let offset = offset + TextSize::of('.');
2020-03-20 19:04:11 +00:00
let range = punct.text_range();
2020-03-20 20:22:21 +00:00
self.punct_offset = Some((punct.clone(), offset));
2020-04-24 21:40:41 +00:00
let range = TextRange::at(range.start() + offset, TextSize::of('.'));
2020-03-20 20:22:21 +00:00
return Some((SynToken::Punch(punct, offset), range));
2019-11-21 15:56:01 +00:00
}
2020-03-20 19:04:11 +00:00
}
2020-03-20 20:22:21 +00:00
let curr = self.current.clone()?;
2020-04-24 21:40:41 +00:00
if !&self.range.contains_range(curr.text_range()) {
2020-03-20 20:22:21 +00:00
return None;
}
2020-03-20 19:04:11 +00:00
self.current = curr.next_token();
let token = if curr.kind().is_punct() {
let range = curr.text_range();
2020-04-24 22:57:47 +00:00
let range = TextRange::at(range.start(), TextSize::of('.'));
self.punct_offset = Some((curr.clone(), 0.into()));
(SynToken::Punch(curr, 0.into()), range)
2020-03-20 19:04:11 +00:00
} else {
self.punct_offset = None;
let range = curr.text_range();
2021-01-08 14:41:46 +00:00
(SynToken::Ordinary(curr), range)
2019-11-18 13:08:41 +00:00
};
2019-04-24 15:01:32 +00:00
2020-03-20 19:04:11 +00:00
Some(token)
}
2019-05-04 07:00:16 +00:00
2020-03-20 19:04:11 +00:00
fn peek(&self) -> Option<Self::Token> {
if let Some((punct, mut offset)) = self.punct_offset.clone() {
2020-04-24 22:57:47 +00:00
offset = offset + TextSize::of('.');
2020-04-24 21:40:41 +00:00
if usize::from(offset) < punct.text().len() {
2020-03-20 19:04:11 +00:00
return Some(SynToken::Punch(punct, offset));
}
2019-11-18 13:08:41 +00:00
}
2020-03-20 20:22:21 +00:00
let curr = self.current.clone()?;
2020-04-24 21:40:41 +00:00
if !self.range.contains_range(curr.text_range()) {
2020-03-20 20:22:21 +00:00
return None;
}
2020-03-20 19:04:11 +00:00
let token = if curr.kind().is_punct() {
2020-04-24 22:57:47 +00:00
SynToken::Punch(curr, 0.into())
2020-03-20 19:04:11 +00:00
} else {
2021-01-08 14:41:46 +00:00
SynToken::Ordinary(curr)
2020-03-20 19:04:11 +00:00
};
Some(token)
}
fn id_alloc(&mut self) -> &mut TokenIdAlloc {
&mut self.id_alloc
2019-01-31 18:29:04 +00:00
}
}
2019-02-23 11:18:32 +00:00
2019-05-27 14:56:21 +00:00
struct TtTreeSink<'a> {
2019-02-23 14:21:56 +00:00
buf: String,
2019-05-27 14:56:21 +00:00
cursor: Cursor<'a>,
2020-04-24 21:40:41 +00:00
open_delims: FxHashMap<tt::TokenId, TextSize>,
text_pos: TextSize,
2019-02-23 14:21:56 +00:00
inner: SyntaxTreeBuilder,
2019-11-18 13:08:41 +00:00
token_map: TokenMap,
// Number of roots
// Use for detect ill-form tree which is not single root
roots: smallvec::SmallVec<[usize; 1]>,
2019-02-23 14:21:56 +00:00
}
2019-05-27 14:56:21 +00:00
impl<'a> TtTreeSink<'a> {
2019-11-03 19:12:19 +00:00
fn new(cursor: Cursor<'a>) -> Self {
2019-02-23 14:21:56 +00:00
TtTreeSink {
buf: String::new(),
2019-05-27 14:56:21 +00:00
cursor,
2019-12-12 17:41:44 +00:00
open_delims: FxHashMap::default(),
2019-02-23 14:21:56 +00:00
text_pos: 0.into(),
inner: SyntaxTreeBuilder::default(),
roots: smallvec::SmallVec::new(),
2019-11-18 13:08:41 +00:00
token_map: TokenMap::default(),
2019-02-23 14:21:56 +00:00
}
}
2019-11-03 19:12:19 +00:00
fn finish(mut self) -> (Parse<SyntaxNode>, TokenMap) {
self.token_map.entries.shrink_to_fit();
2019-11-18 13:08:41 +00:00
(self.inner.finish(), self.token_map)
2019-11-03 19:12:19 +00:00
}
2019-02-23 14:21:56 +00:00
}
2021-01-04 16:11:56 +00:00
fn delim_to_str(d: Option<tt::DelimiterKind>, closing: bool) -> &'static str {
2019-12-18 03:47:26 +00:00
let texts = match d {
2019-12-12 17:41:44 +00:00
Some(tt::DelimiterKind::Parenthesis) => "()",
Some(tt::DelimiterKind::Brace) => "{}",
Some(tt::DelimiterKind::Bracket) => "[]",
2021-01-04 16:11:56 +00:00
None => return "",
2019-05-27 14:56:21 +00:00
};
let idx = closing as usize;
2021-01-04 16:11:56 +00:00
&texts[idx..texts.len() - (1 - idx)]
}
2019-05-27 14:56:21 +00:00
impl<'a> TreeSink for TtTreeSink<'a> {
2020-04-18 11:28:07 +00:00
fn token(&mut self, kind: SyntaxKind, mut n_tokens: u8) {
if kind == L_DOLLAR || kind == R_DOLLAR {
2019-05-27 16:28:46 +00:00
self.cursor = self.cursor.bump_subtree();
return;
}
2020-12-15 18:23:51 +00:00
if kind == LIFETIME_IDENT {
2020-04-18 11:28:07 +00:00
n_tokens = 2;
}
2020-03-04 15:38:58 +00:00
let mut last = self.cursor;
2019-02-23 14:21:56 +00:00
for _ in 0..n_tokens {
2021-01-04 16:11:56 +00:00
let tmp_str: SmolStr;
2019-05-27 14:56:21 +00:00
if self.cursor.eof() {
break;
}
2020-03-04 15:38:58 +00:00
last = self.cursor;
2021-01-04 16:11:56 +00:00
let text: &str = match self.cursor.token_tree() {
2021-01-04 16:22:42 +00:00
Some(tt::buffer::TokenTreeRef::Leaf(leaf, _)) => {
2019-11-03 17:45:58 +00:00
// Mark the range if needed
let (text, id) = match leaf {
2021-01-04 16:11:56 +00:00
tt::Leaf::Ident(ident) => (&ident.text, ident.id),
tt::Leaf::Punct(punct) => {
2020-09-23 06:45:35 +00:00
assert!(punct.char.is_ascii());
let char = &(punct.char as u8);
2021-01-04 16:11:56 +00:00
tmp_str = SmolStr::new_inline(
std::str::from_utf8(std::slice::from_ref(char)).unwrap(),
);
(&tmp_str, punct.id)
}
2021-01-04 16:11:56 +00:00
tt::Leaf::Literal(lit) => (&lit.text, lit.id),
2019-12-12 17:41:44 +00:00
};
2020-04-24 21:40:41 +00:00
let range = TextRange::at(self.text_pos, TextSize::of(text.as_str()));
2019-12-12 17:41:44 +00:00
self.token_map.insert(id, range);
2019-11-03 17:45:58 +00:00
self.cursor = self.cursor.bump();
2019-12-18 03:47:26 +00:00
text
2019-05-27 14:56:21 +00:00
}
2021-01-04 16:22:42 +00:00
Some(tt::buffer::TokenTreeRef::Subtree(subtree, _)) => {
2019-05-27 14:56:21 +00:00
self.cursor = self.cursor.subtree().unwrap();
2019-12-12 17:41:44 +00:00
if let Some(id) = subtree.delimiter.map(|it| it.id) {
self.open_delims.insert(id, self.text_pos);
2019-05-27 14:56:21 +00:00
}
2019-12-18 03:47:26 +00:00
delim_to_str(subtree.delimiter_kind(), false)
2019-05-27 14:56:21 +00:00
}
2019-12-18 03:47:26 +00:00
None => {
if let Some(parent) = self.cursor.end() {
self.cursor = self.cursor.bump();
if let Some(id) = parent.delimiter.map(|it| it.id) {
if let Some(open_delim) = self.open_delims.get(&id) {
2020-04-24 21:40:41 +00:00
let open_range = TextRange::at(*open_delim, TextSize::of('('));
let close_range = TextRange::at(self.text_pos, TextSize::of('('));
2019-12-18 03:47:26 +00:00
self.token_map.insert_delim(id, open_range, close_range);
}
2019-12-12 17:41:44 +00:00
}
2019-12-18 03:47:26 +00:00
delim_to_str(parent.delimiter_kind(), true)
} else {
continue;
2019-12-12 17:41:44 +00:00
}
2019-12-18 03:47:26 +00:00
}
2019-05-27 14:56:21 +00:00
};
2019-12-18 03:47:26 +00:00
self.buf += &text;
2021-01-04 16:11:56 +00:00
self.text_pos += TextSize::of(text);
2019-02-23 14:21:56 +00:00
}
2019-05-27 14:56:21 +00:00
self.inner.token(kind, self.buf.as_str());
2019-02-23 14:21:56 +00:00
self.buf.clear();
2019-05-27 14:56:21 +00:00
// Add whitespace between adjoint puncts
2020-03-04 15:38:58 +00:00
let next = last.bump();
2019-05-27 14:56:21 +00:00
if let (
2021-01-04 16:22:42 +00:00
Some(tt::buffer::TokenTreeRef::Leaf(tt::Leaf::Punct(curr), _)),
Some(tt::buffer::TokenTreeRef::Leaf(tt::Leaf::Punct(_), _)),
2020-03-04 15:38:58 +00:00
) = (last.token_tree(), next.token_tree())
2019-05-27 14:56:21 +00:00
{
2020-03-05 20:32:08 +00:00
// Note: We always assume the semi-colon would be the last token in
// other parts of RA such that we don't add whitespace here.
if curr.spacing == tt::Spacing::Alone && curr.char != ';' {
self.inner.token(WHITESPACE, " ");
2020-04-24 21:40:41 +00:00
self.text_pos += TextSize::of(' ');
}
}
2019-02-23 14:21:56 +00:00
}
2019-03-30 10:25:53 +00:00
fn start_node(&mut self, kind: SyntaxKind) {
self.inner.start_node(kind);
match self.roots.last_mut() {
None | Some(0) => self.roots.push(1),
Some(ref mut n) => **n += 1,
};
2019-02-23 14:21:56 +00:00
}
2019-03-30 10:25:53 +00:00
fn finish_node(&mut self) {
self.inner.finish_node();
*self.roots.last_mut().unwrap() -= 1;
2019-02-23 14:21:56 +00:00
}
fn error(&mut self, error: ParseError) {
self.inner.error(error, self.text_pos)
}
}
#[cfg(test)]
mod tests {
use super::*;
2019-12-19 15:17:22 +00:00
use crate::tests::parse_macro;
2020-08-12 15:06:49 +00:00
use parser::TokenSource;
2020-08-12 16:26:51 +00:00
use syntax::{
algo::{insert_children, InsertPosition},
ast::AstNode,
};
#[test]
fn convert_tt_token_source() {
2019-12-19 15:17:22 +00:00
let expansion = parse_macro(
r#"
macro_rules! literals {
($i:ident) => {
{
let a = 'c';
let c = 1000;
let f = 12E+99_f64;
let s = "rust1";
}
}
}
"#,
2019-12-19 15:17:22 +00:00
)
.expand_tt("literals!(foo);");
2019-10-30 17:36:37 +00:00
let tts = &[expansion.into()];
2021-01-04 16:22:42 +00:00
let buffer = tt::buffer::TokenBuffer::from_tokens(tts);
2019-05-27 14:56:21 +00:00
let mut tt_src = SubtreeTokenSource::new(&buffer);
let mut tokens = vec![];
while tt_src.current().kind != EOF {
tokens.push((tt_src.current().kind, tt_src.text()));
tt_src.bump();
}
// [${]
// [let] [a] [=] ['c'] [;]
2019-05-27 14:56:21 +00:00
assert_eq!(tokens[2 + 3].1, "'c'");
assert_eq!(tokens[2 + 3].0, CHAR);
// [let] [c] [=] [1000] [;]
2019-05-27 14:56:21 +00:00
assert_eq!(tokens[2 + 5 + 3].1, "1000");
assert_eq!(tokens[2 + 5 + 3].0, INT_NUMBER);
// [let] [f] [=] [12E+99_f64] [;]
2019-05-27 14:56:21 +00:00
assert_eq!(tokens[2 + 10 + 3].1, "12E+99_f64");
assert_eq!(tokens[2 + 10 + 3].0, FLOAT_NUMBER);
// [let] [s] [=] ["rust1"] [;]
2019-05-27 14:56:21 +00:00
assert_eq!(tokens[2 + 15 + 3].1, "\"rust1\"");
assert_eq!(tokens[2 + 15 + 3].0, STRING);
}
#[test]
fn stmts_token_trees_to_expr_is_err() {
2019-12-19 15:17:22 +00:00
let expansion = parse_macro(
r#"
macro_rules! stmts {
() => {
let a = 0;
let b = 0;
let c = 0;
let d = 0;
}
}
"#,
2019-12-19 15:17:22 +00:00
)
.expand_tt("stmts!();");
assert!(token_tree_to_syntax_node(&expansion, FragmentKind::Expr).is_err());
}
2019-11-21 15:56:01 +00:00
#[test]
fn test_token_tree_last_child_is_white_space() {
let source_file = ast::SourceFile::parse("f!({} );").ok().unwrap();
let macro_call = source_file.syntax().descendants().find_map(ast::MacroCall::cast).unwrap();
let token_tree = macro_call.token_tree().unwrap();
// Token Tree now is :
// TokenTree
// - T!['(']
// - TokenTree
// - T!['{']
// - T!['}']
// - WHITE_SPACE
// - T![')']
let rbrace =
token_tree.syntax().descendants_with_tokens().find(|it| it.kind() == T!['}']).unwrap();
let space = token_tree
.syntax()
.descendants_with_tokens()
.find(|it| it.kind() == SyntaxKind::WHITESPACE)
.unwrap();
// reorder th white space, such that the white is inside the inner token-tree.
let token_tree = insert_children(
&rbrace.parent().unwrap(),
InsertPosition::Last,
2020-02-29 12:49:43 +00:00
std::iter::once(space),
2019-11-21 15:56:01 +00:00
);
// Token Tree now is :
// TokenTree
// - T!['{']
// - T!['}']
// - WHITE_SPACE
let token_tree = ast::TokenTree::cast(token_tree).unwrap();
let tt = ast_to_token_tree(&token_tree).0;
2019-11-21 15:56:01 +00:00
2019-12-18 03:47:26 +00:00
assert_eq!(tt.delimiter_kind(), Some(tt::DelimiterKind::Brace));
2019-11-21 15:56:01 +00:00
}
#[test]
fn test_token_tree_multi_char_punct() {
let source_file = ast::SourceFile::parse("struct Foo { a: x::Y }").ok().unwrap();
2020-07-30 15:50:40 +00:00
let struct_def = source_file.syntax().descendants().find_map(ast::Struct::cast).unwrap();
let tt = ast_to_token_tree(&struct_def).0;
token_tree_to_syntax_node(&tt, FragmentKind::Item).unwrap();
}
}