diff options
Diffstat (limited to 'crates/ra_syntax/src/reparsing.rs')
-rw-r--r-- | crates/ra_syntax/src/reparsing.rs | 343 |
1 files changed, 343 insertions, 0 deletions
diff --git a/crates/ra_syntax/src/reparsing.rs b/crates/ra_syntax/src/reparsing.rs new file mode 100644 index 000000000..da44913c5 --- /dev/null +++ b/crates/ra_syntax/src/reparsing.rs | |||
@@ -0,0 +1,343 @@ | |||
1 | use algo; | ||
2 | use grammar; | ||
3 | use lexer::{tokenize, Token}; | ||
4 | use text_unit::{TextRange, TextUnit}; | ||
5 | use yellow::{self, SyntaxNodeRef, GreenNode, SyntaxError}; | ||
6 | use parser_impl; | ||
7 | use parser_api::Parser; | ||
8 | use { | ||
9 | SyntaxKind::*, | ||
10 | }; | ||
11 | use text_utils::replace_range; | ||
12 | |||
13 | #[derive(Debug, Clone)] | ||
14 | pub struct AtomEdit { | ||
15 | pub delete: TextRange, | ||
16 | pub insert: String, | ||
17 | } | ||
18 | |||
19 | impl AtomEdit { | ||
20 | pub fn replace(range: TextRange, replace_with: String) -> AtomEdit { | ||
21 | AtomEdit { delete: range, insert: replace_with } | ||
22 | } | ||
23 | |||
24 | pub fn delete(range: TextRange) -> AtomEdit { | ||
25 | AtomEdit::replace(range, String::new()) | ||
26 | } | ||
27 | |||
28 | pub fn insert(offset: TextUnit, text: String) -> AtomEdit { | ||
29 | AtomEdit::replace(TextRange::offset_len(offset, 0.into()), text) | ||
30 | } | ||
31 | } | ||
32 | |||
33 | pub(crate) fn incremental_reparse( | ||
34 | node: SyntaxNodeRef, | ||
35 | edit: &AtomEdit, | ||
36 | errors: Vec<SyntaxError>, | ||
37 | ) -> Option<(GreenNode, Vec<SyntaxError>)> { | ||
38 | let (node, green, new_errors) = | ||
39 | reparse_leaf(node, &edit).or_else(|| reparse_block(node, &edit))?; | ||
40 | let green_root = node.replace_with(green); | ||
41 | let errors = merge_errors(errors, new_errors, node, edit); | ||
42 | Some((green_root, errors)) | ||
43 | } | ||
44 | |||
45 | fn reparse_leaf<'node>( | ||
46 | node: SyntaxNodeRef<'node>, | ||
47 | edit: &AtomEdit, | ||
48 | ) -> Option<(SyntaxNodeRef<'node>, GreenNode, Vec<SyntaxError>)> { | ||
49 | let node = algo::find_covering_node(node, edit.delete); | ||
50 | match node.kind() { | ||
51 | | WHITESPACE | ||
52 | | COMMENT | ||
53 | | DOC_COMMENT | ||
54 | | IDENT | ||
55 | | STRING | ||
56 | | RAW_STRING => { | ||
57 | let text = get_text_after_edit(node, &edit); | ||
58 | let tokens = tokenize(&text); | ||
59 | let token = match tokens[..] { | ||
60 | [token] if token.kind == node.kind() => token, | ||
61 | _ => return None, | ||
62 | }; | ||
63 | |||
64 | if token.kind == IDENT && is_contextual_kw(&text) { | ||
65 | return None; | ||
66 | } | ||
67 | |||
68 | let green = GreenNode::new_leaf(node.kind(), &text); | ||
69 | let new_errors = vec![]; | ||
70 | Some((node, green, new_errors)) | ||
71 | } | ||
72 | _ => None, | ||
73 | } | ||
74 | } | ||
75 | |||
76 | fn reparse_block<'node>( | ||
77 | node: SyntaxNodeRef<'node>, | ||
78 | edit: &AtomEdit, | ||
79 | ) -> Option<(SyntaxNodeRef<'node>, GreenNode, Vec<SyntaxError>)> { | ||
80 | let (node, reparser) = find_reparsable_node(node, edit.delete)?; | ||
81 | let text = get_text_after_edit(node, &edit); | ||
82 | let tokens = tokenize(&text); | ||
83 | if !is_balanced(&tokens) { | ||
84 | return None; | ||
85 | } | ||
86 | let (green, new_errors) = | ||
87 | parser_impl::parse_with::<yellow::GreenBuilder>( | ||
88 | &text, &tokens, reparser, | ||
89 | ); | ||
90 | Some((node, green, new_errors)) | ||
91 | } | ||
92 | |||
93 | fn get_text_after_edit(node: SyntaxNodeRef, edit: &AtomEdit) -> String { | ||
94 | replace_range( | ||
95 | node.text().to_string(), | ||
96 | edit.delete - node.range().start(), | ||
97 | &edit.insert, | ||
98 | ) | ||
99 | } | ||
100 | |||
101 | fn is_contextual_kw(text: &str) -> bool { | ||
102 | match text { | ||
103 | | "auto" | ||
104 | | "default" | ||
105 | | "union" => true, | ||
106 | _ => false, | ||
107 | } | ||
108 | } | ||
109 | |||
110 | fn find_reparsable_node<'node>( | ||
111 | node: SyntaxNodeRef<'node>, | ||
112 | range: TextRange, | ||
113 | ) -> Option<(SyntaxNodeRef<'node>, fn(&mut Parser))> { | ||
114 | let node = algo::find_covering_node(node, range); | ||
115 | return algo::ancestors(node) | ||
116 | .filter_map(|node| reparser(node).map(|r| (node, r))) | ||
117 | .next(); | ||
118 | |||
119 | fn reparser(node: SyntaxNodeRef) -> Option<fn(&mut Parser)> { | ||
120 | let res = match node.kind() { | ||
121 | BLOCK => grammar::block, | ||
122 | NAMED_FIELD_DEF_LIST => grammar::named_field_def_list, | ||
123 | NAMED_FIELD_LIST => grammar::named_field_list, | ||
124 | ENUM_VARIANT_LIST => grammar::enum_variant_list, | ||
125 | MATCH_ARM_LIST => grammar::match_arm_list, | ||
126 | USE_TREE_LIST => grammar::use_tree_list, | ||
127 | EXTERN_ITEM_LIST => grammar::extern_item_list, | ||
128 | TOKEN_TREE if node.first_child().unwrap().kind() == L_CURLY => grammar::token_tree, | ||
129 | ITEM_LIST => { | ||
130 | let parent = node.parent().unwrap(); | ||
131 | match parent.kind() { | ||
132 | IMPL_ITEM => grammar::impl_item_list, | ||
133 | TRAIT_DEF => grammar::trait_item_list, | ||
134 | MODULE => grammar::mod_item_list, | ||
135 | _ => return None, | ||
136 | } | ||
137 | } | ||
138 | _ => return None, | ||
139 | }; | ||
140 | Some(res) | ||
141 | } | ||
142 | } | ||
143 | |||
144 | fn is_balanced(tokens: &[Token]) -> bool { | ||
145 | if tokens.len() == 0 | ||
146 | || tokens.first().unwrap().kind != L_CURLY | ||
147 | || tokens.last().unwrap().kind != R_CURLY { | ||
148 | return false; | ||
149 | } | ||
150 | let mut balance = 0usize; | ||
151 | for t in tokens.iter() { | ||
152 | match t.kind { | ||
153 | L_CURLY => balance += 1, | ||
154 | R_CURLY => balance = match balance.checked_sub(1) { | ||
155 | Some(b) => b, | ||
156 | None => return false, | ||
157 | }, | ||
158 | _ => (), | ||
159 | } | ||
160 | } | ||
161 | balance == 0 | ||
162 | } | ||
163 | |||
164 | fn merge_errors( | ||
165 | old_errors: Vec<SyntaxError>, | ||
166 | new_errors: Vec<SyntaxError>, | ||
167 | old_node: SyntaxNodeRef, | ||
168 | edit: &AtomEdit, | ||
169 | ) -> Vec<SyntaxError> { | ||
170 | let mut res = Vec::new(); | ||
171 | for e in old_errors { | ||
172 | if e.offset <= old_node.range().start() { | ||
173 | res.push(e) | ||
174 | } else if e.offset >= old_node.range().end() { | ||
175 | res.push(SyntaxError { | ||
176 | msg: e.msg, | ||
177 | offset: e.offset + TextUnit::of_str(&edit.insert) - edit.delete.len(), | ||
178 | }) | ||
179 | } | ||
180 | } | ||
181 | for e in new_errors { | ||
182 | res.push(SyntaxError { | ||
183 | msg: e.msg, | ||
184 | offset: e.offset + old_node.range().start(), | ||
185 | }) | ||
186 | } | ||
187 | res | ||
188 | } | ||
189 | |||
190 | #[cfg(test)] | ||
191 | mod tests { | ||
192 | use super::{ | ||
193 | super::{ | ||
194 | File, | ||
195 | test_utils::extract_range, | ||
196 | text_utils::replace_range, | ||
197 | utils::dump_tree, | ||
198 | }, | ||
199 | reparse_leaf, reparse_block, AtomEdit, GreenNode, SyntaxError, SyntaxNodeRef, | ||
200 | }; | ||
201 | |||
202 | fn do_check<F>( | ||
203 | before: &str, | ||
204 | replace_with: &str, | ||
205 | reparser: F, | ||
206 | ) where | ||
207 | for<'a> F: Fn( | ||
208 | SyntaxNodeRef<'a>, | ||
209 | &AtomEdit, | ||
210 | ) -> Option<(SyntaxNodeRef<'a>, GreenNode, Vec<SyntaxError>)> | ||
211 | { | ||
212 | let (range, before) = extract_range(before); | ||
213 | let after = replace_range(before.clone(), range, replace_with); | ||
214 | |||
215 | let fully_reparsed = File::parse(&after); | ||
216 | let incrementally_reparsed = { | ||
217 | let f = File::parse(&before); | ||
218 | let edit = AtomEdit { delete: range, insert: replace_with.to_string() }; | ||
219 | let (node, green, new_errors) = | ||
220 | reparser(f.syntax(), &edit).expect("cannot incrementally reparse"); | ||
221 | let green_root = node.replace_with(green); | ||
222 | let errors = super::merge_errors(f.errors(), new_errors, node, &edit); | ||
223 | File::new(green_root, errors) | ||
224 | }; | ||
225 | |||
226 | assert_eq_text!( | ||
227 | &dump_tree(fully_reparsed.syntax()), | ||
228 | &dump_tree(incrementally_reparsed.syntax()), | ||
229 | ) | ||
230 | } | ||
231 | |||
232 | #[test] | ||
233 | fn reparse_block_tests() { | ||
234 | let do_check = |before, replace_to| | ||
235 | do_check(before, replace_to, reparse_block); | ||
236 | |||
237 | do_check(r" | ||
238 | fn foo() { | ||
239 | let x = foo + <|>bar<|> | ||
240 | } | ||
241 | ", "baz"); | ||
242 | do_check(r" | ||
243 | fn foo() { | ||
244 | let x = foo<|> + bar<|> | ||
245 | } | ||
246 | ", "baz"); | ||
247 | do_check(r" | ||
248 | struct Foo { | ||
249 | f: foo<|><|> | ||
250 | } | ||
251 | ", ",\n g: (),"); | ||
252 | do_check(r" | ||
253 | fn foo { | ||
254 | let; | ||
255 | 1 + 1; | ||
256 | <|>92<|>; | ||
257 | } | ||
258 | ", "62"); | ||
259 | do_check(r" | ||
260 | mod foo { | ||
261 | fn <|><|> | ||
262 | } | ||
263 | ", "bar"); | ||
264 | do_check(r" | ||
265 | trait Foo { | ||
266 | type <|>Foo<|>; | ||
267 | } | ||
268 | ", "Output"); | ||
269 | do_check(r" | ||
270 | impl IntoIterator<Item=i32> for Foo { | ||
271 | f<|><|> | ||
272 | } | ||
273 | ", "n next("); | ||
274 | do_check(r" | ||
275 | use a::b::{foo,<|>,bar<|>}; | ||
276 | ", "baz"); | ||
277 | do_check(r" | ||
278 | pub enum A { | ||
279 | Foo<|><|> | ||
280 | } | ||
281 | ", "\nBar;\n"); | ||
282 | do_check(r" | ||
283 | foo!{a, b<|><|> d} | ||
284 | ", ", c[3]"); | ||
285 | do_check(r" | ||
286 | fn foo() { | ||
287 | vec![<|><|>] | ||
288 | } | ||
289 | ", "123"); | ||
290 | do_check(r" | ||
291 | extern { | ||
292 | fn<|>;<|> | ||
293 | } | ||
294 | ", " exit(code: c_int)"); | ||
295 | } | ||
296 | |||
297 | #[test] | ||
298 | fn reparse_leaf_tests() { | ||
299 | let do_check = |before, replace_to| | ||
300 | do_check(before, replace_to, reparse_leaf); | ||
301 | |||
302 | do_check(r"<|><|> | ||
303 | fn foo() -> i32 { 1 } | ||
304 | ", "\n\n\n \n"); | ||
305 | do_check(r" | ||
306 | fn foo() -> <|><|> {} | ||
307 | ", " \n"); | ||
308 | do_check(r" | ||
309 | fn <|>foo<|>() -> i32 { 1 } | ||
310 | ", "bar"); | ||
311 | do_check(r" | ||
312 | fn foo<|><|>foo() { } | ||
313 | ", "bar"); | ||
314 | do_check(r" | ||
315 | fn foo /* <|><|> */ () {} | ||
316 | ", "some comment"); | ||
317 | do_check(r" | ||
318 | fn baz <|><|> () {} | ||
319 | ", " \t\t\n\n"); | ||
320 | do_check(r" | ||
321 | fn baz <|><|> () {} | ||
322 | ", " \t\t\n\n"); | ||
323 | do_check(r" | ||
324 | /// foo <|><|>omment | ||
325 | mod { } | ||
326 | ", "c"); | ||
327 | do_check(r#" | ||
328 | fn -> &str { "Hello<|><|>" } | ||
329 | "#, ", world"); | ||
330 | do_check(r#" | ||
331 | fn -> &str { // "Hello<|><|>" | ||
332 | "#, ", world"); | ||
333 | do_check(r##" | ||
334 | fn -> &str { r#"Hello<|><|>"# | ||
335 | "##, ", world"); | ||
336 | do_check(r" | ||
337 | #[derive(<|>Copy<|>)] | ||
338 | enum Foo { | ||
339 | |||
340 | } | ||
341 | ", "Clone"); | ||
342 | } | ||
343 | } \ No newline at end of file | ||