@@ -1,6 +1,9 @@
1+use std::borrow::Cow;
2+
13use crate::{
24 chars::char_is_line_ending,
3− graphemes::{nth_next_grapheme_boundary, RopeGraphemes},
5+ graphemes::{ensure_grapheme_boundary_prev, grapheme_width, RopeGraphemes},
6+ line_ending::line_end_char_index,
47 RopeSlice,
58};
69
@@ -52,23 +55,58 @@ impl From<Position> for tree_sitter::Point {
5255 }
5356}
5457/// Convert a character index to (line, column) coordinates.
58+///
59+/// TODO: this should be split into two methods: one for visual
60+/// row/column, and one for "objective" row/column (possibly with
61+/// the column specified in `char`s). The former would be used
62+/// for cursor movement, and the latter would be used for e.g. the
63+/// row:column display in the status line.
5564pub fn coords_at_pos(text: RopeSlice, pos: usize) -> Position {
56− // TODO: this isn't correct. This needs to work in terms of
57− // visual horizontal position, not graphemes.
5865 let line = text.char_to_line(pos);
66+
5967 let line_start = text.line_to_char(line);
60− let col = RopeGraphemes::new(text.slice(line_start..pos)).count();
68+ let pos = ensure_grapheme_boundary_prev(text, pos);
69+ let col = RopeGraphemes::new(text.slice(line_start..pos))
70+ .map(|g| {
71+ let g: Cow<str> = g.into();
72+ grapheme_width(&g)
73+ })
74+ .sum();
75+
6176 Position::new(line, col)
6277}
6378
6479/// Convert (line, column) coordinates to a character index.
65−pub fn pos_at_coords(text: RopeSlice, coords: Position) -> usize {
66− // TODO: this isn't correct. This needs to work in terms of
67− // visual horizontal position, not graphemes.
80+///
81+/// `is_1_width` specifies whether the position should be treated
82+/// as a block cursor or not. This effects how line-ends are handled.
83+/// `false` corresponds to properly round-tripping with `coords_at_pos()`,
84+/// whereas `true` will ensure that block cursors don't jump off the
85+/// end of the line.
86+pub fn pos_at_coords(text: RopeSlice, coords: Position, is_1_width: bool) -> usize {
6887 let Position { row, col } = coords;
6988 let line_start = text.line_to_char(row);
70− // line_start + col
71− nth_next_grapheme_boundary(text, line_start, col)
89+ let line_end = if is_1_width {
90+ line_end_char_index(&text, row)
91+ } else {
92+ text.line_to_char((row + 1).min(text.len_lines()))
93+ };
94+
95+ let mut prev_col = 0;
96+ let mut col_char_offset = 0;
97+ for g in RopeGraphemes::new(text.slice(line_start..line_end)) {
98+ let g: Cow<str> = g.into();
99+ let next_col = prev_col + grapheme_width(&g);
100+
101+ if next_col > col {
102+ break;
103+ }
104+
105+ prev_col = next_col;
106+ col_char_offset += g.chars().count();
107+ }
108+
109+ line_start + col_char_offset
72110}
73111
74112#[cfg(test)]
@@ -84,13 +122,24 @@ mod test {
84122
85123 #[test]
86124 fn test_coords_at_pos() {
87− // let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
88− // let slice = text.slice(..);
89− // assert_eq!(coords_at_pos(slice, 0), (0, 0).into());
90− // assert_eq!(coords_at_pos(slice, 5), (0, 5).into()); // position on \n
91− // assert_eq!(coords_at_pos(slice, 6), (1, 0).into()); // position on w
92− // assert_eq!(coords_at_pos(slice, 7), (1, 1).into()); // position on o
93− // assert_eq!(coords_at_pos(slice, 10), (1, 4).into()); // position on d
125+ let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
126+ let slice = text.slice(..);
127+ assert_eq!(coords_at_pos(slice, 0), (0, 0).into());
128+ assert_eq!(coords_at_pos(slice, 5), (0, 5).into()); // position on \n
129+ assert_eq!(coords_at_pos(slice, 6), (1, 0).into()); // position on w
130+ assert_eq!(coords_at_pos(slice, 7), (1, 1).into()); // position on o
131+ assert_eq!(coords_at_pos(slice, 10), (1, 4).into()); // position on d
132+
133+ // Test with wide characters.
134+ let text = Rope::from("今日はいい\n");
135+ let slice = text.slice(..);
136+ assert_eq!(coords_at_pos(slice, 0), (0, 0).into());
137+ assert_eq!(coords_at_pos(slice, 1), (0, 2).into());
138+ assert_eq!(coords_at_pos(slice, 2), (0, 4).into());
139+ assert_eq!(coords_at_pos(slice, 3), (0, 6).into());
140+ assert_eq!(coords_at_pos(slice, 4), (0, 8).into());
141+ assert_eq!(coords_at_pos(slice, 5), (0, 10).into());
142+ assert_eq!(coords_at_pos(slice, 6), (1, 0).into());
94143
95144 // test with grapheme clusters
96145 let text = Rope::from("a̐éö̲\r\n");
@@ -99,40 +148,64 @@ mod test {
99148 assert_eq!(coords_at_pos(slice, 2), (0, 1).into());
100149 assert_eq!(coords_at_pos(slice, 4), (0, 2).into());
101150 assert_eq!(coords_at_pos(slice, 7), (0, 3).into());
151+ assert_eq!(coords_at_pos(slice, 9), (1, 0).into());
102152
103− let text = Rope::from("किमपि");
153+ let text = Rope::from("किमपि\n");
104154 let slice = text.slice(..);
105155 assert_eq!(coords_at_pos(slice, 0), (0, 0).into());
106− assert_eq!(coords_at_pos(slice, 2), (0, 1).into());
107− assert_eq!(coords_at_pos(slice, 3), (0, 2).into());
108− assert_eq!(coords_at_pos(slice, 5), (0, 3).into());
156+ assert_eq!(coords_at_pos(slice, 2), (0, 2).into());
157+ assert_eq!(coords_at_pos(slice, 3), (0, 3).into());
158+ assert_eq!(coords_at_pos(slice, 5), (0, 5).into());
159+ assert_eq!(coords_at_pos(slice, 6), (1, 0).into());
109160 }
110161
111162 #[test]
112163 fn test_pos_at_coords() {
113164 let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
114165 let slice = text.slice(..);
115− assert_eq!(pos_at_coords(slice, (0, 0).into()), 0);
116− assert_eq!(pos_at_coords(slice, (0, 5).into()), 5); // position on \n
117− assert_eq!(pos_at_coords(slice, (1, 0).into()), 6); // position on w
118− assert_eq!(pos_at_coords(slice, (1, 1).into()), 7); // position on o
119− assert_eq!(pos_at_coords(slice, (1, 4).into()), 10); // position on d
166+ assert_eq!(pos_at_coords(slice, (0, 0).into(), false), 0);
167+ assert_eq!(pos_at_coords(slice, (0, 5).into(), false), 5); // position on \n
168+ assert_eq!(pos_at_coords(slice, (0, 6).into(), false), 6); // position after \n
169+ assert_eq!(pos_at_coords(slice, (0, 6).into(), true), 5); // position after \n
170+ assert_eq!(pos_at_coords(slice, (1, 0).into(), false), 6); // position on w
171+ assert_eq!(pos_at_coords(slice, (1, 1).into(), false), 7); // position on o
172+ assert_eq!(pos_at_coords(slice, (1, 4).into(), false), 10); // position on d
173+
174+ // Test with wide characters.
175+ let text = Rope::from("今日はいい\n");
176+ let slice = text.slice(..);
177+ assert_eq!(pos_at_coords(slice, (0, 0).into(), false), 0);
178+ assert_eq!(pos_at_coords(slice, (0, 1).into(), false), 0);
179+ assert_eq!(pos_at_coords(slice, (0, 2).into(), false), 1);
180+ assert_eq!(pos_at_coords(slice, (0, 3).into(), false), 1);
181+ assert_eq!(pos_at_coords(slice, (0, 4).into(), false), 2);
182+ assert_eq!(pos_at_coords(slice, (0, 6).into(), false), 3);
183+ assert_eq!(pos_at_coords(slice, (0, 8).into(), false), 4);
184+ assert_eq!(pos_at_coords(slice, (0, 10).into(), false), 5);
185+ assert_eq!(pos_at_coords(slice, (0, 11).into(), false), 6);
186+ assert_eq!(pos_at_coords(slice, (0, 11).into(), true), 5);
187+ assert_eq!(pos_at_coords(slice, (1, 0).into(), false), 6);
120188
121189 // test with grapheme clusters
122190 let text = Rope::from("a̐éö̲\r\n");
123191 let slice = text.slice(..);
124− assert_eq!(pos_at_coords(slice, (0, 0).into()), 0);
125− assert_eq!(pos_at_coords(slice, (0, 1).into()), 2);
126− assert_eq!(pos_at_coords(slice, (0, 2).into()), 4);
127− assert_eq!(pos_at_coords(slice, (0, 3).into()), 7); // \r\n is one char here
128− assert_eq!(pos_at_coords(slice, (0, 4).into()), 9);
192+ assert_eq!(pos_at_coords(slice, (0, 0).into(), false), 0);
193+ assert_eq!(pos_at_coords(slice, (0, 1).into(), false), 2);
194+ assert_eq!(pos_at_coords(slice, (0, 2).into(), false), 4);
195+ assert_eq!(pos_at_coords(slice, (0, 3).into(), false), 7); // \r\n is one char here
196+ assert_eq!(pos_at_coords(slice, (0, 4).into(), false), 9);
197+ assert_eq!(pos_at_coords(slice, (0, 4).into(), true), 7);
198+ assert_eq!(pos_at_coords(slice, (1, 0).into(), false), 9);
199+
129200 let text = Rope::from("किमपि");
130201 // 2 - 1 - 2 codepoints
131202 // TODO: delete handling as per https://news.ycombinator.com/item?id=20058454
132203 let slice = text.slice(..);
133− assert_eq!(pos_at_coords(slice, (0, 0).into()), 0);
134− assert_eq!(pos_at_coords(slice, (0, 1).into()), 2);
135− assert_eq!(pos_at_coords(slice, (0, 2).into()), 3);
136− assert_eq!(pos_at_coords(slice, (0, 3).into()), 5); // eol
204+ assert_eq!(pos_at_coords(slice, (0, 0).into(), false), 0);
205+ assert_eq!(pos_at_coords(slice, (0, 1).into(), false), 0);
206+ assert_eq!(pos_at_coords(slice, (0, 2).into(), false), 2);
207+ assert_eq!(pos_at_coords(slice, (0, 3).into(), false), 3);
208+ assert_eq!(pos_at_coords(slice, (0, 4).into(), false), 3);
209+ assert_eq!(pos_at_coords(slice, (0, 5).into(), false), 5); // eol
137210 }
138211}