2019-09-30 03:58:53 -05:00
|
|
|
//! FIXME: write short doc here
|
|
|
|
|
2019-11-09 04:14:10 -06:00
|
|
|
use ra_parser::{FragmentKind, ParseError, TreeSink};
|
2019-02-11 12:12:06 -06:00
|
|
|
use ra_syntax::{
|
2020-03-08 11:13:04 -05:00
|
|
|
ast::{self, make::tokens::doc_comment},
|
|
|
|
tokenize, AstToken, NodeOrToken, Parse, SmolStr, SyntaxKind,
|
|
|
|
SyntaxKind::*,
|
|
|
|
SyntaxNode, SyntaxTreeBuilder, TextRange, TextUnit, Token, T,
|
2019-02-11 12:12:06 -06:00
|
|
|
};
|
2019-12-12 11:41:44 -06:00
|
|
|
use rustc_hash::FxHashMap;
|
2019-11-21 09:56:01 -06:00
|
|
|
use std::iter::successors;
|
2019-07-04 15:05:17 -05:00
|
|
|
use tt::buffer::{Cursor, TokenBuffer};
|
2019-04-07 08:42:53 -05:00
|
|
|
|
2019-07-20 12:04:34 -05:00
|
|
|
use crate::subtree_source::SubtreeTokenSource;
|
|
|
|
use crate::ExpandError;
|
|
|
|
|
2019-12-13 09:55:51 -06:00
|
|
|
#[derive(Debug, PartialEq, Eq, Clone, Copy)]
|
|
|
|
pub enum TokenTextRange {
|
|
|
|
Token(TextRange),
|
|
|
|
Delimiter(TextRange, TextRange),
|
|
|
|
}
|
|
|
|
|
|
|
|
impl TokenTextRange {
|
2019-12-17 21:36:10 -06:00
|
|
|
pub fn by_kind(self, kind: SyntaxKind) -> Option<TextRange> {
|
2019-12-13 09:55:51 -06:00
|
|
|
match self {
|
|
|
|
TokenTextRange::Token(it) => Some(it),
|
|
|
|
TokenTextRange::Delimiter(open, close) => match kind {
|
|
|
|
T!['{'] | T!['('] | T!['['] => Some(open),
|
|
|
|
T!['}'] | T![')'] | T![']'] => Some(close),
|
|
|
|
_ => None,
|
|
|
|
},
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-02-11 12:31:54 -06:00
|
|
|
/// Maps `tt::TokenId` to the relative range of the original token.
|
2019-11-04 12:09:16 -06:00
|
|
|
#[derive(Debug, PartialEq, Eq, Default)]
|
2019-02-11 12:12:06 -06:00
|
|
|
pub struct TokenMap {
|
|
|
|
/// Maps `tt::TokenId` to the *relative* source range.
|
2019-12-13 09:55:51 -06:00
|
|
|
entries: Vec<(tt::TokenId, TokenTextRange)>,
|
2019-11-03 08:46:12 -06:00
|
|
|
}
|
|
|
|
|
2019-02-11 12:31:54 -06:00
|
|
|
/// Convert the syntax tree (what user has written) to a `TokenTree` (what macro
|
|
|
|
/// will consume).
|
2019-12-05 08:10:33 -06:00
|
|
|
pub fn ast_to_token_tree(ast: &impl ast::AstNode) -> Option<(tt::Subtree, TokenMap)> {
|
2019-11-18 07:08:41 -06:00
|
|
|
syntax_node_to_token_tree(ast.syntax())
|
2019-02-11 12:12:06 -06:00
|
|
|
}
|
|
|
|
|
2019-04-12 12:50:05 -05:00
|
|
|
/// Convert the syntax node to a `TokenTree` (what macro
|
|
|
|
/// will consume).
|
|
|
|
pub fn syntax_node_to_token_tree(node: &SyntaxNode) -> Option<(tt::Subtree, TokenMap)> {
|
2019-11-18 07:08:41 -06:00
|
|
|
let global_offset = node.text_range().start();
|
2020-03-08 11:13:04 -05:00
|
|
|
let mut c = Convertor {
|
|
|
|
id_alloc: { TokenIdAlloc { map: TokenMap::default(), global_offset, next_id: 0 } },
|
|
|
|
};
|
2019-11-18 07:08:41 -06:00
|
|
|
let subtree = c.go(node)?;
|
2020-03-08 11:13:04 -05:00
|
|
|
Some((subtree, c.id_alloc.map))
|
2019-04-12 12:50:05 -05:00
|
|
|
}
|
|
|
|
|
2019-04-18 13:47:29 -05:00
|
|
|
// 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 14:49:56 -05:00
|
|
|
// * 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
|
2019-04-18 13:47:29 -05:00
|
|
|
//
|
|
|
|
// * TraitItems(SmallVec<[ast::TraitItem; 1]>)
|
|
|
|
// * ImplItems(SmallVec<[ast::ImplItem; 1]>)
|
|
|
|
// * ForeignItems(SmallVec<[ast::ForeignItem; 1]>
|
2019-05-25 08:55:46 -05:00
|
|
|
|
2019-11-09 04:14:10 -06:00
|
|
|
pub fn token_tree_to_syntax_node(
|
2019-09-02 10:51:03 -05:00
|
|
|
tt: &tt::Subtree,
|
|
|
|
fragment_kind: FragmentKind,
|
2019-11-18 07:08:41 -06:00
|
|
|
) -> Result<(Parse<SyntaxNode>, TokenMap), ExpandError> {
|
2019-09-10 06:27:08 -05:00
|
|
|
let tmp;
|
|
|
|
let tokens = match tt {
|
2019-12-13 07:53:34 -06:00
|
|
|
tt::Subtree { delimiter: None, token_trees } => token_trees.as_slice(),
|
2019-09-10 06:27:08 -05:00
|
|
|
_ => {
|
|
|
|
tmp = [tt.clone().into()];
|
|
|
|
&tmp[..]
|
|
|
|
}
|
|
|
|
};
|
2019-09-02 10:51:03 -05:00
|
|
|
let buffer = TokenBuffer::new(&tokens);
|
|
|
|
let mut token_source = SubtreeTokenSource::new(&buffer);
|
2019-11-03 13:12:19 -06:00
|
|
|
let mut tree_sink = TtTreeSink::new(buffer.begin());
|
2019-09-02 10:51:03 -05:00
|
|
|
ra_parser::parse_fragment(&mut token_source, &mut tree_sink, fragment_kind);
|
|
|
|
if tree_sink.roots.len() != 1 {
|
|
|
|
return Err(ExpandError::ConversionError);
|
|
|
|
}
|
|
|
|
//FIXME: would be cool to report errors
|
2019-11-03 13:12:19 -06:00
|
|
|
let (parse, range_map) = tree_sink.finish();
|
2019-11-03 08:46:12 -06:00
|
|
|
Ok((parse, range_map))
|
2019-09-02 10:51:03 -05:00
|
|
|
}
|
|
|
|
|
2020-03-08 11:13:04 -05:00
|
|
|
/// 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,
|
|
|
|
offset: TextUnit::default(),
|
|
|
|
inner: tokens.iter(),
|
|
|
|
id_alloc: TokenIdAlloc {
|
|
|
|
map: Default::default(),
|
|
|
|
global_offset: TextUnit::default(),
|
|
|
|
next_id: 0,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
let subtree = conv.go()?;
|
|
|
|
Some((subtree, conv.id_alloc.map))
|
|
|
|
}
|
|
|
|
|
2019-02-11 12:12:06 -06:00
|
|
|
impl TokenMap {
|
2019-11-17 11:15:55 -06:00
|
|
|
pub fn token_by_range(&self, relative_range: TextRange) -> Option<tt::TokenId> {
|
2019-12-13 09:55:51 -06:00
|
|
|
let &(token_id, _) = self.entries.iter().find(|(_, range)| match range {
|
|
|
|
TokenTextRange::Token(it) => *it == relative_range,
|
|
|
|
TokenTextRange::Delimiter(open, close) => {
|
|
|
|
*open == relative_range || *close == relative_range
|
|
|
|
}
|
|
|
|
})?;
|
2019-11-18 07:08:41 -06:00
|
|
|
Some(token_id)
|
2019-11-18 06:08:39 -06:00
|
|
|
}
|
|
|
|
|
2019-12-13 09:55:51 -06:00
|
|
|
pub fn range_by_token(&self, token_id: tt::TokenId) -> Option<TokenTextRange> {
|
2019-11-18 07:08:41 -06:00
|
|
|
let &(_, range) = self.entries.iter().find(|(tid, _)| *tid == token_id)?;
|
|
|
|
Some(range)
|
2019-11-17 11:15:55 -06:00
|
|
|
}
|
|
|
|
|
2019-11-18 07:08:41 -06:00
|
|
|
fn insert(&mut self, token_id: tt::TokenId, relative_range: TextRange) {
|
2019-12-13 09:55:51 -06:00
|
|
|
self.entries.push((token_id, TokenTextRange::Token(relative_range)));
|
2019-11-03 08:46:12 -06:00
|
|
|
}
|
2019-12-12 11:41:44 -06:00
|
|
|
|
|
|
|
fn insert_delim(
|
|
|
|
&mut self,
|
2019-12-13 09:55:51 -06:00
|
|
|
token_id: tt::TokenId,
|
|
|
|
open_relative_range: TextRange,
|
|
|
|
close_relative_range: TextRange,
|
2019-12-12 11:41:44 -06:00
|
|
|
) {
|
2019-12-13 09:55:51 -06:00
|
|
|
self.entries
|
|
|
|
.push((token_id, TokenTextRange::Delimiter(open_relative_range, close_relative_range)));
|
2019-12-12 11:41:44 -06:00
|
|
|
}
|
2020-03-08 11:13:04 -05:00
|
|
|
|
|
|
|
fn update_close_delim(&mut self, token_id: tt::TokenId, close_relative_range: TextRange) {
|
|
|
|
if let Some(entry) = self.entries.iter_mut().find(|(tid, _)| *tid == token_id) {
|
|
|
|
if let TokenTextRange::Delimiter(dim, _) = entry.1 {
|
|
|
|
entry.1 = TokenTextRange::Delimiter(dim, close_relative_range);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2019-11-03 08:46:12 -06:00
|
|
|
}
|
|
|
|
|
2019-05-04 02:00:16 -05:00
|
|
|
/// 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
|
2019-05-04 06:34:02 -05:00
|
|
|
let text = format!("{:?}", text.escape_default().to_string());
|
2019-05-04 02:00:16 -05:00
|
|
|
text.into()
|
|
|
|
}
|
|
|
|
|
2019-07-18 12:09:50 -05:00
|
|
|
fn convert_doc_comment(token: &ra_syntax::SyntaxToken) -> Option<Vec<tt::TokenTree>> {
|
|
|
|
let comment = ast::Comment::cast(token.clone())?;
|
2019-05-04 02:00:16 -05:00
|
|
|
let doc = comment.kind().doc?;
|
|
|
|
|
|
|
|
// Make `doc="\" Comments\""
|
|
|
|
let mut meta_tkns = Vec::new();
|
|
|
|
meta_tkns.push(mk_ident("doc"));
|
|
|
|
meta_tkns.push(mk_punct('='));
|
|
|
|
meta_tkns.push(mk_doc_literal(&comment));
|
|
|
|
|
|
|
|
// Make `#![]`
|
|
|
|
let mut token_trees = Vec::new();
|
|
|
|
token_trees.push(mk_punct('#'));
|
|
|
|
if let ast::CommentPlacement::Inner = doc {
|
|
|
|
token_trees.push(mk_punct('!'));
|
|
|
|
}
|
2019-06-03 09:21:08 -05:00
|
|
|
token_trees.push(tt::TokenTree::from(tt::Subtree {
|
2019-12-12 11:41:44 -06:00
|
|
|
delimiter: Some(tt::Delimiter {
|
|
|
|
kind: tt::DelimiterKind::Bracket,
|
|
|
|
id: tt::TokenId::unspecified(),
|
|
|
|
}),
|
2019-06-03 09:21:08 -05:00
|
|
|
token_trees: meta_tkns,
|
|
|
|
}));
|
2019-05-04 02:00:16 -05:00
|
|
|
|
|
|
|
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 07:47:54 -06:00
|
|
|
tt::TokenTree::from(tt::Leaf::from(tt::Punct {
|
|
|
|
char: c,
|
|
|
|
spacing: tt::Spacing::Alone,
|
|
|
|
id: tt::TokenId::unspecified(),
|
|
|
|
}))
|
2019-05-04 02:00:16 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
fn mk_doc_literal(comment: &ast::Comment) -> tt::TokenTree {
|
2019-12-12 07:47:54 -06:00
|
|
|
let lit = tt::Literal { text: doc_comment_text(comment), id: tt::TokenId::unspecified() };
|
2019-05-04 02:00:16 -05:00
|
|
|
|
|
|
|
tt::TokenTree::from(tt::Leaf::from(lit))
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-03-08 11:13:04 -05:00
|
|
|
struct TokenIdAlloc {
|
2019-11-18 07:08:41 -06:00
|
|
|
map: TokenMap,
|
2019-02-11 12:12:06 -06:00
|
|
|
global_offset: TextUnit,
|
2019-11-18 07:08:41 -06:00
|
|
|
next_id: u32,
|
|
|
|
}
|
2019-04-25 10:12:57 -05:00
|
|
|
|
2020-03-08 11:13:04 -05:00
|
|
|
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 delim(&mut self, open_abs_range: TextRange, close_abs_range: TextRange) -> tt::TokenId {
|
|
|
|
let open_relative_range = open_abs_range - self.global_offset;
|
|
|
|
let close_relative_range = close_abs_range - self.global_offset;
|
|
|
|
let token_id = tt::TokenId(self.next_id);
|
|
|
|
self.next_id += 1;
|
|
|
|
|
|
|
|
self.map.insert_delim(token_id, open_relative_range, close_relative_range);
|
|
|
|
token_id
|
|
|
|
}
|
|
|
|
|
|
|
|
fn open_delim(&mut self, open_abs_range: TextRange) -> tt::TokenId {
|
|
|
|
let token_id = tt::TokenId(self.next_id);
|
|
|
|
self.next_id += 1;
|
|
|
|
self.map.insert_delim(token_id, open_abs_range, open_abs_range);
|
|
|
|
token_id
|
|
|
|
}
|
|
|
|
|
|
|
|
fn close_delim(&mut self, id: tt::TokenId, close_abs_range: TextRange) {
|
|
|
|
self.map.update_close_delim(id, close_abs_range);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/// A Raw Token (straightly from lexer) convertor
|
|
|
|
struct RawConvertor<'a> {
|
|
|
|
text: &'a str,
|
|
|
|
offset: TextUnit,
|
|
|
|
id_alloc: TokenIdAlloc,
|
|
|
|
inner: std::slice::Iter<'a, Token>,
|
|
|
|
}
|
|
|
|
|
|
|
|
impl RawConvertor<'_> {
|
|
|
|
fn go(&mut self) -> Option<tt::Subtree> {
|
|
|
|
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.is_empty() {
|
|
|
|
return None;
|
|
|
|
}
|
|
|
|
if subtree.token_trees.len() == 1 {
|
|
|
|
if let tt::TokenTree::Subtree(first) = &subtree.token_trees[0] {
|
|
|
|
return Some(first.clone());
|
|
|
|
}
|
|
|
|
}
|
|
|
|
Some(subtree)
|
|
|
|
}
|
|
|
|
|
|
|
|
fn bump(&mut self) -> Option<(Token, TextRange)> {
|
|
|
|
let token = self.inner.next()?;
|
|
|
|
let range = TextRange::offset_len(self.offset, token.len);
|
|
|
|
self.offset += token.len;
|
|
|
|
Some((*token, range))
|
|
|
|
}
|
|
|
|
|
|
|
|
fn peek(&self) -> Option<Token> {
|
|
|
|
self.inner.as_slice().get(0).cloned()
|
|
|
|
}
|
|
|
|
|
|
|
|
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 {
|
|
|
|
let node = doc_comment(&self.text[range]);
|
|
|
|
if let Some(tokens) = convert_doc_comment(&node) {
|
|
|
|
result.extend(tokens);
|
|
|
|
}
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
result.push(if k.is_punct() {
|
|
|
|
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 = self.id_alloc.open_delim(range);
|
|
|
|
subtree.delimiter = Some(tt::Delimiter { kind, id });
|
|
|
|
|
|
|
|
while self.peek().map(|it| it.kind != closed).unwrap_or(false) {
|
|
|
|
self.collect_leaf(&mut subtree.token_trees);
|
|
|
|
}
|
|
|
|
let last_range = match self.bump() {
|
|
|
|
None => return,
|
|
|
|
Some(it) => it.1,
|
|
|
|
};
|
|
|
|
self.id_alloc.close_delim(id, last_range);
|
|
|
|
subtree.into()
|
|
|
|
} else {
|
|
|
|
let spacing = match self.peek() {
|
|
|
|
Some(next)
|
|
|
|
if next.kind.is_trivia()
|
|
|
|
|| next.kind == T!['[']
|
|
|
|
|| next.kind == T!['{']
|
|
|
|
|| next.kind == T!['('] =>
|
|
|
|
{
|
|
|
|
tt::Spacing::Alone
|
|
|
|
}
|
|
|
|
Some(next) if next.kind.is_punct() => tt::Spacing::Joint,
|
|
|
|
_ => tt::Spacing::Alone,
|
|
|
|
};
|
|
|
|
let char =
|
|
|
|
self.text[range].chars().next().expect("Token from lexer must be single char");
|
|
|
|
|
|
|
|
tt::Leaf::from(tt::Punct { char, spacing, id: self.id_alloc.alloc(range) }).into()
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
macro_rules! make_leaf {
|
|
|
|
($i:ident) => {
|
|
|
|
tt::$i { id: self.id_alloc.alloc(range), text: self.text[range].into() }.into()
|
|
|
|
};
|
|
|
|
}
|
|
|
|
let leaf: tt::Leaf = match k {
|
|
|
|
T![true] | T![false] => make_leaf!(Literal),
|
|
|
|
IDENT | LIFETIME => make_leaf!(Ident),
|
|
|
|
k if k.is_keyword() => make_leaf!(Ident),
|
|
|
|
k if k.is_literal() => make_leaf!(Literal),
|
|
|
|
_ => return,
|
|
|
|
};
|
|
|
|
|
|
|
|
leaf.into()
|
|
|
|
});
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// FIXME: There are some duplicate logic between RawConvertor and Convertor
|
|
|
|
// It would be nice to refactor to converting SyntaxNode to ra_parser::Token and thus
|
|
|
|
// use RawConvertor directly. But performance-wise it may not be a good idea ?
|
|
|
|
struct Convertor {
|
|
|
|
id_alloc: TokenIdAlloc,
|
|
|
|
}
|
|
|
|
|
2019-11-18 07:08:41 -06:00
|
|
|
impl Convertor {
|
|
|
|
fn go(&mut self, tt: &SyntaxNode) -> Option<tt::Subtree> {
|
|
|
|
// This tree is empty
|
|
|
|
if tt.first_child_or_token().is_none() {
|
2019-12-13 07:53:34 -06:00
|
|
|
return Some(tt::Subtree { token_trees: vec![], delimiter: None });
|
2019-11-18 07:08:41 -06:00
|
|
|
}
|
2019-04-12 12:50:05 -05:00
|
|
|
|
2019-11-18 07:08:41 -06:00
|
|
|
let first_child = tt.first_child_or_token()?;
|
|
|
|
let last_child = tt.last_child_or_token()?;
|
2019-11-21 09:56:01 -06:00
|
|
|
|
|
|
|
// ignore trivial first_child and last_child
|
|
|
|
let first_child = successors(Some(first_child), |it| {
|
|
|
|
if it.kind().is_trivia() {
|
|
|
|
it.next_sibling_or_token()
|
|
|
|
} else {
|
|
|
|
None
|
|
|
|
}
|
|
|
|
})
|
|
|
|
.last()
|
|
|
|
.unwrap();
|
|
|
|
if first_child.kind().is_trivia() {
|
2019-12-13 07:53:34 -06:00
|
|
|
return Some(tt::Subtree { token_trees: vec![], delimiter: None });
|
2019-11-21 09:56:01 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
let last_child = successors(Some(last_child), |it| {
|
|
|
|
if it.kind().is_trivia() {
|
|
|
|
it.prev_sibling_or_token()
|
|
|
|
} else {
|
|
|
|
None
|
|
|
|
}
|
|
|
|
})
|
|
|
|
.last()
|
|
|
|
.unwrap();
|
|
|
|
|
2019-12-12 11:41:44 -06:00
|
|
|
let (delimiter_kind, skip_first) = match (first_child.kind(), last_child.kind()) {
|
|
|
|
(T!['('], T![')']) => (Some(tt::DelimiterKind::Parenthesis), true),
|
|
|
|
(T!['{'], T!['}']) => (Some(tt::DelimiterKind::Brace), true),
|
|
|
|
(T!['['], T![']']) => (Some(tt::DelimiterKind::Bracket), true),
|
2019-12-13 07:53:34 -06:00
|
|
|
_ => (None, false),
|
2019-11-18 07:08:41 -06:00
|
|
|
};
|
2019-12-12 11:41:44 -06:00
|
|
|
let delimiter = delimiter_kind.map(|kind| tt::Delimiter {
|
|
|
|
kind,
|
2020-03-08 11:13:04 -05:00
|
|
|
id: self.id_alloc.delim(first_child.text_range(), last_child.text_range()),
|
2019-12-12 11:41:44 -06:00
|
|
|
});
|
2019-04-24 10:01:32 -05:00
|
|
|
|
2019-11-18 07:08:41 -06:00
|
|
|
let mut token_trees = Vec::new();
|
|
|
|
let mut child_iter = tt.children_with_tokens().skip(skip_first as usize).peekable();
|
2019-05-04 02:00:16 -05:00
|
|
|
|
2019-11-18 07:08:41 -06:00
|
|
|
while let Some(child) = child_iter.next() {
|
|
|
|
if skip_first && (child == first_child || child == last_child) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
match child {
|
|
|
|
NodeOrToken::Token(token) => {
|
|
|
|
if let Some(doc_tokens) = convert_doc_comment(&token) {
|
|
|
|
token_trees.extend(doc_tokens);
|
|
|
|
} else if token.kind().is_trivia() {
|
|
|
|
continue;
|
|
|
|
} else if token.kind().is_punct() {
|
2019-12-05 08:10:33 -06:00
|
|
|
// we need to pull apart joined punctuation tokens
|
|
|
|
let last_spacing = match child_iter.peek() {
|
2019-11-18 07:08:41 -06:00
|
|
|
Some(NodeOrToken::Token(token)) => {
|
|
|
|
if token.kind().is_punct() {
|
|
|
|
tt::Spacing::Joint
|
|
|
|
} else {
|
|
|
|
tt::Spacing::Alone
|
|
|
|
}
|
2019-04-24 10:01:32 -05:00
|
|
|
}
|
2019-11-18 07:08:41 -06:00
|
|
|
_ => tt::Spacing::Alone,
|
2019-05-15 07:35:47 -05:00
|
|
|
};
|
2019-12-05 08:10:33 -06:00
|
|
|
let spacing_iter = std::iter::repeat(tt::Spacing::Joint)
|
|
|
|
.take(token.text().len() - 1)
|
|
|
|
.chain(std::iter::once(last_spacing));
|
|
|
|
for (char, spacing) in token.text().chars().zip(spacing_iter) {
|
2019-12-17 21:47:26 -06:00
|
|
|
token_trees.push(
|
|
|
|
tt::Leaf::from(tt::Punct {
|
|
|
|
char,
|
|
|
|
spacing,
|
2020-03-08 11:13:04 -05:00
|
|
|
id: self.id_alloc.alloc(token.text_range()),
|
2019-12-17 21:47:26 -06:00
|
|
|
})
|
|
|
|
.into(),
|
|
|
|
);
|
2019-12-05 08:10:33 -06:00
|
|
|
}
|
2019-11-18 07:08:41 -06:00
|
|
|
} else {
|
2019-12-17 21:47:26 -06:00
|
|
|
macro_rules! make_leaf {
|
|
|
|
($i:ident) => {
|
|
|
|
tt::$i {
|
2020-03-08 11:13:04 -05:00
|
|
|
id: self.id_alloc.alloc(token.text_range()),
|
2019-12-17 21:47:26 -06:00
|
|
|
text: token.text().clone(),
|
|
|
|
}
|
|
|
|
.into()
|
|
|
|
};
|
|
|
|
}
|
|
|
|
|
2019-12-12 11:41:44 -06:00
|
|
|
let child: tt::Leaf = match token.kind() {
|
2019-12-17 21:47:26 -06:00
|
|
|
T![true] | T![false] => make_leaf!(Literal),
|
|
|
|
IDENT | LIFETIME => make_leaf!(Ident),
|
|
|
|
k if k.is_keyword() => make_leaf!(Ident),
|
|
|
|
k if k.is_literal() => make_leaf!(Literal),
|
2019-12-12 11:41:44 -06:00
|
|
|
_ => return None,
|
2019-12-12 07:47:54 -06:00
|
|
|
};
|
2019-12-12 11:41:44 -06:00
|
|
|
token_trees.push(child.into());
|
2019-11-18 07:08:41 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
NodeOrToken::Node(node) => {
|
2019-12-05 12:27:39 -06:00
|
|
|
let child_subtree = self.go(&node)?;
|
2019-12-13 07:53:34 -06:00
|
|
|
if child_subtree.delimiter.is_none() && node.kind() != SyntaxKind::TOKEN_TREE {
|
2019-12-05 12:27:39 -06:00
|
|
|
token_trees.extend(child_subtree.token_trees);
|
|
|
|
} else {
|
|
|
|
token_trees.push(child_subtree.into());
|
|
|
|
}
|
2019-01-31 12:29:04 -06:00
|
|
|
}
|
2019-11-18 07:08:41 -06:00
|
|
|
};
|
|
|
|
}
|
|
|
|
|
|
|
|
let res = tt::Subtree { delimiter, token_trees };
|
|
|
|
Some(res)
|
2019-01-31 12:29:04 -06:00
|
|
|
}
|
|
|
|
}
|
2019-02-23 05:18:32 -06:00
|
|
|
|
2019-05-27 09:56:21 -05:00
|
|
|
struct TtTreeSink<'a> {
|
2019-02-23 08:21:56 -06:00
|
|
|
buf: String,
|
2019-05-27 09:56:21 -05:00
|
|
|
cursor: Cursor<'a>,
|
2019-12-12 11:41:44 -06:00
|
|
|
open_delims: FxHashMap<tt::TokenId, TextUnit>,
|
2019-02-23 08:21:56 -06:00
|
|
|
text_pos: TextUnit,
|
|
|
|
inner: SyntaxTreeBuilder,
|
2019-11-18 07:08:41 -06:00
|
|
|
token_map: TokenMap,
|
2019-05-13 18:59:28 -05:00
|
|
|
|
|
|
|
// Number of roots
|
|
|
|
// Use for detect ill-form tree which is not single root
|
|
|
|
roots: smallvec::SmallVec<[usize; 1]>,
|
2019-02-23 08:21:56 -06:00
|
|
|
}
|
|
|
|
|
2019-05-27 09:56:21 -05:00
|
|
|
impl<'a> TtTreeSink<'a> {
|
2019-11-03 13:12:19 -06:00
|
|
|
fn new(cursor: Cursor<'a>) -> Self {
|
2019-02-23 08:21:56 -06:00
|
|
|
TtTreeSink {
|
|
|
|
buf: String::new(),
|
2019-05-27 09:56:21 -05:00
|
|
|
cursor,
|
2019-12-12 11:41:44 -06:00
|
|
|
open_delims: FxHashMap::default(),
|
2019-02-23 08:21:56 -06:00
|
|
|
text_pos: 0.into(),
|
|
|
|
inner: SyntaxTreeBuilder::default(),
|
2019-05-13 18:59:28 -05:00
|
|
|
roots: smallvec::SmallVec::new(),
|
2019-11-18 07:08:41 -06:00
|
|
|
token_map: TokenMap::default(),
|
2019-02-23 08:21:56 -06:00
|
|
|
}
|
|
|
|
}
|
2019-11-03 13:12:19 -06:00
|
|
|
|
2019-11-18 07:08:41 -06:00
|
|
|
fn finish(self) -> (Parse<SyntaxNode>, TokenMap) {
|
|
|
|
(self.inner.finish(), self.token_map)
|
2019-11-03 13:12:19 -06:00
|
|
|
}
|
2019-02-23 08:21:56 -06:00
|
|
|
}
|
|
|
|
|
2019-12-17 21:47:26 -06:00
|
|
|
fn delim_to_str(d: Option<tt::DelimiterKind>, closing: bool) -> SmolStr {
|
|
|
|
let texts = match d {
|
2019-12-12 11:41:44 -06:00
|
|
|
Some(tt::DelimiterKind::Parenthesis) => "()",
|
|
|
|
Some(tt::DelimiterKind::Brace) => "{}",
|
|
|
|
Some(tt::DelimiterKind::Bracket) => "[]",
|
2019-12-13 07:53:34 -06:00
|
|
|
None => return "".into(),
|
2019-05-27 09:56:21 -05:00
|
|
|
};
|
|
|
|
|
|
|
|
let idx = closing as usize;
|
2019-12-13 07:53:34 -06:00
|
|
|
let text = &texts[idx..texts.len() - (1 - idx)];
|
2019-05-27 09:56:21 -05:00
|
|
|
text.into()
|
2019-05-01 21:02:17 -05:00
|
|
|
}
|
|
|
|
|
2019-05-27 09:56:21 -05:00
|
|
|
impl<'a> TreeSink for TtTreeSink<'a> {
|
2019-03-30 05:25:53 -05:00
|
|
|
fn token(&mut self, kind: SyntaxKind, n_tokens: u8) {
|
2019-04-12 12:50:05 -05:00
|
|
|
if kind == L_DOLLAR || kind == R_DOLLAR {
|
2019-05-27 11:28:46 -05:00
|
|
|
self.cursor = self.cursor.bump_subtree();
|
2019-04-12 12:50:05 -05:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2020-03-04 09:38:58 -06:00
|
|
|
let mut last = self.cursor;
|
2019-02-23 08:21:56 -06:00
|
|
|
for _ in 0..n_tokens {
|
2019-05-27 09:56:21 -05:00
|
|
|
if self.cursor.eof() {
|
|
|
|
break;
|
|
|
|
}
|
2020-03-04 09:38:58 -06:00
|
|
|
last = self.cursor;
|
2019-12-17 21:47:26 -06:00
|
|
|
let text: SmolStr = match self.cursor.token_tree() {
|
2019-05-27 09:56:21 -05:00
|
|
|
Some(tt::TokenTree::Leaf(leaf)) => {
|
2019-11-03 11:45:58 -06:00
|
|
|
// Mark the range if needed
|
2019-12-12 11:41:44 -06:00
|
|
|
let id = match leaf {
|
|
|
|
tt::Leaf::Ident(ident) => ident.id,
|
|
|
|
tt::Leaf::Punct(punct) => punct.id,
|
|
|
|
tt::Leaf::Literal(lit) => lit.id,
|
|
|
|
};
|
|
|
|
let text = SmolStr::new(format!("{}", leaf));
|
|
|
|
let range = TextRange::offset_len(self.text_pos, TextUnit::of_str(&text));
|
|
|
|
self.token_map.insert(id, range);
|
2019-11-03 11:45:58 -06:00
|
|
|
self.cursor = self.cursor.bump();
|
2019-12-17 21:47:26 -06:00
|
|
|
text
|
2019-05-27 09:56:21 -05:00
|
|
|
}
|
|
|
|
Some(tt::TokenTree::Subtree(subtree)) => {
|
|
|
|
self.cursor = self.cursor.subtree().unwrap();
|
2019-12-12 11:41:44 -06:00
|
|
|
if let Some(id) = subtree.delimiter.map(|it| it.id) {
|
|
|
|
self.open_delims.insert(id, self.text_pos);
|
2019-05-27 09:56:21 -05:00
|
|
|
}
|
2019-12-17 21:47:26 -06:00
|
|
|
delim_to_str(subtree.delimiter_kind(), false)
|
2019-05-27 09:56:21 -05:00
|
|
|
}
|
2019-12-17 21:47:26 -06:00
|
|
|
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) {
|
|
|
|
let open_range =
|
|
|
|
TextRange::offset_len(*open_delim, TextUnit::from_usize(1));
|
|
|
|
let close_range =
|
|
|
|
TextRange::offset_len(self.text_pos, TextUnit::from_usize(1));
|
|
|
|
self.token_map.insert_delim(id, open_range, close_range);
|
|
|
|
}
|
2019-12-12 11:41:44 -06:00
|
|
|
}
|
2019-12-17 21:47:26 -06:00
|
|
|
delim_to_str(parent.delimiter_kind(), true)
|
|
|
|
} else {
|
|
|
|
continue;
|
2019-12-12 11:41:44 -06:00
|
|
|
}
|
2019-12-17 21:47:26 -06:00
|
|
|
}
|
2019-05-27 09:56:21 -05:00
|
|
|
};
|
2019-12-17 21:47:26 -06:00
|
|
|
self.buf += &text;
|
|
|
|
self.text_pos += TextUnit::of_str(&text);
|
2019-02-23 08:21:56 -06:00
|
|
|
}
|
2019-05-27 09:56:21 -05:00
|
|
|
|
2019-02-23 08:21:56 -06:00
|
|
|
let text = SmolStr::new(self.buf.as_str());
|
|
|
|
self.buf.clear();
|
2019-04-25 10:12:57 -05:00
|
|
|
self.inner.token(kind, text);
|
|
|
|
|
2019-05-27 09:56:21 -05:00
|
|
|
// Add whitespace between adjoint puncts
|
2020-03-04 09:38:58 -06:00
|
|
|
let next = last.bump();
|
2019-05-27 09:56:21 -05:00
|
|
|
if let (
|
|
|
|
Some(tt::TokenTree::Leaf(tt::Leaf::Punct(curr))),
|
|
|
|
Some(tt::TokenTree::Leaf(tt::Leaf::Punct(_))),
|
2020-03-04 09:38:58 -06:00
|
|
|
) = (last.token_tree(), next.token_tree())
|
2019-05-27 09:56:21 -05:00
|
|
|
{
|
2020-03-05 14:32:08 -06:00
|
|
|
// 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 != ';' {
|
2019-05-27 09:56:21 -05:00
|
|
|
self.inner.token(WHITESPACE, " ".into());
|
2019-11-03 08:46:12 -06:00
|
|
|
self.text_pos += TextUnit::of_char(' ');
|
2019-05-01 21:02:17 -05:00
|
|
|
}
|
|
|
|
}
|
2019-02-23 08:21:56 -06:00
|
|
|
}
|
|
|
|
|
2019-03-30 05:25:53 -05:00
|
|
|
fn start_node(&mut self, kind: SyntaxKind) {
|
|
|
|
self.inner.start_node(kind);
|
2019-05-13 18:59:28 -05:00
|
|
|
|
|
|
|
match self.roots.last_mut() {
|
|
|
|
None | Some(0) => self.roots.push(1),
|
|
|
|
Some(ref mut n) => **n += 1,
|
|
|
|
};
|
2019-02-23 08:21:56 -06:00
|
|
|
}
|
|
|
|
|
2019-03-30 05:25:53 -05:00
|
|
|
fn finish_node(&mut self) {
|
|
|
|
self.inner.finish_node();
|
2019-05-13 18:59:28 -05:00
|
|
|
*self.roots.last_mut().unwrap() -= 1;
|
2019-02-23 08:21:56 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
fn error(&mut self, error: ParseError) {
|
|
|
|
self.inner.error(error, self.text_pos)
|
|
|
|
}
|
|
|
|
}
|
2019-04-05 07:58:24 -05:00
|
|
|
|
|
|
|
#[cfg(test)]
|
|
|
|
mod tests {
|
|
|
|
use super::*;
|
2019-12-19 09:17:22 -06:00
|
|
|
use crate::tests::parse_macro;
|
2019-05-27 09:56:21 -05:00
|
|
|
use ra_parser::TokenSource;
|
2019-12-05 08:10:33 -06:00
|
|
|
use ra_syntax::{
|
|
|
|
algo::{insert_children, InsertPosition},
|
|
|
|
ast::AstNode,
|
|
|
|
};
|
2019-04-05 07:58:24 -05:00
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn convert_tt_token_source() {
|
2019-12-19 09:17:22 -06:00
|
|
|
let expansion = parse_macro(
|
2019-04-05 07:58:24 -05:00
|
|
|
r#"
|
|
|
|
macro_rules! literals {
|
|
|
|
($i:ident) => {
|
|
|
|
{
|
|
|
|
let a = 'c';
|
|
|
|
let c = 1000;
|
|
|
|
let f = 12E+99_f64;
|
|
|
|
let s = "rust1";
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
"#,
|
2019-12-19 09:17:22 -06:00
|
|
|
)
|
|
|
|
.expand_tt("literals!(foo);");
|
2019-10-30 12:36:37 -05:00
|
|
|
let tts = &[expansion.into()];
|
2019-06-02 11:54:33 -05:00
|
|
|
let buffer = tt::buffer::TokenBuffer::new(tts);
|
2019-05-27 09:56:21 -05:00
|
|
|
let mut tt_src = SubtreeTokenSource::new(&buffer);
|
|
|
|
let mut tokens = vec![];
|
|
|
|
while tt_src.current().kind != EOF {
|
|
|
|
tokens.push((tt_src.current().kind, tt_src.text()));
|
|
|
|
tt_src.bump();
|
|
|
|
}
|
2019-04-05 07:58:24 -05:00
|
|
|
|
2019-04-12 12:50:05 -05:00
|
|
|
// [${]
|
2019-04-05 07:58:24 -05:00
|
|
|
// [let] [a] [=] ['c'] [;]
|
2019-05-27 09:56:21 -05:00
|
|
|
assert_eq!(tokens[2 + 3].1, "'c'");
|
|
|
|
assert_eq!(tokens[2 + 3].0, CHAR);
|
2019-04-05 07:58:24 -05:00
|
|
|
// [let] [c] [=] [1000] [;]
|
2019-05-27 09:56:21 -05:00
|
|
|
assert_eq!(tokens[2 + 5 + 3].1, "1000");
|
|
|
|
assert_eq!(tokens[2 + 5 + 3].0, INT_NUMBER);
|
2019-04-05 07:58:24 -05:00
|
|
|
// [let] [f] [=] [12E+99_f64] [;]
|
2019-05-27 09:56:21 -05:00
|
|
|
assert_eq!(tokens[2 + 10 + 3].1, "12E+99_f64");
|
|
|
|
assert_eq!(tokens[2 + 10 + 3].0, FLOAT_NUMBER);
|
2019-04-05 07:58:24 -05:00
|
|
|
|
|
|
|
// [let] [s] [=] ["rust1"] [;]
|
2019-05-27 09:56:21 -05:00
|
|
|
assert_eq!(tokens[2 + 15 + 3].1, "\"rust1\"");
|
|
|
|
assert_eq!(tokens[2 + 15 + 3].0, STRING);
|
2019-04-05 07:58:24 -05:00
|
|
|
}
|
2019-05-13 18:59:28 -05:00
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn stmts_token_trees_to_expr_is_err() {
|
2019-12-19 09:17:22 -06:00
|
|
|
let expansion = parse_macro(
|
2019-05-13 18:59:28 -05:00
|
|
|
r#"
|
|
|
|
macro_rules! stmts {
|
|
|
|
() => {
|
|
|
|
let a = 0;
|
|
|
|
let b = 0;
|
|
|
|
let c = 0;
|
|
|
|
let d = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
"#,
|
2019-12-19 09:17:22 -06:00
|
|
|
)
|
|
|
|
.expand_tt("stmts!();");
|
2019-11-09 04:14:10 -06:00
|
|
|
assert!(token_tree_to_syntax_node(&expansion, FragmentKind::Expr).is_err());
|
2019-05-13 18:59:28 -05:00
|
|
|
}
|
2019-11-21 09:56:01 -06:00
|
|
|
|
|
|
|
#[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 06:49:43 -06:00
|
|
|
std::iter::once(space),
|
2019-11-21 09:56:01 -06:00
|
|
|
);
|
|
|
|
|
|
|
|
// 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).unwrap().0;
|
|
|
|
|
2019-12-17 21:47:26 -06:00
|
|
|
assert_eq!(tt.delimiter_kind(), Some(tt::DelimiterKind::Brace));
|
2019-11-21 09:56:01 -06:00
|
|
|
}
|
2019-12-05 08:10:33 -06:00
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn test_token_tree_multi_char_punct() {
|
|
|
|
let source_file = ast::SourceFile::parse("struct Foo { a: x::Y }").ok().unwrap();
|
|
|
|
let struct_def = source_file.syntax().descendants().find_map(ast::StructDef::cast).unwrap();
|
|
|
|
let tt = ast_to_token_tree(&struct_def).unwrap().0;
|
|
|
|
token_tree_to_syntax_node(&tt, FragmentKind::Item).unwrap();
|
|
|
|
}
|
2019-04-05 07:58:24 -05:00
|
|
|
}
|