Merge pull request #376 from cessen/great_line_ending_and_cursor_range_cleanup

The Great Line Ending & Cursor Range Cleanup

1 files changed, 115 insertions, 29 deletions

diff --git a/helix-core/src/position.rs b/helix-core/src/position.rs
index 3d114b52..611e6b76 100644
--- a/helix-core/src/position.rs
+++ b/helix-core/src/position.rs
@@ -1,6 +1,7 @@
 use crate::{
     chars::char_is_line_ending,
-    graphemes::{nth_next_grapheme_boundary, RopeGraphemes},
+    graphemes::{ensure_grapheme_boundary_prev, RopeGraphemes},
+    line_ending::line_end_char_index,
     RopeSlice,
 };
 
@@ -52,19 +53,50 @@ impl From<Position> for tree_sitter::Point {
     }
 }
 /// Convert a character index to (line, column) coordinates.
+///
+/// TODO: this should be split into two methods: one for visual
+/// row/column, and one for "objective" row/column (possibly with
+/// the column specified in `char`s).  The former would be used
+/// for cursor movement, and the latter would be used for e.g. the
+/// row:column display in the status line.
 pub fn coords_at_pos(text: RopeSlice, pos: usize) -> Position {
     let line = text.char_to_line(pos);
+
     let line_start = text.line_to_char(line);
+    let pos = ensure_grapheme_boundary_prev(text, pos);
     let col = RopeGraphemes::new(text.slice(line_start..pos)).count();
+
     Position::new(line, col)
 }
 
 /// Convert (line, column) coordinates to a character index.
-pub fn pos_at_coords(text: RopeSlice, coords: Position) -> usize {
+///
+/// `is_1_width` specifies whether the position should be treated
+/// as a block cursor or not.  This effects how line-ends are handled.
+/// `false` corresponds to properly round-tripping with `coords_at_pos()`,
+/// whereas `true` will ensure that block cursors don't jump off the
+/// end of the line.
+///
+/// TODO: this should be changed to work in terms of visual row/column, not
+/// graphemes.
+pub fn pos_at_coords(text: RopeSlice, coords: Position, is_1_width: bool) -> usize {
     let Position { row, col } = coords;
     let line_start = text.line_to_char(row);
-    // line_start + col
-    nth_next_grapheme_boundary(text, line_start, col)
+    let line_end = if is_1_width {
+        line_end_char_index(&text, row)
+    } else {
+        text.line_to_char((row + 1).min(text.len_lines()))
+    };
+
+    let mut col_char_offset = 0;
+    for (i, g) in RopeGraphemes::new(text.slice(line_start..line_end)).enumerate() {
+        if i == col {
+            break;
+        }
+        col_char_offset += g.chars().count();
+    }
+
+    line_start + col_char_offset
 }
 
 #[cfg(test)]
@@ -80,55 +112,109 @@ mod test {
 
     #[test]
     fn test_coords_at_pos() {
-        // let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
-        // let slice = text.slice(..);
-        // assert_eq!(coords_at_pos(slice, 0), (0, 0).into());
-        // assert_eq!(coords_at_pos(slice, 5), (0, 5).into()); // position on \n
-        // assert_eq!(coords_at_pos(slice, 6), (1, 0).into()); // position on w
-        // assert_eq!(coords_at_pos(slice, 7), (1, 1).into()); // position on o
-        // assert_eq!(coords_at_pos(slice, 10), (1, 4).into()); // position on d
-
-        // test with grapheme clusters
+        let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
+        let slice = text.slice(..);
+        assert_eq!(coords_at_pos(slice, 0), (0, 0).into());
+        assert_eq!(coords_at_pos(slice, 5), (0, 5).into()); // position on \n
+        assert_eq!(coords_at_pos(slice, 6), (1, 0).into()); // position on w
+        assert_eq!(coords_at_pos(slice, 7), (1, 1).into()); // position on o
+        assert_eq!(coords_at_pos(slice, 10), (1, 4).into()); // position on d
+
+        // Test with wide characters.
+        // TODO: account for character width.
+        let text = Rope::from("今日はいい\n");
+        let slice = text.slice(..);
+        assert_eq!(coords_at_pos(slice, 0), (0, 0).into());
+        assert_eq!(coords_at_pos(slice, 1), (0, 1).into());
+        assert_eq!(coords_at_pos(slice, 2), (0, 2).into());
+        assert_eq!(coords_at_pos(slice, 3), (0, 3).into());
+        assert_eq!(coords_at_pos(slice, 4), (0, 4).into());
+        assert_eq!(coords_at_pos(slice, 5), (0, 5).into());
+        assert_eq!(coords_at_pos(slice, 6), (1, 0).into());
+
+        // Test with grapheme clusters.
         let text = Rope::from("a̐éö̲\r\n");
         let slice = text.slice(..);
         assert_eq!(coords_at_pos(slice, 0), (0, 0).into());
         assert_eq!(coords_at_pos(slice, 2), (0, 1).into());
         assert_eq!(coords_at_pos(slice, 4), (0, 2).into());
         assert_eq!(coords_at_pos(slice, 7), (0, 3).into());
+        assert_eq!(coords_at_pos(slice, 9), (1, 0).into());
 
-        let text = Rope::from("किमपि");
+        // Test with wide-character grapheme clusters.
+        // TODO: account for character width.
+        let text = Rope::from("किमपि\n");
         let slice = text.slice(..);
         assert_eq!(coords_at_pos(slice, 0), (0, 0).into());
         assert_eq!(coords_at_pos(slice, 2), (0, 1).into());
         assert_eq!(coords_at_pos(slice, 3), (0, 2).into());
         assert_eq!(coords_at_pos(slice, 5), (0, 3).into());
+        assert_eq!(coords_at_pos(slice, 6), (1, 0).into());
+
+        // Test with tabs.
+        // Todo: account for tab stops.
+        let text = Rope::from("\tHello\n");
+        let slice = text.slice(..);
+        assert_eq!(coords_at_pos(slice, 0), (0, 0).into());
+        assert_eq!(coords_at_pos(slice, 1), (0, 1).into());
+        assert_eq!(coords_at_pos(slice, 2), (0, 2).into());
     }
 
     #[test]
     fn test_pos_at_coords() {
         let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
         let slice = text.slice(..);
-        assert_eq!(pos_at_coords(slice, (0, 0).into()), 0);
-        assert_eq!(pos_at_coords(slice, (0, 5).into()), 5); // position on \n
-        assert_eq!(pos_at_coords(slice, (1, 0).into()), 6); // position on w
-        assert_eq!(pos_at_coords(slice, (1, 1).into()), 7); // position on o
-        assert_eq!(pos_at_coords(slice, (1, 4).into()), 10); // position on d
+        assert_eq!(pos_at_coords(slice, (0, 0).into(), false), 0);
+        assert_eq!(pos_at_coords(slice, (0, 5).into(), false), 5); // position on \n
+        assert_eq!(pos_at_coords(slice, (0, 6).into(), false), 6); // position after \n
+        assert_eq!(pos_at_coords(slice, (0, 6).into(), true), 5); // position after \n
+        assert_eq!(pos_at_coords(slice, (1, 0).into(), false), 6); // position on w
+        assert_eq!(pos_at_coords(slice, (1, 1).into(), false), 7); // position on o
+        assert_eq!(pos_at_coords(slice, (1, 4).into(), false), 10); // position on d
 
-        // test with grapheme clusters
+        // Test with wide characters.
+        // TODO: account for character width.
+        let text = Rope::from("今日はいい\n");
+        let slice = text.slice(..);
+        assert_eq!(pos_at_coords(slice, (0, 0).into(), false), 0);
+        assert_eq!(pos_at_coords(slice, (0, 1).into(), false), 1);
+        assert_eq!(pos_at_coords(slice, (0, 2).into(), false), 2);
+        assert_eq!(pos_at_coords(slice, (0, 3).into(), false), 3);
+        assert_eq!(pos_at_coords(slice, (0, 4).into(), false), 4);
+        assert_eq!(pos_at_coords(slice, (0, 5).into(), false), 5);
+        assert_eq!(pos_at_coords(slice, (0, 6).into(), false), 6);
+        assert_eq!(pos_at_coords(slice, (0, 6).into(), true), 5);
+        assert_eq!(pos_at_coords(slice, (1, 0).into(), false), 6);
+
+        // Test with grapheme clusters.
         let text = Rope::from("a̐éö̲\r\n");
         let slice = text.slice(..);
-        assert_eq!(pos_at_coords(slice, (0, 0).into()), 0);
-        assert_eq!(pos_at_coords(slice, (0, 1).into()), 2);
-        assert_eq!(pos_at_coords(slice, (0, 2).into()), 4);
-        assert_eq!(pos_at_coords(slice, (0, 3).into()), 7); // \r\n is one char here
-        assert_eq!(pos_at_coords(slice, (0, 4).into()), 9);
+        assert_eq!(pos_at_coords(slice, (0, 0).into(), false), 0);
+        assert_eq!(pos_at_coords(slice, (0, 1).into(), false), 2);
+        assert_eq!(pos_at_coords(slice, (0, 2).into(), false), 4);
+        assert_eq!(pos_at_coords(slice, (0, 3).into(), false), 7); // \r\n is one char here
+        assert_eq!(pos_at_coords(slice, (0, 4).into(), false), 9);
+        assert_eq!(pos_at_coords(slice, (0, 4).into(), true), 7);
+        assert_eq!(pos_at_coords(slice, (1, 0).into(), false), 9);
+
+        // Test with wide-character grapheme clusters.
+        // TODO: account for character width.
         let text = Rope::from("किमपि");
         // 2 - 1 - 2 codepoints
         // TODO: delete handling as per https://news.ycombinator.com/item?id=20058454
         let slice = text.slice(..);
-        assert_eq!(pos_at_coords(slice, (0, 0).into()), 0);
-        assert_eq!(pos_at_coords(slice, (0, 1).into()), 2);
-        assert_eq!(pos_at_coords(slice, (0, 2).into()), 3);
-        assert_eq!(pos_at_coords(slice, (0, 3).into()), 5); // eol
+        assert_eq!(pos_at_coords(slice, (0, 0).into(), false), 0);
+        assert_eq!(pos_at_coords(slice, (0, 1).into(), false), 2);
+        assert_eq!(pos_at_coords(slice, (0, 2).into(), false), 3);
+        assert_eq!(pos_at_coords(slice, (0, 3).into(), false), 5);
+        assert_eq!(pos_at_coords(slice, (0, 3).into(), true), 5);
+
+        // Test with tabs.
+        // Todo: account for tab stops.
+        let text = Rope::from("\tHello\n");
+        let slice = text.slice(..);
+        assert_eq!(pos_at_coords(slice, (0, 0).into(), false), 0);
+        assert_eq!(pos_at_coords(slice, (0, 1).into(), false), 1);
+        assert_eq!(pos_at_coords(slice, (0, 2).into(), false), 2);
     }
 }