diff options
author | Aleksey Kladov <[email protected]> | 2020-02-06 11:17:40 +0000 |
---|---|---|
committer | Aleksey Kladov <[email protected]> | 2020-02-06 11:17:40 +0000 |
commit | ee2ee1a8ff7ad13211d2f66d2f2f1daaf3a00bd9 (patch) | |
tree | c0956465525077f4318ff6575ec6a3ac24bc57ff /crates/ra_ide/src/ide_db/line_index_utils.rs | |
parent | 551f33d754c5131f548baee4f58fd5892a9c7efc (diff) |
Move line_index
Diffstat (limited to 'crates/ra_ide/src/ide_db/line_index_utils.rs')
-rw-r--r-- | crates/ra_ide/src/ide_db/line_index_utils.rs | 332 |
1 files changed, 332 insertions, 0 deletions
diff --git a/crates/ra_ide/src/ide_db/line_index_utils.rs b/crates/ra_ide/src/ide_db/line_index_utils.rs new file mode 100644 index 000000000..70bf7253c --- /dev/null +++ b/crates/ra_ide/src/ide_db/line_index_utils.rs | |||
@@ -0,0 +1,332 @@ | |||
1 | //! FIXME: write short doc here | ||
2 | |||
3 | use ra_syntax::{TextRange, TextUnit}; | ||
4 | use ra_text_edit::{AtomTextEdit, TextEdit}; | ||
5 | |||
6 | use crate::ide_db::line_index::{LineCol, LineIndex, Utf16Char}; | ||
7 | |||
8 | #[derive(Debug, Clone)] | ||
9 | enum Step { | ||
10 | Newline(TextUnit), | ||
11 | Utf16Char(TextRange), | ||
12 | } | ||
13 | |||
14 | #[derive(Debug)] | ||
15 | struct LineIndexStepIter<'a> { | ||
16 | line_index: &'a LineIndex, | ||
17 | next_newline_idx: usize, | ||
18 | utf16_chars: Option<(TextUnit, std::slice::Iter<'a, Utf16Char>)>, | ||
19 | } | ||
20 | |||
21 | impl<'a> LineIndexStepIter<'a> { | ||
22 | fn from(line_index: &LineIndex) -> LineIndexStepIter { | ||
23 | let mut x = LineIndexStepIter { line_index, next_newline_idx: 0, utf16_chars: None }; | ||
24 | // skip first newline since it's not real | ||
25 | x.next(); | ||
26 | x | ||
27 | } | ||
28 | } | ||
29 | |||
30 | impl<'a> Iterator for LineIndexStepIter<'a> { | ||
31 | type Item = Step; | ||
32 | fn next(&mut self) -> Option<Step> { | ||
33 | self.utf16_chars | ||
34 | .as_mut() | ||
35 | .and_then(|(newline, x)| { | ||
36 | let x = x.next()?; | ||
37 | Some(Step::Utf16Char(TextRange::from_to(*newline + x.start, *newline + x.end))) | ||
38 | }) | ||
39 | .or_else(|| { | ||
40 | let next_newline = *self.line_index.newlines.get(self.next_newline_idx)?; | ||
41 | self.utf16_chars = self | ||
42 | .line_index | ||
43 | .utf16_lines | ||
44 | .get(&(self.next_newline_idx as u32)) | ||
45 | .map(|x| (next_newline, x.iter())); | ||
46 | self.next_newline_idx += 1; | ||
47 | Some(Step::Newline(next_newline)) | ||
48 | }) | ||
49 | } | ||
50 | } | ||
51 | |||
52 | #[derive(Debug)] | ||
53 | struct OffsetStepIter<'a> { | ||
54 | text: &'a str, | ||
55 | offset: TextUnit, | ||
56 | } | ||
57 | |||
58 | impl<'a> Iterator for OffsetStepIter<'a> { | ||
59 | type Item = Step; | ||
60 | fn next(&mut self) -> Option<Step> { | ||
61 | let (next, next_offset) = self | ||
62 | .text | ||
63 | .char_indices() | ||
64 | .filter_map(|(i, c)| { | ||
65 | if c == '\n' { | ||
66 | let next_offset = self.offset + TextUnit::from_usize(i + 1); | ||
67 | let next = Step::Newline(next_offset); | ||
68 | Some((next, next_offset)) | ||
69 | } else { | ||
70 | let char_len = TextUnit::of_char(c); | ||
71 | if char_len.to_usize() > 1 { | ||
72 | let start = self.offset + TextUnit::from_usize(i); | ||
73 | let end = start + char_len; | ||
74 | let next = Step::Utf16Char(TextRange::from_to(start, end)); | ||
75 | let next_offset = end; | ||
76 | Some((next, next_offset)) | ||
77 | } else { | ||
78 | None | ||
79 | } | ||
80 | } | ||
81 | }) | ||
82 | .next()?; | ||
83 | let next_idx = (next_offset - self.offset).to_usize(); | ||
84 | self.text = &self.text[next_idx..]; | ||
85 | self.offset = next_offset; | ||
86 | Some(next) | ||
87 | } | ||
88 | } | ||
89 | |||
90 | #[derive(Debug)] | ||
91 | enum NextSteps<'a> { | ||
92 | Use, | ||
93 | ReplaceMany(OffsetStepIter<'a>), | ||
94 | AddMany(OffsetStepIter<'a>), | ||
95 | } | ||
96 | |||
97 | #[derive(Debug)] | ||
98 | struct TranslatedEdit<'a> { | ||
99 | delete: TextRange, | ||
100 | insert: &'a str, | ||
101 | diff: i64, | ||
102 | } | ||
103 | |||
104 | struct Edits<'a> { | ||
105 | edits: &'a [AtomTextEdit], | ||
106 | current: Option<TranslatedEdit<'a>>, | ||
107 | acc_diff: i64, | ||
108 | } | ||
109 | |||
110 | impl<'a> Edits<'a> { | ||
111 | fn from_text_edit(text_edit: &'a TextEdit) -> Edits<'a> { | ||
112 | let mut x = Edits { edits: text_edit.as_atoms(), current: None, acc_diff: 0 }; | ||
113 | x.advance_edit(); | ||
114 | x | ||
115 | } | ||
116 | fn advance_edit(&mut self) { | ||
117 | self.acc_diff += self.current.as_ref().map_or(0, |x| x.diff); | ||
118 | match self.edits.split_first() { | ||
119 | Some((next, rest)) => { | ||
120 | let delete = self.translate_range(next.delete); | ||
121 | let diff = next.insert.len() as i64 - next.delete.len().to_usize() as i64; | ||
122 | self.current = Some(TranslatedEdit { delete, insert: &next.insert, diff }); | ||
123 | self.edits = rest; | ||
124 | } | ||
125 | None => { | ||
126 | self.current = None; | ||
127 | } | ||
128 | } | ||
129 | } | ||
130 | |||
131 | fn next_inserted_steps(&mut self) -> Option<OffsetStepIter<'a>> { | ||
132 | let cur = self.current.as_ref()?; | ||
133 | let res = Some(OffsetStepIter { offset: cur.delete.start(), text: &cur.insert }); | ||
134 | self.advance_edit(); | ||
135 | res | ||
136 | } | ||
137 | |||
138 | fn next_steps(&mut self, step: &Step) -> NextSteps { | ||
139 | let step_pos = match *step { | ||
140 | Step::Newline(n) => n, | ||
141 | Step::Utf16Char(r) => r.end(), | ||
142 | }; | ||
143 | match &mut self.current { | ||
144 | Some(edit) => { | ||
145 | if step_pos <= edit.delete.start() { | ||
146 | NextSteps::Use | ||
147 | } else if step_pos <= edit.delete.end() { | ||
148 | let iter = OffsetStepIter { offset: edit.delete.start(), text: &edit.insert }; | ||
149 | // empty slice to avoid returning steps again | ||
150 | edit.insert = &edit.insert[edit.insert.len()..]; | ||
151 | NextSteps::ReplaceMany(iter) | ||
152 | } else { | ||
153 | let iter = OffsetStepIter { offset: edit.delete.start(), text: &edit.insert }; | ||
154 | // empty slice to avoid returning steps again | ||
155 | edit.insert = &edit.insert[edit.insert.len()..]; | ||
156 | self.advance_edit(); | ||
157 | NextSteps::AddMany(iter) | ||
158 | } | ||
159 | } | ||
160 | None => NextSteps::Use, | ||
161 | } | ||
162 | } | ||
163 | |||
164 | fn translate_range(&self, range: TextRange) -> TextRange { | ||
165 | if self.acc_diff == 0 { | ||
166 | range | ||
167 | } else { | ||
168 | let start = self.translate(range.start()); | ||
169 | let end = self.translate(range.end()); | ||
170 | TextRange::from_to(start, end) | ||
171 | } | ||
172 | } | ||
173 | |||
174 | fn translate(&self, x: TextUnit) -> TextUnit { | ||
175 | if self.acc_diff == 0 { | ||
176 | x | ||
177 | } else { | ||
178 | TextUnit::from((x.to_usize() as i64 + self.acc_diff) as u32) | ||
179 | } | ||
180 | } | ||
181 | |||
182 | fn translate_step(&self, x: &Step) -> Step { | ||
183 | if self.acc_diff == 0 { | ||
184 | x.clone() | ||
185 | } else { | ||
186 | match *x { | ||
187 | Step::Newline(n) => Step::Newline(self.translate(n)), | ||
188 | Step::Utf16Char(r) => Step::Utf16Char(self.translate_range(r)), | ||
189 | } | ||
190 | } | ||
191 | } | ||
192 | } | ||
193 | |||
194 | #[derive(Debug)] | ||
195 | struct RunningLineCol { | ||
196 | line: u32, | ||
197 | last_newline: TextUnit, | ||
198 | col_adjust: TextUnit, | ||
199 | } | ||
200 | |||
201 | impl RunningLineCol { | ||
202 | fn new() -> RunningLineCol { | ||
203 | RunningLineCol { line: 0, last_newline: TextUnit::from(0), col_adjust: TextUnit::from(0) } | ||
204 | } | ||
205 | |||
206 | fn to_line_col(&self, offset: TextUnit) -> LineCol { | ||
207 | LineCol { | ||
208 | line: self.line, | ||
209 | col_utf16: ((offset - self.last_newline) - self.col_adjust).into(), | ||
210 | } | ||
211 | } | ||
212 | |||
213 | fn add_line(&mut self, newline: TextUnit) { | ||
214 | self.line += 1; | ||
215 | self.last_newline = newline; | ||
216 | self.col_adjust = TextUnit::from(0); | ||
217 | } | ||
218 | |||
219 | fn adjust_col(&mut self, range: TextRange) { | ||
220 | self.col_adjust += range.len() - TextUnit::from(1); | ||
221 | } | ||
222 | } | ||
223 | |||
224 | pub fn translate_offset_with_edit( | ||
225 | line_index: &LineIndex, | ||
226 | offset: TextUnit, | ||
227 | text_edit: &TextEdit, | ||
228 | ) -> LineCol { | ||
229 | let mut state = Edits::from_text_edit(&text_edit); | ||
230 | |||
231 | let mut res = RunningLineCol::new(); | ||
232 | |||
233 | macro_rules! test_step { | ||
234 | ($x:ident) => { | ||
235 | match &$x { | ||
236 | Step::Newline(n) => { | ||
237 | if offset < *n { | ||
238 | return res.to_line_col(offset); | ||
239 | } else { | ||
240 | res.add_line(*n); | ||
241 | } | ||
242 | } | ||
243 | Step::Utf16Char(x) => { | ||
244 | if offset < x.end() { | ||
245 | // if the offset is inside a multibyte char it's invalid | ||
246 | // clamp it to the start of the char | ||
247 | let clamp = offset.min(x.start()); | ||
248 | return res.to_line_col(clamp); | ||
249 | } else { | ||
250 | res.adjust_col(*x); | ||
251 | } | ||
252 | } | ||
253 | } | ||
254 | }; | ||
255 | } | ||
256 | |||
257 | for orig_step in LineIndexStepIter::from(line_index) { | ||
258 | loop { | ||
259 | let translated_step = state.translate_step(&orig_step); | ||
260 | match state.next_steps(&translated_step) { | ||
261 | NextSteps::Use => { | ||
262 | test_step!(translated_step); | ||
263 | break; | ||
264 | } | ||
265 | NextSteps::ReplaceMany(ns) => { | ||
266 | for n in ns { | ||
267 | test_step!(n); | ||
268 | } | ||
269 | break; | ||
270 | } | ||
271 | NextSteps::AddMany(ns) => { | ||
272 | for n in ns { | ||
273 | test_step!(n); | ||
274 | } | ||
275 | } | ||
276 | } | ||
277 | } | ||
278 | } | ||
279 | |||
280 | loop { | ||
281 | match state.next_inserted_steps() { | ||
282 | None => break, | ||
283 | Some(ns) => { | ||
284 | for n in ns { | ||
285 | test_step!(n); | ||
286 | } | ||
287 | } | ||
288 | } | ||
289 | } | ||
290 | |||
291 | res.to_line_col(offset) | ||
292 | } | ||
293 | |||
294 | #[cfg(test)] | ||
295 | mod test { | ||
296 | use super::*; | ||
297 | use crate::line_index; | ||
298 | use proptest::{prelude::*, proptest}; | ||
299 | use ra_text_edit::test_utils::{arb_offset, arb_text_with_edit}; | ||
300 | use ra_text_edit::TextEdit; | ||
301 | |||
302 | #[derive(Debug)] | ||
303 | struct ArbTextWithEditAndOffset { | ||
304 | text: String, | ||
305 | edit: TextEdit, | ||
306 | edited_text: String, | ||
307 | offset: TextUnit, | ||
308 | } | ||
309 | |||
310 | fn arb_text_with_edit_and_offset() -> BoxedStrategy<ArbTextWithEditAndOffset> { | ||
311 | arb_text_with_edit() | ||
312 | .prop_flat_map(|x| { | ||
313 | let edited_text = x.edit.apply(&x.text); | ||
314 | let arb_offset = arb_offset(&edited_text); | ||
315 | (Just(x), Just(edited_text), arb_offset).prop_map(|(x, edited_text, offset)| { | ||
316 | ArbTextWithEditAndOffset { text: x.text, edit: x.edit, edited_text, offset } | ||
317 | }) | ||
318 | }) | ||
319 | .boxed() | ||
320 | } | ||
321 | |||
322 | proptest! { | ||
323 | #[test] | ||
324 | fn test_translate_offset_with_edit(x in arb_text_with_edit_and_offset()) { | ||
325 | let expected = line_index::to_line_col(&x.edited_text, x.offset); | ||
326 | let line_index = LineIndex::new(&x.text); | ||
327 | let actual = translate_offset_with_edit(&line_index, x.offset, &x.edit); | ||
328 | |||
329 | assert_eq!(actual, expected); | ||
330 | } | ||
331 | } | ||
332 | } | ||