aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorEdwin Cheng <[email protected]>2019-04-07 14:42:53 +0100
committerEdwin Cheng <[email protected]>2019-04-07 14:42:53 +0100
commit7f1e93a3c934342a9b9c418865ff395b8ad245ce (patch)
treef31ec549b70a9260939c0fc8c0d4bb6433133a0d
parentaac9dfa46418603940ab2333cfea2190d9464d9e (diff)
Refactoring subtree_source
-rw-r--r--crates/ra_mbe/src/lib.rs4
-rw-r--r--crates/ra_mbe/src/subtree_source.rs352
-rw-r--r--crates/ra_mbe/src/syntax_bridge.rs262
-rw-r--r--crates/ra_mbe/src/tt_cursor.rs132
4 files changed, 386 insertions, 364 deletions
diff --git a/crates/ra_mbe/src/lib.rs b/crates/ra_mbe/src/lib.rs
index a5b7fab52..38d3ec7e1 100644
--- a/crates/ra_mbe/src/lib.rs
+++ b/crates/ra_mbe/src/lib.rs
@@ -15,10 +15,12 @@ macro_rules! impl_froms {
15 } 15 }
16} 16}
17 17
18mod tt_cursor; 18// mod tt_cursor;
19mod mbe_parser; 19mod mbe_parser;
20mod mbe_expander; 20mod mbe_expander;
21mod syntax_bridge; 21mod syntax_bridge;
22mod tt_cursor;
23mod subtree_source;
22 24
23use ra_syntax::SmolStr; 25use ra_syntax::SmolStr;
24 26
diff --git a/crates/ra_mbe/src/subtree_source.rs b/crates/ra_mbe/src/subtree_source.rs
new file mode 100644
index 000000000..8f5ce4ed5
--- /dev/null
+++ b/crates/ra_mbe/src/subtree_source.rs
@@ -0,0 +1,352 @@
1use ra_parser::{TokenSource};
2use ra_syntax::{classify_literal, SmolStr, SyntaxKind, SyntaxKind::*};
3
4#[derive(Debug)]
5struct TtToken {
6 pub kind: SyntaxKind,
7 pub is_joint_to_next: bool,
8 pub text: SmolStr,
9 pub n_tokens: usize,
10}
11
12/// SubtreeSourceQuerier let outside to query internal tokens as string
13pub(crate) struct SubtreeSourceQuerier<'a> {
14 src: &'a SubtreeTokenSource<'a>,
15}
16
17impl<'a> SubtreeSourceQuerier<'a> {
18 pub(crate) fn token(&self, uidx: usize) -> (SyntaxKind, &SmolStr) {
19 let tkn = &self.src.tokens[uidx];
20 (tkn.kind, &tkn.text)
21 }
22}
23
24pub(crate) struct SubtreeTokenSource<'a> {
25 tt_pos: usize,
26 tokens: Vec<TtToken>,
27 subtree: &'a tt::Subtree,
28}
29
30impl<'a> SubtreeTokenSource<'a> {
31 pub fn new(subtree: &tt::Subtree) -> SubtreeTokenSource {
32 SubtreeTokenSource { tokens: TtTokenBuilder::build(subtree), tt_pos: 0, subtree }
33 }
34
35 pub fn advance(&mut self, curr: usize, skip_first_delimiter: bool) {
36 if skip_first_delimiter {
37 self.tt_pos += 1;
38 }
39
40 // Matching `TtToken` cursor to `tt::TokenTree` cursor
41 // It is because TtToken is not One to One mapping to tt::Token
42 // There are 3 case (`TtToken` <=> `tt::TokenTree`) :
43 // * One to One => ident, single char punch
44 // * Many to One => `tt::TokenTree::SubTree`
45 // * One to Many => multibyte punct
46 //
47 // Such that we cannot simpliy advance the cursor
48 // We have to bump it one by one
49 let mut pos = 0;
50 while pos < curr {
51 pos += self.bump(&self.subtree.token_trees[pos]);
52 }
53 }
54
55 pub fn querier(&self) -> SubtreeSourceQuerier {
56 SubtreeSourceQuerier { src: self }
57 }
58
59 fn count(&self, tt: &tt::TokenTree) -> usize {
60 assert!(!self.tokens.is_empty());
61 TtTokenBuilder::count_tt_tokens(tt, None)
62 }
63
64 pub(crate) fn bump(&mut self, tt: &tt::TokenTree) -> usize {
65 let cur = &self.tokens[self.tt_pos];
66 let n_tokens = cur.n_tokens;
67 self.tt_pos += self.count(tt);
68 n_tokens
69 }
70
71 pub(crate) fn bump_n(
72 &mut self,
73 n_tokens: usize,
74 mut token_pos: usize,
75 ) -> (usize, Vec<&tt::TokenTree>) {
76 let mut res = vec![];
77 // Matching `TtToken` cursor to `tt::TokenTree` cursor
78 // It is because TtToken is not One to One mapping to tt::Token
79 // There are 3 case (`TtToken` <=> `tt::TokenTree`) :
80 // * One to One => ident, single char punch
81 // * Many to One => `tt::TokenTree::SubTree`
82 // * One to Many => multibyte punct
83 //
84 // Such that we cannot simpliy advance the cursor
85 // We have to bump it one by one
86 let next_pos = self.tt_pos + n_tokens;
87 let old_token_pos = token_pos;
88
89 while self.tt_pos < next_pos {
90 let current = &self.subtree.token_trees[token_pos];
91 let n = self.bump(current);
92 res.extend((0..n).map(|i| &self.subtree.token_trees[token_pos + i]));
93 token_pos += n;
94 }
95
96 (token_pos - old_token_pos, res)
97 }
98}
99
100impl<'a> TokenSource for SubtreeTokenSource<'a> {
101 fn token_kind(&self, pos: usize) -> SyntaxKind {
102 if let Some(tok) = self.tokens.get(self.tt_pos + pos) {
103 tok.kind
104 } else {
105 SyntaxKind::EOF
106 }
107 }
108 fn is_token_joint_to_next(&self, pos: usize) -> bool {
109 self.tokens[self.tt_pos + pos].is_joint_to_next
110 }
111 fn is_keyword(&self, pos: usize, kw: &str) -> bool {
112 self.tokens[self.tt_pos + pos].text == *kw
113 }
114}
115
116struct TokenPeek<'a, I>
117where
118 I: Iterator<Item = &'a tt::TokenTree>,
119{
120 iter: itertools::MultiPeek<I>,
121}
122
123// helper function
124fn to_punct(tt: &tt::TokenTree) -> Option<&tt::Punct> {
125 if let tt::TokenTree::Leaf(tt::Leaf::Punct(pp)) = tt {
126 return Some(pp);
127 }
128 None
129}
130
131impl<'a, I> TokenPeek<'a, I>
132where
133 I: Iterator<Item = &'a tt::TokenTree>,
134{
135 pub fn new(iter: I) -> Self {
136 TokenPeek { iter: itertools::multipeek(iter) }
137 }
138
139 pub fn next(&mut self) -> Option<&tt::TokenTree> {
140 self.iter.next()
141 }
142
143 fn current_punct2(&mut self, p: &tt::Punct) -> Option<((char, char), bool)> {
144 if p.spacing != tt::Spacing::Joint {
145 return None;
146 }
147
148 self.iter.reset_peek();
149 let p1 = to_punct(self.iter.peek()?)?;
150 Some(((p.char, p1.char), p1.spacing == tt::Spacing::Joint))
151 }
152
153 fn current_punct3(&mut self, p: &tt::Punct) -> Option<((char, char, char), bool)> {
154 self.current_punct2(p).and_then(|((p0, p1), last_joint)| {
155 if !last_joint {
156 None
157 } else {
158 let p2 = to_punct(*self.iter.peek()?)?;
159 Some(((p0, p1, p2.char), p2.spacing == tt::Spacing::Joint))
160 }
161 })
162 }
163}
164
165struct TtTokenBuilder {
166 tokens: Vec<TtToken>,
167}
168
169impl TtTokenBuilder {
170 fn build(sub: &tt::Subtree) -> Vec<TtToken> {
171 let mut res = TtTokenBuilder { tokens: vec![] };
172 res.convert_subtree(sub);
173 res.tokens
174 }
175
176 fn convert_subtree(&mut self, sub: &tt::Subtree) {
177 self.push_delim(sub.delimiter, false);
178 let mut peek = TokenPeek::new(sub.token_trees.iter());
179 while let Some(tt) = peek.iter.next() {
180 self.convert_tt(tt, &mut peek);
181 }
182 self.push_delim(sub.delimiter, true)
183 }
184
185 fn convert_tt<'b, I>(&mut self, tt: &tt::TokenTree, iter: &mut TokenPeek<'b, I>)
186 where
187 I: Iterator<Item = &'b tt::TokenTree>,
188 {
189 match tt {
190 tt::TokenTree::Leaf(token) => self.convert_token(token, iter),
191 tt::TokenTree::Subtree(sub) => self.convert_subtree(sub),
192 }
193 }
194
195 fn convert_token<'b, I>(&mut self, token: &tt::Leaf, iter: &mut TokenPeek<'b, I>)
196 where
197 I: Iterator<Item = &'b tt::TokenTree>,
198 {
199 let tok = match token {
200 tt::Leaf::Literal(l) => TtToken {
201 kind: classify_literal(&l.text).unwrap().kind,
202 is_joint_to_next: false,
203 text: l.text.clone(),
204 n_tokens: 1,
205 },
206 tt::Leaf::Punct(p) => {
207 if let Some((kind, is_joint_to_next, text, size)) =
208 Self::convert_multi_char_punct(p, iter)
209 {
210 for _ in 0..size - 1 {
211 iter.next();
212 }
213
214 TtToken { kind, is_joint_to_next, text: text.into(), n_tokens: size }
215 } else {
216 let kind = match p.char {
217 // lexer may produce combpund tokens for these ones
218 '.' => DOT,
219 ':' => COLON,
220 '=' => EQ,
221 '!' => EXCL,
222 '-' => MINUS,
223 c => SyntaxKind::from_char(c).unwrap(),
224 };
225 let text = {
226 let mut buf = [0u8; 4];
227 let s: &str = p.char.encode_utf8(&mut buf);
228 SmolStr::new(s)
229 };
230 TtToken {
231 kind,
232 is_joint_to_next: p.spacing == tt::Spacing::Joint,
233 text,
234 n_tokens: 1,
235 }
236 }
237 }
238 tt::Leaf::Ident(ident) => {
239 let kind = SyntaxKind::from_keyword(ident.text.as_str()).unwrap_or(IDENT);
240 TtToken { kind, is_joint_to_next: false, text: ident.text.clone(), n_tokens: 1 }
241 }
242 };
243 self.tokens.push(tok)
244 }
245
246 fn convert_multi_char_punct<'b, I>(
247 p: &tt::Punct,
248 iter: &mut TokenPeek<'b, I>,
249 ) -> Option<(SyntaxKind, bool, &'static str, usize)>
250 where
251 I: Iterator<Item = &'b tt::TokenTree>,
252 {
253 if let Some((m, is_joint_to_next)) = iter.current_punct3(p) {
254 if let Some((kind, text)) = match m {
255 ('<', '<', '=') => Some((SHLEQ, "<<=")),
256 ('>', '>', '=') => Some((SHREQ, ">>=")),
257 ('.', '.', '.') => Some((DOTDOTDOT, "...")),
258 ('.', '.', '=') => Some((DOTDOTEQ, "..=")),
259 _ => None,
260 } {
261 return Some((kind, is_joint_to_next, text, 3));
262 }
263 }
264
265 if let Some((m, is_joint_to_next)) = iter.current_punct2(p) {
266 if let Some((kind, text)) = match m {
267 ('<', '<') => Some((SHL, "<<")),
268 ('>', '>') => Some((SHR, ">>")),
269
270 ('|', '|') => Some((PIPEPIPE, "||")),
271 ('&', '&') => Some((AMPAMP, "&&")),
272 ('%', '=') => Some((PERCENTEQ, "%=")),
273 ('*', '=') => Some((STAREQ, "*=")),
274 ('/', '=') => Some((SLASHEQ, "/=")),
275 ('^', '=') => Some((CARETEQ, "^=")),
276
277 ('&', '=') => Some((AMPEQ, "&=")),
278 ('|', '=') => Some((PIPEEQ, "|=")),
279 ('-', '=') => Some((MINUSEQ, "-=")),
280 ('+', '=') => Some((PLUSEQ, "+=")),
281 ('>', '=') => Some((GTEQ, ">=")),
282 ('<', '=') => Some((LTEQ, "<=")),
283
284 ('-', '>') => Some((THIN_ARROW, "->")),
285 ('!', '=') => Some((NEQ, "!=")),
286 ('=', '>') => Some((FAT_ARROW, "=>")),
287 ('=', '=') => Some((EQEQ, "==")),
288 ('.', '.') => Some((DOTDOT, "..")),
289 (':', ':') => Some((COLONCOLON, "::")),
290
291 _ => None,
292 } {
293 return Some((kind, is_joint_to_next, text, 2));
294 }
295 }
296
297 None
298 }
299
300 fn push_delim(&mut self, d: tt::Delimiter, closing: bool) {
301 let (kinds, texts) = match d {
302 tt::Delimiter::Parenthesis => ([L_PAREN, R_PAREN], "()"),
303 tt::Delimiter::Brace => ([L_CURLY, R_CURLY], "{}"),
304 tt::Delimiter::Bracket => ([L_BRACK, R_BRACK], "[]"),
305 tt::Delimiter::None => return,
306 };
307 let idx = closing as usize;
308 let kind = kinds[idx];
309 let text = &texts[idx..texts.len() - (1 - idx)];
310 let tok = TtToken { kind, is_joint_to_next: false, text: SmolStr::new(text), n_tokens: 1 };
311 self.tokens.push(tok)
312 }
313
314 fn skip_sibling_leaf(leaf: &tt::Leaf, iter: &mut std::slice::Iter<tt::TokenTree>) {
315 if let tt::Leaf::Punct(p) = leaf {
316 let mut peek = TokenPeek::new(iter);
317 if let Some((_, _, _, size)) = TtTokenBuilder::convert_multi_char_punct(p, &mut peek) {
318 for _ in 0..size - 1 {
319 peek.next();
320 }
321 }
322 }
323 }
324
325 fn count_tt_tokens(
326 tt: &tt::TokenTree,
327 iter: Option<&mut std::slice::Iter<tt::TokenTree>>,
328 ) -> usize {
329 match tt {
330 tt::TokenTree::Subtree(sub_tree) => {
331 let mut iter = sub_tree.token_trees.iter();
332 let mut count = match sub_tree.delimiter {
333 tt::Delimiter::None => 0,
334 _ => 2,
335 };
336
337 while let Some(tt) = iter.next() {
338 count += Self::count_tt_tokens(&tt, Some(&mut iter));
339 }
340 count
341 }
342
343 tt::TokenTree::Leaf(leaf) => {
344 iter.map(|iter| {
345 Self::skip_sibling_leaf(leaf, iter);
346 });
347
348 1
349 }
350 }
351 }
352}
diff --git a/crates/ra_mbe/src/syntax_bridge.rs b/crates/ra_mbe/src/syntax_bridge.rs
index 3a0702a30..102bba341 100644
--- a/crates/ra_mbe/src/syntax_bridge.rs
+++ b/crates/ra_mbe/src/syntax_bridge.rs
@@ -1,9 +1,11 @@
1use ra_parser::{TokenSource, TreeSink, ParseError}; 1use ra_parser::{TreeSink, ParseError};
2use ra_syntax::{ 2use ra_syntax::{
3 AstNode, SyntaxNode, TextRange, SyntaxKind, SmolStr, SyntaxTreeBuilder, TreeArc, SyntaxElement, 3 AstNode, SyntaxNode, TextRange, SyntaxKind, SmolStr, SyntaxTreeBuilder, TreeArc, SyntaxElement,
4 ast, SyntaxKind::*, TextUnit, classify_literal 4 ast, SyntaxKind::*, TextUnit
5}; 5};
6 6
7use crate::subtree_source::{SubtreeTokenSource, SubtreeSourceQuerier};
8
7/// Maps `tt::TokenId` to the relative range of the original token. 9/// Maps `tt::TokenId` to the relative range of the original token.
8#[derive(Default)] 10#[derive(Default)]
9pub struct TokenMap { 11pub struct TokenMap {
@@ -22,8 +24,8 @@ pub fn ast_to_token_tree(ast: &ast::TokenTree) -> Option<(tt::Subtree, TokenMap)
22 24
23/// Parses the token tree (result of macro expansion) as a sequence of items 25/// Parses the token tree (result of macro expansion) as a sequence of items
24pub fn token_tree_to_ast_item_list(tt: &tt::Subtree) -> TreeArc<ast::SourceFile> { 26pub fn token_tree_to_ast_item_list(tt: &tt::Subtree) -> TreeArc<ast::SourceFile> {
25 let token_source = TtTokenSource::new(tt); 27 let token_source = SubtreeTokenSource::new(tt);
26 let mut tree_sink = TtTreeSink::new(&token_source.tokens); 28 let mut tree_sink = TtTreeSink::new(token_source.querier());
27 ra_parser::parse(&token_source, &mut tree_sink); 29 ra_parser::parse(&token_source, &mut tree_sink);
28 let syntax = tree_sink.inner.finish(); 30 let syntax = tree_sink.inner.finish();
29 ast::SourceFile::cast(&syntax).unwrap().to_owned() 31 ast::SourceFile::cast(&syntax).unwrap().to_owned()
@@ -103,243 +105,19 @@ fn convert_tt(
103 Some(res) 105 Some(res)
104} 106}
105 107
106#[derive(Debug)]
107pub(crate) struct TtTokenSource {
108 pub tokens: Vec<TtToken>,
109}
110
111#[derive(Debug)]
112pub(crate) struct TtToken {
113 pub kind: SyntaxKind,
114 pub is_joint_to_next: bool,
115 pub text: SmolStr,
116 pub n_tokens: usize,
117}
118
119// Some helper functions
120fn to_punct(tt: &tt::TokenTree) -> Option<&tt::Punct> {
121 if let tt::TokenTree::Leaf(tt::Leaf::Punct(pp)) = tt {
122 return Some(pp);
123 }
124 None
125}
126
127pub(crate) struct TokenPeek<'a, I>
128where
129 I: Iterator<Item = &'a tt::TokenTree>,
130{
131 iter: itertools::MultiPeek<I>,
132}
133
134impl<'a, I> TokenPeek<'a, I>
135where
136 I: Iterator<Item = &'a tt::TokenTree>,
137{
138 pub fn new(iter: I) -> Self {
139 TokenPeek { iter: itertools::multipeek(iter) }
140 }
141
142 pub fn next(&mut self) -> Option<&tt::TokenTree> {
143 self.iter.next()
144 }
145
146 fn current_punct2(&mut self, p: &tt::Punct) -> Option<((char, char), bool)> {
147 if p.spacing != tt::Spacing::Joint {
148 return None;
149 }
150
151 self.iter.reset_peek();
152 let p1 = to_punct(self.iter.peek()?)?;
153 Some(((p.char, p1.char), p1.spacing == tt::Spacing::Joint))
154 }
155
156 fn current_punct3(&mut self, p: &tt::Punct) -> Option<((char, char, char), bool)> {
157 self.current_punct2(p).and_then(|((p0, p1), last_joint)| {
158 if !last_joint {
159 None
160 } else {
161 let p2 = to_punct(*self.iter.peek()?)?;
162 Some(((p0, p1, p2.char), p2.spacing == tt::Spacing::Joint))
163 }
164 })
165 }
166}
167
168impl TtTokenSource {
169 pub fn new(tt: &tt::Subtree) -> TtTokenSource {
170 let mut res = TtTokenSource { tokens: Vec::new() };
171 res.convert_subtree(tt);
172 res
173 }
174 fn convert_subtree(&mut self, sub: &tt::Subtree) {
175 self.push_delim(sub.delimiter, false);
176 let mut peek = TokenPeek::new(sub.token_trees.iter());
177 while let Some(tt) = peek.iter.next() {
178 self.convert_tt(tt, &mut peek);
179 }
180 self.push_delim(sub.delimiter, true)
181 }
182
183 fn convert_tt<'a, I>(&mut self, tt: &tt::TokenTree, iter: &mut TokenPeek<'a, I>)
184 where
185 I: Iterator<Item = &'a tt::TokenTree>,
186 {
187 match tt {
188 tt::TokenTree::Leaf(token) => self.convert_token(token, iter),
189 tt::TokenTree::Subtree(sub) => self.convert_subtree(sub),
190 }
191 }
192
193 fn convert_token<'a, I>(&mut self, token: &tt::Leaf, iter: &mut TokenPeek<'a, I>)
194 where
195 I: Iterator<Item = &'a tt::TokenTree>,
196 {
197 let tok = match token {
198 tt::Leaf::Literal(l) => TtToken {
199 kind: classify_literal(&l.text).unwrap().kind,
200 is_joint_to_next: false,
201 text: l.text.clone(),
202 n_tokens: 1,
203 },
204 tt::Leaf::Punct(p) => {
205 if let Some((kind, is_joint_to_next, text, size)) =
206 Self::convert_multi_char_punct(p, iter)
207 {
208 for _ in 0..size - 1 {
209 iter.next();
210 }
211
212 TtToken { kind, is_joint_to_next, text: text.into(), n_tokens: size }
213 } else {
214 let kind = match p.char {
215 // lexer may produce combpund tokens for these ones
216 '.' => DOT,
217 ':' => COLON,
218 '=' => EQ,
219 '!' => EXCL,
220 '-' => MINUS,
221 c => SyntaxKind::from_char(c).unwrap(),
222 };
223 let text = {
224 let mut buf = [0u8; 4];
225 let s: &str = p.char.encode_utf8(&mut buf);
226 SmolStr::new(s)
227 };
228 TtToken {
229 kind,
230 is_joint_to_next: p.spacing == tt::Spacing::Joint,
231 text,
232 n_tokens: 1,
233 }
234 }
235 }
236 tt::Leaf::Ident(ident) => {
237 let kind = SyntaxKind::from_keyword(ident.text.as_str()).unwrap_or(IDENT);
238 TtToken { kind, is_joint_to_next: false, text: ident.text.clone(), n_tokens: 1 }
239 }
240 };
241 self.tokens.push(tok)
242 }
243
244 pub(crate) fn convert_multi_char_punct<'a, I>(
245 p: &tt::Punct,
246 iter: &mut TokenPeek<'a, I>,
247 ) -> Option<(SyntaxKind, bool, &'static str, usize)>
248 where
249 I: Iterator<Item = &'a tt::TokenTree>,
250 {
251 if let Some((m, is_joint_to_next)) = iter.current_punct3(p) {
252 if let Some((kind, text)) = match m {
253 ('<', '<', '=') => Some((SHLEQ, "<<=")),
254 ('>', '>', '=') => Some((SHREQ, ">>=")),
255 ('.', '.', '.') => Some((DOTDOTDOT, "...")),
256 ('.', '.', '=') => Some((DOTDOTEQ, "..=")),
257 _ => None,
258 } {
259 return Some((kind, is_joint_to_next, text, 3));
260 }
261 }
262
263 if let Some((m, is_joint_to_next)) = iter.current_punct2(p) {
264 if let Some((kind, text)) = match m {
265 ('<', '<') => Some((SHL, "<<")),
266 ('>', '>') => Some((SHR, ">>")),
267
268 ('|', '|') => Some((PIPEPIPE, "||")),
269 ('&', '&') => Some((AMPAMP, "&&")),
270 ('%', '=') => Some((PERCENTEQ, "%=")),
271 ('*', '=') => Some((STAREQ, "*=")),
272 ('/', '=') => Some((SLASHEQ, "/=")),
273 ('^', '=') => Some((CARETEQ, "^=")),
274
275 ('&', '=') => Some((AMPEQ, "&=")),
276 ('|', '=') => Some((PIPEEQ, "|=")),
277 ('-', '=') => Some((MINUSEQ, "-=")),
278 ('+', '=') => Some((PLUSEQ, "+=")),
279 ('>', '=') => Some((GTEQ, ">=")),
280 ('<', '=') => Some((LTEQ, "<=")),
281
282 ('-', '>') => Some((THIN_ARROW, "->")),
283 ('!', '=') => Some((NEQ, "!=")),
284 ('=', '>') => Some((FAT_ARROW, "=>")),
285 ('=', '=') => Some((EQEQ, "==")),
286 ('.', '.') => Some((DOTDOT, "..")),
287 (':', ':') => Some((COLONCOLON, "::")),
288
289 _ => None,
290 } {
291 return Some((kind, is_joint_to_next, text, 2));
292 }
293 }
294
295 None
296 }
297
298 fn push_delim(&mut self, d: tt::Delimiter, closing: bool) {
299 let (kinds, texts) = match d {
300 tt::Delimiter::Parenthesis => ([L_PAREN, R_PAREN], "()"),
301 tt::Delimiter::Brace => ([L_CURLY, R_CURLY], "{}"),
302 tt::Delimiter::Bracket => ([L_BRACK, R_BRACK], "[]"),
303 tt::Delimiter::None => return,
304 };
305 let idx = closing as usize;
306 let kind = kinds[idx];
307 let text = &texts[idx..texts.len() - (1 - idx)];
308 let tok = TtToken { kind, is_joint_to_next: false, text: SmolStr::new(text), n_tokens: 1 };
309 self.tokens.push(tok)
310 }
311}
312
313impl TokenSource for TtTokenSource {
314 fn token_kind(&self, pos: usize) -> SyntaxKind {
315 if let Some(tok) = self.tokens.get(pos) {
316 tok.kind
317 } else {
318 SyntaxKind::EOF
319 }
320 }
321 fn is_token_joint_to_next(&self, pos: usize) -> bool {
322 self.tokens[pos].is_joint_to_next
323 }
324 fn is_keyword(&self, pos: usize, kw: &str) -> bool {
325 self.tokens[pos].text == *kw
326 }
327}
328
329#[derive(Default)]
330struct TtTreeSink<'a> { 108struct TtTreeSink<'a> {
331 buf: String, 109 buf: String,
332 tokens: &'a [TtToken], 110 src_querier: SubtreeSourceQuerier<'a>,
333 text_pos: TextUnit, 111 text_pos: TextUnit,
334 token_pos: usize, 112 token_pos: usize,
335 inner: SyntaxTreeBuilder, 113 inner: SyntaxTreeBuilder,
336} 114}
337 115
338impl<'a> TtTreeSink<'a> { 116impl<'a> TtTreeSink<'a> {
339 fn new(tokens: &'a [TtToken]) -> TtTreeSink { 117 fn new(src_querier: SubtreeSourceQuerier<'a>) -> TtTreeSink {
340 TtTreeSink { 118 TtTreeSink {
341 buf: String::new(), 119 buf: String::new(),
342 tokens, 120 src_querier,
343 text_pos: 0.into(), 121 text_pos: 0.into(),
344 token_pos: 0, 122 token_pos: 0,
345 inner: SyntaxTreeBuilder::default(), 123 inner: SyntaxTreeBuilder::default(),
@@ -350,7 +128,7 @@ impl<'a> TtTreeSink<'a> {
350impl<'a> TreeSink for TtTreeSink<'a> { 128impl<'a> TreeSink for TtTreeSink<'a> {
351 fn token(&mut self, kind: SyntaxKind, n_tokens: u8) { 129 fn token(&mut self, kind: SyntaxKind, n_tokens: u8) {
352 for _ in 0..n_tokens { 130 for _ in 0..n_tokens {
353 self.buf += self.tokens[self.token_pos].text.as_str(); 131 self.buf += self.src_querier.token(self.token_pos).1;
354 self.token_pos += 1; 132 self.token_pos += 1;
355 } 133 }
356 self.text_pos += TextUnit::of_str(&self.buf); 134 self.text_pos += TextUnit::of_str(&self.buf);
@@ -394,21 +172,23 @@ mod tests {
394 "#, 172 "#,
395 ); 173 );
396 let expansion = expand(&rules, "literals!(foo)"); 174 let expansion = expand(&rules, "literals!(foo)");
397 let tt_src = TtTokenSource::new(&expansion); 175 let tt_src = SubtreeTokenSource::new(&expansion);
176
177 let query = tt_src.querier();
398 178
399 // [{] 179 // [{]
400 // [let] [a] [=] ['c'] [;] 180 // [let] [a] [=] ['c'] [;]
401 assert_eq!(tt_src.tokens[1 + 3].text, "'c'"); 181 assert_eq!(query.token(1 + 3).1, "'c'");
402 assert_eq!(tt_src.tokens[1 + 3].kind, CHAR); 182 assert_eq!(query.token(1 + 3).0, CHAR);
403 // [let] [c] [=] [1000] [;] 183 // [let] [c] [=] [1000] [;]
404 assert_eq!(tt_src.tokens[1 + 5 + 3].text, "1000"); 184 assert_eq!(query.token(1 + 5 + 3).1, "1000");
405 assert_eq!(tt_src.tokens[1 + 5 + 3].kind, INT_NUMBER); 185 assert_eq!(query.token(1 + 5 + 3).0, INT_NUMBER);
406 // [let] [f] [=] [12E+99_f64] [;] 186 // [let] [f] [=] [12E+99_f64] [;]
407 assert_eq!(tt_src.tokens[1 + 10 + 3].text, "12E+99_f64"); 187 assert_eq!(query.token(1 + 10 + 3).1, "12E+99_f64");
408 assert_eq!(tt_src.tokens[1 + 10 + 3].kind, FLOAT_NUMBER); 188 assert_eq!(query.token(1 + 10 + 3).0, FLOAT_NUMBER);
409 189
410 // [let] [s] [=] ["rust1"] [;] 190 // [let] [s] [=] ["rust1"] [;]
411 assert_eq!(tt_src.tokens[1 + 15 + 3].text, "\"rust1\""); 191 assert_eq!(query.token(1 + 15 + 3).1, "\"rust1\"");
412 assert_eq!(tt_src.tokens[1 + 15 + 3].kind, STRING); 192 assert_eq!(query.token(1 + 15 + 3).0, STRING);
413 } 193 }
414} 194}
diff --git a/crates/ra_mbe/src/tt_cursor.rs b/crates/ra_mbe/src/tt_cursor.rs
index 6ac3ac187..52e072599 100644
--- a/crates/ra_mbe/src/tt_cursor.rs
+++ b/crates/ra_mbe/src/tt_cursor.rs
@@ -1,116 +1,17 @@
1use crate::ParseError; 1use crate::ParseError;
2use crate::syntax_bridge::{TtTokenSource, TtToken, TokenPeek}; 2use crate::subtree_source::SubtreeTokenSource;
3
3use ra_parser::{TokenSource, TreeSink}; 4use ra_parser::{TokenSource, TreeSink};
4 5
5use ra_syntax::{ 6use ra_syntax::{
6 SyntaxKind 7 SyntaxKind
7}; 8};
8 9
9struct TtCursorTokenSource { 10struct SubtreeTokenSink {
10 tt_pos: usize,
11 inner: TtTokenSource,
12}
13
14impl TtCursorTokenSource {
15 fn new(subtree: &tt::Subtree, curr: usize) -> TtCursorTokenSource {
16 let mut res = TtCursorTokenSource { inner: TtTokenSource::new(subtree), tt_pos: 1 };
17
18 // Matching `TtToken` cursor to `tt::TokenTree` cursor
19 // It is because TtToken is not One to One mapping to tt::Token
20 // There are 3 case (`TtToken` <=> `tt::TokenTree`) :
21 // * One to One => ident, single char punch
22 // * Many to One => `tt::TokenTree::SubTree`
23 // * One to Many => multibyte punct
24 //
25 // Such that we cannot simpliy advance the cursor
26 // We have to bump it one by one
27 let mut pos = 0;
28 while pos < curr {
29 pos += res.bump(&subtree.token_trees[pos]);
30 }
31
32 res
33 }
34
35 fn skip_sibling_leaf(&self, leaf: &tt::Leaf, iter: &mut std::slice::Iter<tt::TokenTree>) {
36 if let tt::Leaf::Punct(p) = leaf {
37 let mut peek = TokenPeek::new(iter);
38 if let Some((_, _, _, size)) = TtTokenSource::convert_multi_char_punct(p, &mut peek) {
39 for _ in 0..size - 1 {
40 peek.next();
41 }
42 }
43 }
44 }
45
46 fn count_tt_tokens(
47 &self,
48 tt: &tt::TokenTree,
49 iter: Option<&mut std::slice::Iter<tt::TokenTree>>,
50 ) -> usize {
51 assert!(!self.inner.tokens.is_empty());
52
53 match tt {
54 tt::TokenTree::Subtree(sub_tree) => {
55 let mut iter = sub_tree.token_trees.iter();
56 let mut count = match sub_tree.delimiter {
57 tt::Delimiter::None => 0,
58 _ => 2,
59 };
60
61 while let Some(tt) = iter.next() {
62 count += self.count_tt_tokens(&tt, Some(&mut iter));
63 }
64 count
65 }
66
67 tt::TokenTree::Leaf(leaf) => {
68 iter.map(|iter| {
69 self.skip_sibling_leaf(leaf, iter);
70 });
71
72 1
73 }
74 }
75 }
76
77 fn count(&self, tt: &tt::TokenTree) -> usize {
78 self.count_tt_tokens(tt, None)
79 }
80
81 fn bump(&mut self, tt: &tt::TokenTree) -> usize {
82 let cur = self.current().unwrap();
83 let n_tokens = cur.n_tokens;
84 self.tt_pos += self.count(tt);
85 n_tokens
86 }
87
88 fn current(&self) -> Option<&TtToken> {
89 self.inner.tokens.get(self.tt_pos)
90 }
91}
92
93impl TokenSource for TtCursorTokenSource {
94 fn token_kind(&self, pos: usize) -> SyntaxKind {
95 if let Some(tok) = self.inner.tokens.get(self.tt_pos + pos) {
96 tok.kind
97 } else {
98 SyntaxKind::EOF
99 }
100 }
101 fn is_token_joint_to_next(&self, pos: usize) -> bool {
102 self.inner.tokens[self.tt_pos + pos].is_joint_to_next
103 }
104 fn is_keyword(&self, pos: usize, kw: &str) -> bool {
105 self.inner.tokens[self.tt_pos + pos].text == *kw
106 }
107}
108
109struct TtCursorTokenSink {
110 token_pos: usize, 11 token_pos: usize,
111} 12}
112 13
113impl TreeSink for TtCursorTokenSink { 14impl TreeSink for SubtreeTokenSink {
114 fn token(&mut self, _kind: SyntaxKind, n_tokens: u8) { 15 fn token(&mut self, _kind: SyntaxKind, n_tokens: u8) {
115 self.token_pos += n_tokens as usize; 16 self.token_pos += n_tokens as usize;
116 } 17 }
@@ -201,24 +102,10 @@ impl<'a> TtCursor<'a> {
201 fn eat_parse_result( 102 fn eat_parse_result(
202 &mut self, 103 &mut self,
203 parsed_token: usize, 104 parsed_token: usize,
204 src: &mut TtCursorTokenSource, 105 src: &mut SubtreeTokenSource,
205 ) -> Option<tt::TokenTree> { 106 ) -> Option<tt::TokenTree> {
206 let mut res = vec![]; 107 let (adv, res) = src.bump_n(parsed_token, self.pos);
207 108 self.pos += adv;
208 // Matching `TtToken` cursor to `tt::TokenTree` cursor
209 // It is because TtToken is not One to One mapping to tt::Token
210 // There are 3 case (`TtToken` <=> `tt::TokenTree`) :
211 // * One to One => ident, single char punch
212 // * Many to One => `tt::TokenTree::SubTree`
213 // * One to Many => multibyte punct
214 //
215 // Such that we cannot simpliy advance the cursor
216 // We have to bump it one by one
217 let next_pos = src.tt_pos + parsed_token;
218 while src.tt_pos < next_pos {
219 let n = src.bump(self.current().unwrap());
220 res.extend((0..n).map(|_| self.eat().unwrap()));
221 }
222 109
223 let res: Vec<_> = res.into_iter().cloned().collect(); 110 let res: Vec<_> = res.into_iter().cloned().collect();
224 111
@@ -236,8 +123,9 @@ impl<'a> TtCursor<'a> {
236 where 123 where
237 F: FnOnce(&dyn TokenSource, &mut dyn TreeSink), 124 F: FnOnce(&dyn TokenSource, &mut dyn TreeSink),
238 { 125 {
239 let mut src = TtCursorTokenSource::new(self.subtree, self.pos); 126 let mut src = SubtreeTokenSource::new(self.subtree);
240 let mut sink = TtCursorTokenSink { token_pos: 0 }; 127 src.advance(self.pos, true);
128 let mut sink = SubtreeTokenSink { token_pos: 0 };
241 129
242 f(&src, &mut sink); 130 f(&src, &mut sink);
243 131