git.iko.soy/helix
on master help sign in
Merge pull request #224 from helix-editor/line_ending_detection

Line ending detection

commita70de6e980ec58cabf58c33e8b91bfafbea312eb authorBlaž Hrastnik <blaz@mxxn.io> dateTue 22 Jun 2021, 11:09 (5 years ago) parentc704970f2954fa
17 files changed+562−205
compare with master
unifiedsplit
Cargo.lock+1 helix-core/src/auto_pairs.rs+1−1 helix-core/src/chars.rs+107−15 helix-core/src/lib.rs+3 helix-core/src/line_ending.rs+252 helix-core/src/movement.rs+38−128 helix-core/src/position.rs+4−2 helix-core/src/syntax.rs+5−4 helix-lsp/src/client.rs+4−3 helix-term/src/commands.rs+111−30 helix-term/src/ui/editor.rs+2−2 helix-term/src/ui/markdown.rs+2 helix-tui/Cargo.toml+1 helix-tui/src/text.rs+2−1 helix-tui/src/widgets/reflow.rs+5−4 helix-view/src/document.rs+22−12 helix-view/src/editor.rs+2−3
Cargo.lock+1
@@ -344,6 +344,7 @@ dependencies = [
344344 "bitflags",
345345 "cassowary",
346346 "crossterm",
347+ "helix-core",
347348 "serde",
348349 "unicode-segmentation",
349350 "unicode-width",
helix-core/src/auto_pairs.rs+1−1
@@ -12,7 +12,7 @@ pub const PAIRS: &[(char, char)] = &[
1212 ('`', '`'),
1313];
1414
15−const CLOSE_BEFORE: &str = ")]}'\":;> \n"; // includes space and newline
15+const CLOSE_BEFORE: &str = ")]}'\":;> \n\r\u{000B}\u{000C}\u{0085}\u{2028}\u{2029}"; // includes space and newlines
1616
1717// insert hook:
1818// Fn(doc, selection, char) => Option<Transaction>
helix-core/src/chars.rs+107−15
@@ -1,25 +1,44 @@
1−/// Determine whether a character is a line break.
2−pub fn char_is_linebreak(c: char) -> bool {
3− matches!(
4− c,
5− '\u{000A}' | // LineFeed
6− '\u{000B}' | // VerticalTab
7− '\u{000C}' | // FormFeed
8− '\u{000D}' | // CarriageReturn
9− '\u{0085}' | // NextLine
10− '\u{2028}' | // Line Separator
11− '\u{2029}' // ParagraphSeparator
12− )
1+use crate::LineEnding;
2+
3+#[derive(Debug, Eq, PartialEq)]
4+pub enum CharCategory {
5+ Whitespace,
6+ Eol,
7+ Word,
8+ Punctuation,
9+ Unknown,
10+}
11+
12+#[inline]
13+pub fn categorize_char(ch: char) -> CharCategory {
14+ if char_is_line_ending(ch) {
15+ CharCategory::Eol
16+ } else if ch.is_whitespace() {
17+ CharCategory::Whitespace
18+ } else if char_is_word(ch) {
19+ CharCategory::Word
20+ } else if char_is_punctuation(ch) {
21+ CharCategory::Punctuation
22+ } else {
23+ CharCategory::Unknown
24+ }
25+}
26+
27+/// Determine whether a character is a line ending.
28+#[inline]
29+pub fn char_is_line_ending(ch: char) -> bool {
30+ LineEnding::from_char(ch).is_some()
1331}
1432
1533/// Determine whether a character qualifies as (non-line-break)
1634/// whitespace.
17−pub fn char_is_whitespace(c: char) -> bool {
35+#[inline]
36+pub fn char_is_whitespace(ch: char) -> bool {
1837 // TODO: this is a naive binary categorization of whitespace
1938 // characters. For display, word wrapping, etc. we'll need a better
2039 // categorization based on e.g. breaking vs non-breaking spaces
2140 // and whether they're zero-width or not.
22− match c {
41+ match ch {
2342 //'\u{1680}' | // Ogham Space Mark (here for completeness, but usually displayed as a dash, not as whitespace)
2443 '\u{0009}' | // Character Tabulation
2544 '\u{0020}' | // Space
@@ -34,8 +53,81 @@ pub fn char_is_whitespace(c: char) -> bool {
3453 // En Quad, Em Quad, En Space, Em Space, Three-per-em Space,
3554 // Four-per-em Space, Six-per-em Space, Figure Space,
3655 // Punctuation Space, Thin Space, Hair Space, Zero Width Space.
37− c if ('\u{2000}' ..= '\u{200B}').contains(&c) => true,
56+ ch if ('\u{2000}' ..= '\u{200B}').contains(&ch) => true,
3857
3958 _ => false,
4059 }
4160}
61+
62+#[inline]
63+pub fn char_is_punctuation(ch: char) -> bool {
64+ use unicode_general_category::{get_general_category, GeneralCategory};
65+
66+ matches!(
67+ get_general_category(ch),
68+ GeneralCategory::OtherPunctuation
69+ | GeneralCategory::OpenPunctuation
70+ | GeneralCategory::ClosePunctuation
71+ | GeneralCategory::InitialPunctuation
72+ | GeneralCategory::FinalPunctuation
73+ | GeneralCategory::ConnectorPunctuation
74+ | GeneralCategory::DashPunctuation
75+ | GeneralCategory::MathSymbol
76+ | GeneralCategory::CurrencySymbol
77+ | GeneralCategory::ModifierSymbol
78+ )
79+}
80+
81+#[inline]
82+pub fn char_is_word(ch: char) -> bool {
83+ ch.is_alphanumeric() || ch == '_'
84+}
85+
86+#[cfg(test)]
87+mod test {
88+ use super::*;
89+
90+ #[test]
91+ fn test_categorize() {
92+ const EOL_TEST_CASE: &'static str = "\n\r\u{000B}\u{000C}\u{0085}\u{2028}\u{2029}";
93+ const WORD_TEST_CASE: &'static str =
94+ "_hello_world_あいうえおー1234567890１２３４５６７８９０";
95+ const PUNCTUATION_TEST_CASE: &'static str =
96+ "!\"#$%&\'()*+,-./:;<=>?@[\\]^`{|}~！”＃＄％＆’（）＊＋、。：；＜＝＞？＠「」＾｀｛｜｝～";
97+ const WHITESPACE_TEST_CASE: &'static str = "  　   ";
98+
99+ for ch in EOL_TEST_CASE.chars() {
100+ assert_eq!(CharCategory::Eol, categorize_char(ch));
101+ }
102+
103+ for ch in WHITESPACE_TEST_CASE.chars() {
104+ assert_eq!(
105+ CharCategory::Whitespace,
106+ categorize_char(ch),
107+ "Testing '{}', but got `{:?}` instead of `Category::Whitespace`",
108+ ch,
109+ categorize_char(ch)
110+ );
111+ }
112+
113+ for ch in WORD_TEST_CASE.chars() {
114+ assert_eq!(
115+ CharCategory::Word,
116+ categorize_char(ch),
117+ "Testing '{}', but got `{:?}` instead of `Category::Word`",
118+ ch,
119+ categorize_char(ch)
120+ );
121+ }
122+
123+ for ch in PUNCTUATION_TEST_CASE.chars() {
124+ assert_eq!(
125+ CharCategory::Punctuation,
126+ categorize_char(ch),
127+ "Testing '{}', but got `{:?}` instead of `Category::Punctuation`",
128+ ch,
129+ categorize_char(ch)
130+ );
131+ }
132+ }
133+}
helix-core/src/lib.rs+3
@@ -6,6 +6,7 @@ pub mod diagnostic;
66pub mod graphemes;
77pub mod history;
88pub mod indent;
9+pub mod line_ending;
910pub mod macros;
1011pub mod match_brackets;
1112pub mod movement;
@@ -106,6 +107,7 @@ pub use tendril::StrTendril as Tendril;
106107#[doc(inline)]
107108pub use {regex, tree_sitter};
108109
110+pub use graphemes::RopeGraphemes;
109111pub use position::{coords_at_pos, pos_at_coords, Position};
110112pub use selection::{Range, Selection};
111113pub use smallvec::SmallVec;
@@ -114,4 +116,5 @@ pub use syntax::Syntax;
114116pub use diagnostic::Diagnostic;
115117pub use state::State;
116118
119+pub use line_ending::{LineEnding, DEFAULT_LINE_ENDING};
117120pub use transaction::{Assoc, Change, ChangeSet, Operation, Transaction};
helix-core/src/line_ending.rsnew file+252
@@ -0,0 +1,252 @@
1+use crate::{Rope, RopeGraphemes, RopeSlice};
2+
3+#[cfg(target_os = "windows")]
4+pub const DEFAULT_LINE_ENDING: LineEnding = LineEnding::Crlf;
5+#[cfg(not(target_os = "windows"))]
6+pub const DEFAULT_LINE_ENDING: LineEnding = LineEnding::LF;
7+
8+/// Represents one of the valid Unicode line endings.
9+#[derive(PartialEq, Copy, Clone, Debug)]
10+pub enum LineEnding {
11+ Crlf, // CarriageReturn followed by LineFeed
12+ LF, // U+000A -- LineFeed
13+ VT, // U+000B -- VerticalTab
14+ FF, // U+000C -- FormFeed
15+ CR, // U+000D -- CarriageReturn
16+ Nel, // U+0085 -- NextLine
17+ LS, // U+2028 -- Line Separator
18+ PS, // U+2029 -- ParagraphSeparator
19+}
20+
21+impl LineEnding {
22+ #[inline]
23+ pub fn len_chars(&self) -> usize {
24+ match self {
25+ Self::Crlf => 2,
26+ _ => 1,
27+ }
28+ }
29+
30+ #[inline]
31+ pub fn as_str(&self) -> &'static str {
32+ match self {
33+ Self::Crlf => "\u{000D}\u{000A}",
34+ Self::LF => "\u{000A}",
35+ Self::VT => "\u{000B}",
36+ Self::FF => "\u{000C}",
37+ Self::CR => "\u{000D}",
38+ Self::Nel => "\u{0085}",
39+ Self::LS => "\u{2028}",
40+ Self::PS => "\u{2029}",
41+ }
42+ }
43+
44+ #[inline]
45+ pub fn from_char(ch: char) -> Option<LineEnding> {
46+ match ch {
47+ '\u{000A}' => Some(LineEnding::LF),
48+ '\u{000B}' => Some(LineEnding::VT),
49+ '\u{000C}' => Some(LineEnding::FF),
50+ '\u{000D}' => Some(LineEnding::CR),
51+ '\u{0085}' => Some(LineEnding::Nel),
52+ '\u{2028}' => Some(LineEnding::LS),
53+ '\u{2029}' => Some(LineEnding::PS),
54+ // Not a line ending
55+ _ => None,
56+ }
57+ }
58+
59+ // Normally we'd want to implement the FromStr trait, but in this case
60+ // that would force us into a different return type than from_char or
61+ // or from_rope_slice, which would be weird.
62+ #[allow(clippy::should_implement_trait)]
63+ #[inline]
64+ pub fn from_str(g: &str) -> Option<LineEnding> {
65+ match g {
66+ "\u{000D}\u{000A}" => Some(LineEnding::Crlf),
67+ "\u{000A}" => Some(LineEnding::LF),
68+ "\u{000B}" => Some(LineEnding::VT),
69+ "\u{000C}" => Some(LineEnding::FF),
70+ "\u{000D}" => Some(LineEnding::CR),
71+ "\u{0085}" => Some(LineEnding::Nel),
72+ "\u{2028}" => Some(LineEnding::LS),
73+ "\u{2029}" => Some(LineEnding::PS),
74+ // Not a line ending
75+ _ => None,
76+ }
77+ }
78+
79+ #[inline]
80+ pub fn from_rope_slice(g: &RopeSlice) -> Option<LineEnding> {
81+ if let Some(text) = g.as_str() {
82+ LineEnding::from_str(text)
83+ } else {
84+ // Non-contiguous, so it can't be a line ending.
85+ // Specifically, Ropey guarantees that CRLF is always
86+ // contiguous. And the remaining line endings are all
87+ // single `char`s, and therefore trivially contiguous.
88+ None
89+ }
90+ }
91+}
92+
93+#[inline]
94+pub fn str_is_line_ending(s: &str) -> bool {
95+ LineEnding::from_str(s).is_some()
96+}
97+
98+/// Attempts to detect what line ending the passed document uses.
99+pub fn auto_detect_line_ending(doc: &Rope) -> Option<LineEnding> {
100+ // Return first matched line ending. Not all possible line endings
101+ // are being matched, as they might be special-use only
102+ for line in doc.lines().take(100) {
103+ match get_line_ending(&line) {
104+ None | Some(LineEnding::VT) | Some(LineEnding::FF) | Some(LineEnding::PS) => {}
105+ ending => return ending,
106+ }
107+ }
108+ None
109+}
110+
111+/// Returns the passed line's line ending, if any.
112+pub fn get_line_ending(line: &RopeSlice) -> Option<LineEnding> {
113+ // Last character as str.
114+ let g1 = line
115+ .slice(line.len_chars().saturating_sub(1)..)
116+ .as_str()
117+ .unwrap();
118+
119+ // Last two characters as str, or empty str if they're not contiguous.
120+ // It's fine to punt on the non-contiguous case, because Ropey guarantees
121+ // that CRLF is always contiguous.
122+ let g2 = line
123+ .slice(line.len_chars().saturating_sub(2)..)
124+ .as_str()
125+ .unwrap_or("");
126+
127+ // First check the two-character case for CRLF, then check the single-character case.
128+ LineEnding::from_str(g2).or_else(|| LineEnding::from_str(g1))
129+}
130+
131+/// Returns the passed line's line ending, if any.
132+pub fn get_line_ending_of_str(line: &str) -> Option<LineEnding> {
133+ if line.ends_with("\u{000D}\u{000A}") {
134+ Some(LineEnding::Crlf)
135+ } else if line.ends_with('\u{000A}') {
136+ Some(LineEnding::LF)
137+ } else if line.ends_with('\u{000B}') {
138+ Some(LineEnding::VT)
139+ } else if line.ends_with('\u{000C}') {
140+ Some(LineEnding::FF)
141+ } else if line.ends_with('\u{000D}') {
142+ Some(LineEnding::CR)
143+ } else if line.ends_with('\u{0085}') {
144+ Some(LineEnding::Nel)
145+ } else if line.ends_with('\u{2028}') {
146+ Some(LineEnding::LS)
147+ } else if line.ends_with('\u{2029}') {
148+ Some(LineEnding::PS)
149+ } else {
150+ None
151+ }
152+}
153+
154+/// Returns the char index of the end of the given line, not including its line ending.
155+pub fn line_end_char_index(slice: &RopeSlice, line: usize) -> usize {
156+ slice.line_to_char(line + 1)
157+ - get_line_ending(&slice.line(line))
158+ .map(|le| le.len_chars())
159+ .unwrap_or(0)
160+}
161+
162+#[cfg(test)]
163+mod line_ending_tests {
164+ use super::*;
165+
166+ #[test]
167+ fn line_ending_autodetect() {
168+ assert_eq!(
169+ auto_detect_line_ending(&Rope::from_str("\n")),
170+ Some(LineEnding::LF)
171+ );
172+ assert_eq!(
173+ auto_detect_line_ending(&Rope::from_str("\r\n")),
174+ Some(LineEnding::Crlf)
175+ );
176+ assert_eq!(auto_detect_line_ending(&Rope::from_str("hello")), None);
177+ assert_eq!(auto_detect_line_ending(&Rope::from_str("")), None);
178+ assert_eq!(
179+ auto_detect_line_ending(&Rope::from_str("hello\nhelix\r\n")),
180+ Some(LineEnding::LF)
181+ );
182+ assert_eq!(
183+ auto_detect_line_ending(&Rope::from_str("a formfeed\u{000C}")),
184+ None
185+ );
186+ assert_eq!(
187+ auto_detect_line_ending(&Rope::from_str("\n\u{000A}\n \u{000A}")),
188+ Some(LineEnding::LF)
189+ );
190+ assert_eq!(
191+ auto_detect_line_ending(&Rope::from_str(
192+ "a formfeed\u{000C} with a\u{000C} linefeed\u{000A}"
193+ )),
194+ Some(LineEnding::LF)
195+ );
196+ assert_eq!(auto_detect_line_ending(&Rope::from_str("a formfeed\u{000C} with a\u{000C} carriage return linefeed\u{000D}\u{000A} and a linefeed\u{000A}")), Some(LineEnding::Crlf));
197+ }
198+
199+ #[test]
200+ fn str_to_line_ending() {
201+ assert_eq!(LineEnding::from_str("\r"), Some(LineEnding::CR));
202+ assert_eq!(LineEnding::from_str("\n"), Some(LineEnding::LF));
203+ assert_eq!(LineEnding::from_str("\r\n"), Some(LineEnding::Crlf));
204+ assert_eq!(LineEnding::from_str("hello\n"), None);
205+ }
206+
207+ #[test]
208+ fn rope_slice_to_line_ending() {
209+ let r = Rope::from_str("hello\r\n");
210+ assert_eq!(
211+ LineEnding::from_rope_slice(&r.slice(5..6)),
212+ Some(LineEnding::CR)
213+ );
214+ assert_eq!(
215+ LineEnding::from_rope_slice(&r.slice(6..7)),
216+ Some(LineEnding::LF)
217+ );
218+ assert_eq!(
219+ LineEnding::from_rope_slice(&r.slice(5..7)),
220+ Some(LineEnding::Crlf)
221+ );
222+ assert_eq!(LineEnding::from_rope_slice(&r.slice(..)), None);
223+ }
224+
225+ #[test]
226+ fn get_line_ending_rope_slice() {
227+ let r = Rope::from_str("Hello\rworld\nhow\r\nare you?");
228+ assert_eq!(get_line_ending(&r.slice(..6)), Some(LineEnding::CR));
229+ assert_eq!(get_line_ending(&r.slice(..12)), Some(LineEnding::LF));
230+ assert_eq!(get_line_ending(&r.slice(..17)), Some(LineEnding::Crlf));
231+ assert_eq!(get_line_ending(&r.slice(..)), None);
232+ }
233+
234+ #[test]
235+ fn get_line_ending_str() {
236+ let text = "Hello\rworld\nhow\r\nare you?";
237+ assert_eq!(get_line_ending_of_str(&text[..6]), Some(LineEnding::CR));
238+ assert_eq!(get_line_ending_of_str(&text[..12]), Some(LineEnding::LF));
239+ assert_eq!(get_line_ending_of_str(&text[..17]), Some(LineEnding::Crlf));
240+ assert_eq!(get_line_ending_of_str(&text[..]), None);
241+ }
242+
243+ #[test]
244+ fn line_end_char_index_rope_slice() {
245+ let r = Rope::from_str("Hello\rworld\nhow\r\nare you?");
246+ let s = &r.slice(..);
247+ assert_eq!(line_end_char_index(s, 0), 5);
248+ assert_eq!(line_end_char_index(s, 1), 11);
249+ assert_eq!(line_end_char_index(s, 2), 15);
250+ assert_eq!(line_end_char_index(s, 3), 25);
251+ }
252+}
helix-core/src/movement.rs+38−128
@@ -3,8 +3,13 @@ use std::iter::{self, from_fn, Peekable, SkipWhile};
33use ropey::iter::Chars;
44
55use crate::{
6+ chars::{
7+ categorize_char, char_is_line_ending, char_is_punctuation, char_is_whitespace,
8+ char_is_word, CharCategory,
9+ },
610 coords_at_pos,
711 graphemes::{nth_next_grapheme_boundary, nth_prev_grapheme_boundary},
12+ line_ending::{get_line_ending, line_end_char_index},
813 pos_at_coords, Position, Range, RopeSlice,
914};
1015
@@ -37,9 +42,8 @@ pub fn move_horizontally(
3742 nth_prev_grapheme_boundary(slice, pos, count).max(start)
3843 }
3944 Direction::Forward => {
40− // Line end is pos at the start of next line - 1
41− let end = slice.line_to_char(line + 1).saturating_sub(1);
42− nth_next_grapheme_boundary(slice, pos, count).min(end)
45+ let end_char_idx = line_end_char_index(&slice, line);
46+ nth_next_grapheme_boundary(slice, pos, count).min(end_char_idx)
4347 }
4448 };
4549 let anchor = match behaviour {
@@ -68,8 +72,11 @@ pub fn move_vertically(
6872 ),
6973 };
7074
71− // convert to 0-indexed, subtract another 1 because len_chars() counts \n
72− let new_line_len = slice.line(new_line).len_chars().saturating_sub(2);
75+ // Length of the line sans line-ending.
76+ let new_line_len = {
77+ let line = slice.line(new_line);
78+ line.len_chars() - get_line_ending(&line).map(|le| le.len_chars()).unwrap_or(0)
79+ };
7380
7481 let new_col = std::cmp::min(horiz as usize, new_line_len);
7582
@@ -104,64 +111,6 @@ fn word_move(slice: RopeSlice, mut range: Range, count: usize, target: WordMotio
104111}
105112
106113// ---- util ------------
107−#[inline]
108−pub(crate) fn is_word(ch: char) -> bool {
109− ch.is_alphanumeric() || ch == '_'
110−}
111−
112−#[inline]
113−pub(crate) fn is_end_of_line(ch: char) -> bool {
114− ch == '\n'
115−}
116−
117−#[inline]
118−// Whitespace, but not end of line
119−pub(crate) fn is_strict_whitespace(ch: char) -> bool {
120− ch.is_whitespace() && !is_end_of_line(ch)
121−}
122−
123−#[inline]
124−pub(crate) fn is_punctuation(ch: char) -> bool {
125− use unicode_general_category::{get_general_category, GeneralCategory};
126−
127− matches!(
128− get_general_category(ch),
129− GeneralCategory::OtherPunctuation
130− | GeneralCategory::OpenPunctuation
131− | GeneralCategory::ClosePunctuation
132− | GeneralCategory::InitialPunctuation
133− | GeneralCategory::FinalPunctuation
134− | GeneralCategory::ConnectorPunctuation
135− | GeneralCategory::DashPunctuation
136− | GeneralCategory::MathSymbol
137− | GeneralCategory::CurrencySymbol
138− | GeneralCategory::ModifierSymbol
139− )
140−}
141−
142−#[derive(Debug, Eq, PartialEq)]
143−pub enum Category {
144− Whitespace,
145− Eol,
146− Word,
147− Punctuation,
148− Unknown,
149−}
150−
151−#[inline]
152−pub(crate) fn categorize(ch: char) -> Category {
153− if is_end_of_line(ch) {
154− Category::Eol
155− } else if ch.is_whitespace() {
156− Category::Whitespace
157− } else if is_word(ch) {
158− Category::Word
159− } else if is_punctuation(ch) {
160− Category::Punctuation
161− } else {
162− Category::Unknown
163− }
164−}
165114
166115#[inline]
167116/// Returns first index that doesn't satisfy a given predicate when
@@ -235,7 +184,8 @@ impl CharHelpers for Chars<'_> {
235184 let mut phase = WordMotionPhase::Start;
236185 let mut head = origin.head;
237186 let mut anchor: Option<usize> = None;
238− let is_boundary = |a: char, b: Option<char>| categorize(a) != categorize(b.unwrap_or(a));
187+ let is_boundary =
188+ |a: char, b: Option<char>| categorize_char(a) != categorize_char(b.unwrap_or(a));
239189 while let Some(peek) = characters.peek().copied() {
240190 phase = match phase {
241191 WordMotionPhase::Start => {
@@ -244,7 +194,8 @@ impl CharHelpers for Chars<'_> {
244194 break; // We're at the end, so there's nothing to do.
245195 }
246196 // Anchor may remain here if the head wasn't at a boundary
247− if !is_boundary(peek, characters.peek().copied()) && !is_end_of_line(peek) {
197+ if !is_boundary(peek, characters.peek().copied()) && !char_is_line_ending(peek)
198+ {
248199 anchor = Some(head);
249200 }
250201 // First character is always skipped by the head
@@ -252,7 +203,7 @@ impl CharHelpers for Chars<'_> {
252203 WordMotionPhase::SkipNewlines
253204 }
254205 WordMotionPhase::SkipNewlines => {
255− if is_end_of_line(peek) {
206+ if char_is_line_ending(peek) {
256207 characters.next();
257208 if characters.peek().is_some() {
258209 advance(&mut head);
@@ -286,12 +237,12 @@ fn reached_target(target: WordMotionTarget, peek: char, next_peek: Option<&char>
286237
287238 match target {
288239 WordMotionTarget::NextWordStart => {
289− ((categorize(peek) != categorize(*next_peek))
290− && (is_end_of_line(*next_peek) || !next_peek.is_whitespace()))
240+ ((categorize_char(peek) != categorize_char(*next_peek))
241+ && (char_is_line_ending(*next_peek) || !next_peek.is_whitespace()))
291242 }
292243 WordMotionTarget::NextWordEnd | WordMotionTarget::PrevWordStart => {
293− ((categorize(peek) != categorize(*next_peek))
294− && (!peek.is_whitespace() || is_end_of_line(*next_peek)))
244+ ((categorize_char(peek) != categorize_char(*next_peek))
245+ && (!peek.is_whitespace() || char_is_line_ending(*next_peek)))
295246 }
296247 }
297248}
@@ -330,7 +281,7 @@ mod test {
330281 slice,
331282 move_vertically(slice, range, Direction::Forward, 1, Movement::Move).head
332283 ),
333− (1, 2).into()
284+ (1, 3).into()
334285 );
335286 }
336287
@@ -343,12 +294,12 @@ mod test {
343294 let mut range = Range::point(position);
344295
345296 let moves_and_expected_coordinates = [
346− ((Direction::Forward, 1usize), (0, 1)),
347− ((Direction::Forward, 2usize), (0, 3)),
348− ((Direction::Forward, 0usize), (0, 3)),
349− ((Direction::Forward, 999usize), (0, 31)),
350− ((Direction::Forward, 999usize), (0, 31)),
351− ((Direction::Backward, 999usize), (0, 0)),
297+ ((Direction::Forward, 1usize), (0, 1)), // T|his is a simple alphabetic line
298+ ((Direction::Forward, 2usize), (0, 3)), // Thi|s is a simple alphabetic line
299+ ((Direction::Forward, 0usize), (0, 3)), // Thi|s is a simple alphabetic line
300+ ((Direction::Forward, 999usize), (0, 32)), // This is a simple alphabetic line|
301+ ((Direction::Forward, 999usize), (0, 32)), // This is a simple alphabetic line|
302+ ((Direction::Backward, 999usize), (0, 0)), // |This is a simple alphabetic line
352303 ];
353304
354305 for ((direction, amount), coordinates) in IntoIter::new(moves_and_expected_coordinates) {
@@ -366,15 +317,15 @@ mod test {
366317 let mut range = Range::point(position);
367318
368319 let moves_and_expected_coordinates = IntoIter::new([
369− ((Direction::Forward, 1usize), (0, 1)), // M_ltiline
370− ((Direction::Forward, 2usize), (0, 3)), // Mul_iline
371− ((Direction::Backward, 6usize), (0, 0)), // _ultiline
372− ((Direction::Backward, 999usize), (0, 0)), // _ultiline
373− ((Direction::Forward, 3usize), (0, 3)), // Mul_iline
374− ((Direction::Forward, 0usize), (0, 3)), // Mul_iline
375− ((Direction::Backward, 0usize), (0, 3)), // Mul_iline
376− ((Direction::Forward, 999usize), (0, 9)), // Multilin_
377− ((Direction::Forward, 999usize), (0, 9)), // Multilin_
320+ ((Direction::Forward, 1usize), (0, 1)), // M|ultiline\n
321+ ((Direction::Forward, 2usize), (0, 3)), // Mul|tiline\n
322+ ((Direction::Backward, 6usize), (0, 0)), // |Multiline\n
323+ ((Direction::Backward, 999usize), (0, 0)), // |Multiline\n
324+ ((Direction::Forward, 3usize), (0, 3)), // Mul|tiline\n
325+ ((Direction::Forward, 0usize), (0, 3)), // Mul|tiline\n
326+ ((Direction::Backward, 0usize), (0, 3)), // Mul|tiline\n
327+ ((Direction::Forward, 999usize), (0, 9)), // Multiline|\n
328+ ((Direction::Forward, 999usize), (0, 9)), // Multiline|\n
378329 ]);
379330
380331 for ((direction, amount), coordinates) in moves_and_expected_coordinates {
@@ -446,7 +397,7 @@ mod test {
446397 // First descent preserves column as the target line is wider
447398 ((Axis::V, Direction::Forward, 1usize), (1, 8)),
448399 // Second descent clamps column as the target line is shorter
449− ((Axis::V, Direction::Forward, 1usize), (2, 4)),
400+ ((Axis::V, Direction::Forward, 1usize), (2, 5)),
450401 // Third descent restores the original column
451402 ((Axis::V, Direction::Forward, 1usize), (3, 8)),
452403 // Behaviour is preserved even through long jumps
@@ -760,45 +711,4 @@ mod test {
760711 }
761712 }
762713 }
763−
764− #[test]
765− fn test_categorize() {
766− const WORD_TEST_CASE: &'static str =
767− "_hello_world_あいうえおー1234567890１２３４５６７８９０";
768− const PUNCTUATION_TEST_CASE: &'static str =
769− "!\"#$%&\'()*+,-./:;<=>?@[\\]^`{|}~！”＃＄％＆’（）＊＋、。：；＜＝＞？＠「」＾｀｛｜｝～";
770− const WHITESPACE_TEST_CASE: &'static str = "  　   ";
771−
772− assert_eq!(Category::Eol, categorize('\n'));
773−
774− for ch in WHITESPACE_TEST_CASE.chars() {
775− assert_eq!(
776− Category::Whitespace,
777− categorize(ch),
778− "Testing '{}', but got `{:?}` instead of `Category::Whitespace`",
779− ch,
780− categorize(ch)
781− );
782− }
783−
784− for ch in WORD_TEST_CASE.chars() {
785− assert_eq!(
786− Category::Word,
787− categorize(ch),
788− "Testing '{}', but got `{:?}` instead of `Category::Word`",
789− ch,
790− categorize(ch)
791− );
792− }
793−
794− for ch in PUNCTUATION_TEST_CASE.chars() {
795− assert_eq!(
796− Category::Punctuation,
797− categorize(ch),
798− "Testing '{}', but got `{:?}` instead of `Category::Punctuation`",
799− ch,
800− categorize(ch)
801− );
802− }
803− }
804714}
helix-core/src/position.rs+4−2
@@ -1,4 +1,5 @@
11use crate::{
2+ chars::char_is_line_ending,
23 graphemes::{nth_next_grapheme_boundary, RopeGraphemes},
34 Rope, RopeSlice,
45};
@@ -23,8 +24,9 @@ impl Position {
2324 pub fn traverse(self, text: &crate::Tendril) -> Self {
2425 let Self { mut row, mut col } = self;
2526 // TODO: there should be a better way here
26− for ch in text.chars() {
27− if ch == '\n' {
27+ let mut chars = text.chars().peekable();
28+ while let Some(ch) = chars.next() {
29+ if char_is_line_ending(ch) && !(ch == '\r' && chars.peek() == Some(&'\n')) {
2830 row += 1;
2931 col = 0;
3032 } else {
helix-core/src/syntax.rs+5−4
@@ -1,4 +1,4 @@
1−use crate::{regex::Regex, Change, Rope, RopeSlice, Transaction};
1+use crate::{chars::char_is_line_ending, regex::Regex, Change, Rope, RopeSlice, Transaction};
22pub use helix_syntax::{get_language, get_language_name, Lang};
33
44use arc_swap::ArcSwap;
@@ -589,9 +589,10 @@ impl LanguageLayer {
589589 mut column,
590590 } = point;
591591
592− // TODO: there should be a better way here
593− for ch in text.bytes() {
594− if ch == b'\n' {
592+ // TODO: there should be a better way here.
593+ let mut chars = text.chars().peekable();
594+ while let Some(ch) = chars.next() {
595+ if char_is_line_ending(ch) && !(ch == '\r' && chars.peek() == Some(&'\n')) {
595596 row += 1;
596597 column = 0;
597598 } else {
helix-lsp/src/client.rs+4−3
@@ -3,7 +3,7 @@ use crate::{
33 Call, Error, OffsetEncoding, Result,
44};
55
6−use helix_core::{find_root, ChangeSet, Rope};
6+use helix_core::{chars::char_is_line_ending, find_root, ChangeSet, Rope};
77use jsonrpc_core as jsonrpc;
88use lsp_types as lsp;
99use serde_json::Value;
@@ -337,8 +337,9 @@ impl Client {
337337 mut character,
338338 } = pos;
339339
340− for ch in text.chars() {
341− if ch == '\n' {
340+ let mut chars = text.chars().peekable();
341+ while let Some(ch) = chars.next() {
342+ if char_is_line_ending(ch) && !(ch == '\r' && chars.peek() == Some(&'\n')) {
342343 line += 1;
343344 character = 0;
344345 } else {
helix-term/src/commands.rs+111−30
@@ -1,12 +1,15 @@
11use helix_core::{
22 comment, coords_at_pos, find_first_non_whitespace_char, find_root, graphemes, indent,
3+ line_ending::{
4+ get_line_ending, get_line_ending_of_str, line_end_char_index, str_is_line_ending,
5+ },
36 match_brackets,
47 movement::{self, Direction},
58 object, pos_at_coords,
69 regex::{self, Regex},
710 register::{self, Register, Registers},
8− search, selection, Change, ChangeSet, Position, Range, Rope, RopeSlice, Selection, SmallVec,
9− Tendril, Transaction,
11+ search, selection, Change, ChangeSet, LineEnding, Position, Range, Rope, RopeGraphemes,
12+ RopeSlice, Selection, SmallVec, Tendril, Transaction, DEFAULT_LINE_ENDING,
1013};
1114
1215use helix_view::{
@@ -303,9 +306,8 @@ fn move_line_end(cx: &mut Context) {
303306 let text = doc.text();
304307 let line = text.char_to_line(range.head);
305308
306− // Line end is pos at the start of next line - 1
307− // subtract another 1 because the line ends with \n
308− let pos = text.line_to_char(line + 1).saturating_sub(2);
309+ let pos = line_end_char_index(&text.slice(..), line);
310+
309311 Range::new(pos, pos)
310312 });
311313
@@ -452,12 +454,28 @@ where
452454 let count = cx.count();
453455
454456 // need to wait for next key
457+ // TODO: should this be done by grapheme rather than char? For example,
458+ // we can't properly handle the line-ending CRLF case here in terms of char.
455459 cx.on_next_key(move |cx, event| {
456460 let ch = match event {
457461 KeyEvent {
458462 code: KeyCode::Enter,
459463 ..
460− } => '\n',
464+ } =>
465+ // TODO: this isn't quite correct when CRLF is involved.
466+ // This hack will work in most cases, since documents don't
467+ // usually mix line endings. But we should fix it eventually
468+ // anyway.
469+ {
470+ current!(cx.editor)
471+ .1
472+ .line_ending
473+ .as_str()
474+ .chars()
475+ .next()
476+ .unwrap()
477+ }
478+
461479 KeyEvent {
462480 code: KeyCode::Char(ch),
463481 ..
@@ -575,32 +593,37 @@ fn extend_first_nonwhitespace(cx: &mut Context) {
575593}
576594
577595fn replace(cx: &mut Context) {
596+ let mut buf = [0u8; 4]; // To hold utf8 encoded char.
597+
578598 // need to wait for next key
579599 cx.on_next_key(move |cx, event| {
600+ let (view, doc) = current!(cx.editor);
580601 let ch = match event {
581602 KeyEvent {
582603 code: KeyCode::Char(ch),
583604 ..
584− } => Some(ch),
605+ } => Some(&ch.encode_utf8(&mut buf[..])[..]),
585606 KeyEvent {
586607 code: KeyCode::Enter,
587608 ..
588− } => Some('\n'),
609+ } => Some(doc.line_ending.as_str()),
589610 _ => None,
590611 };
591612
592613 if let Some(ch) = ch {
593− let (view, doc) = current!(cx.editor);
594−
595614 let transaction =
596615 Transaction::change_by_selection(doc.text(), doc.selection(view.id), |range| {
597616 let max_to = doc.text().len_chars().saturating_sub(1);
598617 let to = std::cmp::min(max_to, range.to() + 1);
599− let text: String = doc
600− .text()
601− .slice(range.from()..to)
602− .chars()
603− .map(|c| if c == '\n' { '\n' } else { ch })
618+ let text: String = RopeGraphemes::new(doc.text().slice(range.from()..to))
619+ .map(|g| {
620+ let cow: Cow<str> = g.into();
621+ if str_is_line_ending(&cow) {
622+ cow
623+ } else {
624+ ch.into()
625+ }
626+ })
604627 .collect();
605628
606629 (range.from(), to, Some(text.into()))
@@ -725,9 +748,8 @@ fn extend_line_end(cx: &mut Context) {
725748 let text = doc.text();
726749 let line = text.char_to_line(range.head);
727750
728− // Line end is pos at the start of next line - 1
729− // subtract another 1 because the line ends with \n
730− let pos = text.line_to_char(line + 1).saturating_sub(2);
751+ let pos = line_end_char_index(&text.slice(..), line);
752+
731753 Range::new(range.anchor, pos)
732754 });
733755
@@ -783,7 +805,8 @@ fn split_selection_on_newline(cx: &mut Context) {
783805 let text = doc.text().slice(..);
784806 // only compile the regex once
785807 #[allow(clippy::trivial_regex)]
786− static REGEX: Lazy<Regex> = Lazy::new(|| Regex::new(r"\n").unwrap());
808+ static REGEX: Lazy<Regex> =
809+ Lazy::new(|| Regex::new(r"\r\n|[\n\r\u{000B}\u{000C}\u{0085}\u{2028}\u{2029}]").unwrap());
787810 let selection = selection::split_on_matches(text, doc.selection(view.id), &REGEX);
788811 doc.set_selection(view.id, selection);
789812}
@@ -922,7 +945,13 @@ fn delete_selection_impl(reg: &mut Register, doc: &mut Document, view_id: ViewId
922945 // then delete
923946 let transaction =
924947 Transaction::change_by_selection(doc.text(), doc.selection(view_id), |range| {
925− let max_to = doc.text().len_chars().saturating_sub(1);
948+ let alltext = doc.text();
949+ let line = alltext.char_to_line(range.head);
950+ let max_to = doc.text().len_chars().saturating_sub(
951+ get_line_ending(&alltext.line(line))
952+ .map(|le| le.len_chars())
953+ .unwrap_or(0),
954+ );
926955 let to = std::cmp::min(max_to, range.to() + 1);
927956 (range.from(), to, None)
928957 });
@@ -1003,7 +1032,7 @@ fn append_mode(cx: &mut Context) {
10031032 if selection.iter().any(|range| range.head == end) {
10041033 let transaction = Transaction::change(
10051034 doc.text(),
1006− std::array::IntoIter::new([(end, end, Some(Tendril::from_char('\n')))]),
1035+ std::array::IntoIter::new([(end, end, Some(doc.line_ending.as_str().into()))]),
10071036 );
10081037 doc.apply(&transaction, view.id);
10091038 }
@@ -1131,6 +1160,45 @@ mod cmd {
11311160 }
11321161 }
11331162
1163+ /// Sets or reports the current document's line ending setting.
1164+ fn set_line_ending(editor: &mut Editor, args: &[&str], event: PromptEvent) {
1165+ use LineEnding::*;
1166+
1167+ // If no argument, report current line ending setting.
1168+ if args.is_empty() {
1169+ let line_ending = current!(editor).1.line_ending;
1170+ editor.set_status(match line_ending {
1171+ Crlf => "crlf".into(),
1172+ LF => "line feed".into(),
1173+ FF => "form feed".into(),
1174+ CR => "carriage return".into(),
1175+ Nel => "next line".into(),
1176+
1177+ // These should never be a document's default line ending.
1178+ VT | LS | PS => "error".into(),
1179+ });
1180+ return;
1181+ }
1182+
1183+ // Attempt to parse argument as a line ending.
1184+ let line_ending = match args.get(0) {
1185+ // We check for CR first because it shares a common prefix with CRLF.
1186+ Some(arg) if "cr".starts_with(&arg.to_lowercase()) => Some(CR),
1187+ Some(arg) if "crlf".starts_with(&arg.to_lowercase()) => Some(Crlf),
1188+ Some(arg) if "lf".starts_with(&arg.to_lowercase()) => Some(LF),
1189+ Some(arg) if "ff".starts_with(&arg.to_lowercase()) => Some(FF),
1190+ Some(arg) if "nel".starts_with(&arg.to_lowercase()) => Some(Nel),
1191+ _ => None,
1192+ };
1193+
1194+ if let Some(le) = line_ending {
1195+ doc_mut!(editor).line_ending = le;
1196+ } else {
1197+ // Invalid argument.
1198+ editor.set_error(format!("invalid line ending '{}'", args[0],));
1199+ }
1200+ }
1201+
11341202 fn earlier(editor: &mut Editor, args: &[&str], event: PromptEvent) {
11351203 let uk = match args.join(" ").parse::<helix_core::history::UndoKind>() {
11361204 Ok(uk) => uk,
@@ -1274,7 +1342,11 @@ mod cmd {
12741342 }
12751343
12761344 fn yank_joined_to_clipboard(editor: &mut Editor, args: &[&str], _: PromptEvent) {
1277− let separator = args.first().copied().unwrap_or("\n");
1345+ let (_, doc) = current!(editor);
1346+ let separator = args
1347+ .first()
1348+ .copied()
1349+ .unwrap_or_else(|| doc.line_ending.as_str());
12781350 yank_joined_to_clipboard_impl(editor, separator);
12791351 }
12801352
@@ -1359,6 +1431,13 @@ mod cmd {
13591431 fun: set_indent_style,
13601432 completer: None,
13611433 },
1434+ TypableCommand {
1435+ name: "line-ending",
1436+ alias: None,
1437+ doc: "Set the document's default line ending. Options: crlf, lf, cr, ff, nel.",
1438+ fun: set_line_ending,
1439+ completer: None,
1440+ },
13621441 TypableCommand {
13631442 name: "earlier",
13641443 alias: Some("ear"),
@@ -1683,8 +1762,7 @@ fn append_to_line(cx: &mut Context) {
16831762 let selection = doc.selection(view.id).transform(|range| {
16841763 let text = doc.text();
16851764 let line = text.char_to_line(range.head);
1686− // we can't use line_to_char(line + 1) - 2 because the last line might not contain \n
1687− let pos = (text.line_to_char(line) + text.line(line).len_chars()).saturating_sub(1);
1765+ let pos = line_end_char_index(&text.slice(..), line);
16881766 Range::new(pos, pos)
16891767 });
16901768 doc.set_selection(view.id, selection);
@@ -1731,7 +1809,7 @@ fn open(cx: &mut Context, open: Open) {
17311809 let indent = doc.indent_unit().repeat(indent_level);
17321810 let indent_len = indent.len();
17331811 let mut text = String::with_capacity(1 + indent_len);
1734− text.push('\n');
1812+ text.push_str(doc.line_ending.as_str());
17351813 text.push_str(&indent);
17361814 let text = text.repeat(count);
17371815
@@ -2344,7 +2422,7 @@ pub mod insert {
23442422 );
23452423 let indent = doc.indent_unit().repeat(indent_level);
23462424 let mut text = String::with_capacity(1 + indent.len());
2347− text.push('\n');
2425+ text.push_str(doc.line_ending.as_str());
23482426 text.push_str(&indent);
23492427
23502428 let head = pos + offs + text.chars().count();
@@ -2365,7 +2443,7 @@ pub mod insert {
23652443 if helix_core::auto_pairs::PAIRS.contains(&(prev, curr)) {
23662444 // another newline, indent the end bracket one level less
23672445 let indent = doc.indent_unit().repeat(indent_level.saturating_sub(1));
2368− text.push('\n');
2446+ text.push_str(doc.line_ending.as_str());
23692447 text.push_str(&indent);
23702448 }
23712449
@@ -2488,7 +2566,8 @@ fn yank_joined_to_clipboard_impl(editor: &mut Editor, separator: &str) {
24882566}
24892567
24902568fn yank_joined_to_clipboard(cx: &mut Context) {
2491− yank_joined_to_clipboard_impl(&mut cx.editor, "\n");
2569+ let line_ending = current!(cx.editor).1.line_ending;
2570+ yank_joined_to_clipboard_impl(&mut cx.editor, line_ending.as_str());
24922571}
24932572
24942573fn yank_main_selection_to_clipboard_impl(editor: &mut Editor) {
@@ -2529,8 +2608,10 @@ fn paste_impl(
25292608 .unwrap(),
25302609 );
25312610
2532− // if any of values ends \n it's linewise paste
2533− let linewise = values.iter().any(|value| value.ends_with('\n'));
2611+ // if any of values ends with a line ending, it's linewise paste
2612+ let linewise = values
2613+ .iter()
2614+ .any(|value| get_line_ending_of_str(value).is_some());
25342615
25352616 let mut values = values.iter().cloned().map(Tendril::from).chain(repeat);
25362617
helix-term/src/ui/editor.rs+2−2
@@ -9,7 +9,7 @@ use crate::{
99use helix_core::{
1010 coords_at_pos,
1111 syntax::{self, HighlightEvent},
12− Position, Range,
12+ LineEnding, Position, Range,
1313};
1414use helix_lsp::LspProgressMap;
1515use helix_view::{document::Mode, Document, Editor, Theme, View};
@@ -184,7 +184,7 @@ impl EditorView {
184184
185185 // iterate over range char by char
186186 for grapheme in RopeGraphemes::new(text) {
187− if grapheme == "\n" {
187+ if LineEnding::from_rope_slice(&grapheme).is_some() {
188188 visual_x = 0;
189189 line += 1;
190190
helix-term/src/ui/markdown.rs+2
@@ -115,6 +115,8 @@ fn parse<'a>(
115115 // TODO: replace tabs with indentation
116116
117117 let mut slice = &text[start..end];
118+ // TODO: do we need to handle all unicode line endings
119+ // here, or is just '\n' okay?
118120 while let Some(end) = slice.find('\n') {
119121 // emit span up to newline
120122 let text = &slice[..end];
helix-tui/Cargo.toml+1
@@ -22,3 +22,4 @@ unicode-segmentation = "1.2"
2222unicode-width = "0.1"
2323crossterm = { version = "0.20", optional = true }
2424serde = { version = "1", "optional" = true, features = ["derive"]}
25+helix-core = { version = "0.2", path = "../helix-core" }
helix-tui/src/text.rs+2−1
@@ -47,6 +47,7 @@
4747//! ]);
4848//! ```
4949use crate::style::Style;
50+use helix_core::line_ending::str_is_line_ending;
5051use std::borrow::Cow;
5152use unicode_segmentation::UnicodeSegmentation;
5253use unicode_width::UnicodeWidthStr;
@@ -177,7 +178,7 @@ impl<'a> Span<'a> {
177178 symbol: g,
178179 style: base_style.patch(self.style),
179180 })
180− .filter(|s| s.symbol != "\n")
181+ .filter(|s| !str_is_line_ending(s.symbol))
181182 }
182183}
183184
helix-tui/src/widgets/reflow.rs+5−4
@@ -1,4 +1,5 @@
11use crate::text::StyledGrapheme;
2+use helix_core::line_ending::str_is_line_ending;
23use unicode_segmentation::UnicodeSegmentation;
34use unicode_width::UnicodeWidthStr;
45
@@ -62,13 +63,13 @@ impl<'a, 'b> LineComposer<'a> for WordWrapper<'a, 'b> {
6263 // Ignore characters wider that the total max width.
6364 if symbol.width() as u16 > self.max_line_width
6465 // Skip leading whitespace when trim is enabled.
65− || self.trim && symbol_whitespace && symbol != "\n" && current_line_width == 0
66+ || self.trim && symbol_whitespace && !str_is_line_ending(symbol) && current_line_width == 0
6667 {
6768 continue;
6869 }
6970
7071 // Break on newline and discard it.
71− if symbol == "\n" {
72+ if str_is_line_ending(symbol) {
7273 if prev_whitespace {
7374 current_line_width = width_to_last_word_end;
7475 self.current_line.truncate(symbols_to_last_word_end);
@@ -170,7 +171,7 @@ impl<'a, 'b> LineComposer<'a> for LineTruncator<'a, 'b> {
170171 }
171172
172173 // Break on newline and discard it.
173− if symbol == "\n" {
174+ if str_is_line_ending(symbol) {
174175 break;
175176 }
176177
@@ -199,7 +200,7 @@ impl<'a, 'b> LineComposer<'a> for LineTruncator<'a, 'b> {
199200
200201 if skip_rest {
201202 for StyledGrapheme { symbol, .. } in &mut self.symbols {
202− if symbol == "\n" {
203+ if str_is_line_ending(symbol) {
203204 break;
204205 }
205206 }
helix-view/src/document.rs+22−12
@@ -7,10 +7,12 @@ use std::str::FromStr;
77use std::sync::Arc;
88
99use helix_core::{
10− chars::{char_is_linebreak, char_is_whitespace},
10+ chars::{char_is_line_ending, char_is_whitespace},
1111 history::History,
12+ line_ending::auto_detect_line_ending,
1213 syntax::{self, LanguageConfiguration},
13− ChangeSet, Diagnostic, Rope, Selection, State, Syntax, Transaction,
14+ ChangeSet, Diagnostic, LineEnding, Rope, Selection, State, Syntax, Transaction,
15+ DEFAULT_LINE_ENDING,
1416};
1517
1618use crate::{DocumentId, Theme, ViewId};
@@ -45,6 +47,9 @@ pub struct Document {
4547 /// Current indent style.
4648 pub indent_style: IndentStyle,
4749
50+ /// The document's default line ending.
51+ pub line_ending: LineEnding,
52+
4853 syntax: Option<Syntax>,
4954 // /// Corresponding language scope name. Usually `source.<lang>`.
5055 pub(crate) language: Option<Arc<LanguageConfiguration>>,
@@ -232,6 +237,7 @@ impl Document {
232237 history: Cell::new(History::default()),
233238 last_saved_revision: 0,
234239 language_server: None,
240+ line_ending: DEFAULT_LINE_ENDING,
235241 }
236242 }
237243
@@ -243,22 +249,26 @@ impl Document {
243249 ) -> Result<Self, Error> {
244250 use std::{fs::File, io::BufReader};
245251
246− let doc = if !path.exists() {
247− Rope::from("\n")
252+ let mut doc = if !path.exists() {
253+ Rope::from(DEFAULT_LINE_ENDING.as_str())
248254 } else {
249255 let file = File::open(&path).context(format!("unable to open {:?}", path))?;
250− let mut doc = Rope::from_reader(BufReader::new(file))?;
251− // add missing newline at the end of file
252− if doc.len_bytes() == 0 || doc.byte(doc.len_bytes() - 1) != b'\n' {
253− doc.insert_char(doc.len_chars(), '\n');
254− }
255− doc
256+ Rope::from_reader(BufReader::new(file))?
256257 };
257258
259+ // search for line endings
260+ let line_ending = auto_detect_line_ending(&doc).unwrap_or(DEFAULT_LINE_ENDING);
261+
262+ // add missing newline at the end of file
263+ if doc.len_bytes() == 0 || char_is_line_ending(doc.char(doc.len_chars() - 1)) {
264+ doc.insert(doc.len_chars(), line_ending.as_str());
265+ }
266+
258267 let mut doc = Self::new(doc);
259268 // set the path and try detecting the language
260269 doc.set_path(&path)?;
261270 doc.detect_indent_style();
271+ doc.line_ending = line_ending;
262272
263273 if let Some(loader) = config_loader {
264274 doc.detect_language(theme, loader);
@@ -366,7 +376,7 @@ impl Document {
366376 Some(' ') => false,
367377
368378 // Ignore blank lines.
369− Some(c) if char_is_linebreak(c) => continue,
379+ Some(c) if char_is_line_ending(c) => continue,
370380
371381 _ => {
372382 prev_line_is_tabs = false;
@@ -390,7 +400,7 @@ impl Document {
390400 c if char_is_whitespace(c) => count_is_done = true,
391401
392402 // Ignore blank lines.
393− c if char_is_linebreak(c) => continue 'outer,
403+ c if char_is_line_ending(c) => continue 'outer,
394404
395405 _ => break,
396406 }
helix-view/src/editor.rs+2−3
@@ -15,10 +15,9 @@ use slotmap::SlotMap;
1515
1616use anyhow::Error;
1717
18−use helix_core::Position;
19−
2018pub use helix_core::diagnostic::Severity;
2119pub use helix_core::register::Registers;
20+use helix_core::{Position, DEFAULT_LINE_ENDING};
2221
2322#[derive(Debug)]
2423pub struct Editor {
@@ -173,7 +172,7 @@ impl Editor {
173172
174173 pub fn new_file(&mut self, action: Action) -> DocumentId {
175174 use helix_core::Rope;
176− let doc = Document::new(Rope::from("\n"));
175+ let doc = Document::new(Rope::from(DEFAULT_LINE_ENDING.as_str()));
177176 let id = self.documents.insert(doc);
178177 self.documents[id].id = id;
179178 self.switch(id, action);