ui: prompt: Better unicode support

We copied over eval_movement from wezterm, that already solves most of
our problems. self.cursor is now byte-based.

Cargo.lock

@@ -329,6 +329,8 @@ dependencies = [
  "serde_json",
  "tokio",
  "toml",
+ "unicode-segmentation",
+ "unicode-width",
 ]
 
 [[package]]

helix-core/Cargo.toml

@@ -19,9 +19,9 @@ helix-syntax = { version = "0.2", path = "../helix-syntax" }
 ropey = "1.3"
 smallvec = "1.4"
 tendril = "0.4.2"
-unicode-segmentation = "1.7.1"
+unicode-segmentation = "1.7"
 unicode-width = "0.1"
-unicode-general-category = "0.4.0"
+unicode-general-category = "0.4"
 # slab = "0.4.2"
 tree-sitter = "0.19"
 once_cell = "1.8"

helix-term/Cargo.toml

@@ -54,3 +54,6 @@ toml = "0.5"
 
 serde_json = "1.0"
 serde = { version = "1.0", features = ["derive"] }
+
+unicode-segmentation = "1.7"
+unicode-width = "0.1"

helix-term/src/ui/prompt.rs

@@ -6,6 +6,9 @@ use helix_view::{Editor, Theme};
 use std::{borrow::Cow, ops::RangeFrom};
 use tui::terminal::CursorKind;
 
+use unicode_segmentation::{GraphemeCursor, GraphemeIncomplete};
+use unicode_width::UnicodeWidthStr;
+
 pub type Completion = (RangeFrom<usize>, Cow<'static, str>);
 
 pub struct Prompt {
@@ -34,6 +37,17 @@ pub enum CompletionDirection {
     Backward,
 }
 
+#[derive(Debug, Clone, Copy)]
+pub enum Movement {
+    BackwardChar(usize),
+    BackwardWord(usize),
+    ForwardChar(usize),
+    ForwardWord(usize),
+    StartOfLine,
+    EndOfLine,
+    None,
+}
+
 impl Prompt {
     pub fn new(
         prompt: String,
@@ -52,30 +66,125 @@ impl Prompt {
         }
     }
 
-    pub fn insert_char(&mut self, c: char) {
+    /// Compute the cursor position after applying movement
-        let pos = if self.line.is_empty() {
+    /// Taken from: https://github.com/wez/wezterm/blob/e0b62d07ca9bf8ce69a61e30a3c20e7abc48ce7e/termwiz/src/lineedit/mod.rs#L516-L611
-            0
+    fn eval_movement(&self, movement: Movement) -> usize {
+        match movement {
+            Movement::BackwardChar(rep) => {
+                let mut position = self.cursor;
+                for _ in 0..rep {
+                    let mut cursor = GraphemeCursor::new(position, self.line.len(), false);
+                    if let Ok(Some(pos)) = cursor.prev_boundary(&self.line, 0) {
+                        position = pos;
                     } else {
-            self.line
+                        break;
-                .char_indices()
+                    }
-                .nth(self.cursor)
+                }
-                .map(|(pos, _)| pos)
+                position
+            }
+            Movement::BackwardWord(rep) => {
+                let char_indices: Vec<(usize, char)> = self.line.char_indices().collect();
+                if char_indices.is_empty() {
+                    return self.cursor;
+                }
+                let mut char_position = char_indices
+                    .iter()
+                    .position(|(idx, _)| *idx == self.cursor)
+                    .unwrap_or(char_indices.len() - 1);
+
+                for _ in 0..rep {
+                    if char_position == 0 {
+                        break;
+                    }
+
+                    let mut found = None;
+                    for prev in (0..char_position - 1).rev() {
+                        if char_indices[prev].1.is_whitespace() {
+                            found = Some(prev + 1);
+                            break;
+                        }
+                    }
+
+                    char_position = found.unwrap_or(0);
+                }
+                char_indices[char_position].0
+            }
+            Movement::ForwardWord(rep) => {
+                let char_indices: Vec<(usize, char)> = self.line.char_indices().collect();
+                if char_indices.is_empty() {
+                    return self.cursor;
+                }
+                let mut char_position = char_indices
+                    .iter()
+                    .position(|(idx, _)| *idx == self.cursor)
+                    .unwrap_or_else(|| char_indices.len());
+
+                for _ in 0..rep {
+                    // Skip any non-whitespace characters
+                    while char_position < char_indices.len()
+                        && !char_indices[char_position].1.is_whitespace()
+                    {
+                        char_position += 1;
+                    }
+
+                    // Skip any whitespace characters
+                    while char_position < char_indices.len()
+                        && char_indices[char_position].1.is_whitespace()
+                    {
+                        char_position += 1;
+                    }
+
+                    // We are now on the start of the next word
+                }
+                char_indices
+                    .get(char_position)
+                    .map(|(i, _)| *i)
                     .unwrap_or_else(|| self.line.len())
-        };
+            }
-        self.line.insert(pos, c);
+            Movement::ForwardChar(rep) => {
-        self.cursor += 1;
+                let mut position = self.cursor;
+                for _ in 0..rep {
+                    let mut cursor = GraphemeCursor::new(position, self.line.len(), false);
+                    if let Ok(Some(pos)) = cursor.next_boundary(&self.line, 0) {
+                        position = pos;
+                    } else {
+                        break;
+                    }
+                }
+                position
+            }
+            Movement::StartOfLine => 0,
+            Movement::EndOfLine => {
+                let mut cursor =
+                    GraphemeCursor::new(self.line.len().saturating_sub(1), self.line.len(), false);
+                if let Ok(Some(pos)) = cursor.next_boundary(&self.line, 0) {
+                    pos
+                } else {
+                    self.cursor
+                }
+            }
+            Movement::None => self.cursor,
+        }
+    }
+
+    pub fn insert_char(&mut self, c: char) {
+        self.line.insert(self.cursor, c);
+        let mut cursor = GraphemeCursor::new(self.cursor, self.line.len(), false);
+        if let Ok(Some(pos)) = cursor.next_boundary(&self.line, 0) {
+            self.cursor = pos;
+        }
         self.completion = (self.completion_fn)(&self.line);
         self.exit_selection();
     }
 
     pub fn move_char_left(&mut self) {
-        self.cursor = self.cursor.saturating_sub(1)
+        let pos = self.eval_movement(Movement::BackwardChar(1));
+        self.cursor = pos
     }
 
     pub fn move_char_right(&mut self) {
-        if self.cursor < self.line.len() {
+        let pos = self.eval_movement(Movement::ForwardChar(1));
-            self.cursor += 1;
+        self.cursor = pos;
-        }
     }
 
     pub fn move_start(&mut self) {
@@ -87,39 +196,21 @@ impl Prompt {
     }
 
     pub fn delete_char_backwards(&mut self) {
-        if self.cursor > 0 {
+        let pos = self.eval_movement(Movement::BackwardChar(1));
-            let pos = self
+        self.line.replace_range(pos..self.cursor, "");
-                .line
+        self.cursor = pos;
-                .char_indices()
+
-                .nth(self.cursor - 1)
-                .map(|(pos, _)| pos)
-                .expect("line is not empty");
-            self.line.remove(pos);
-            self.cursor -= 1;
-            self.completion = (self.completion_fn)(&self.line);
-        }
         self.exit_selection();
+        self.completion = (self.completion_fn)(&self.line);
     }
 
     pub fn delete_word_backwards(&mut self) {
-        use helix_core::get_general_category;
+        let pos = self.eval_movement(Movement::BackwardWord(1));
-        let mut chars = self.line.char_indices().rev();
+        self.line.replace_range(pos..self.cursor, "");
-        // TODO add skipping whitespace logic here
+        self.cursor = pos;
-        let (mut i, cat) = match chars.next() {
+
-            Some((i, c)) => (i, get_general_category(c)),
-            None => return,
-        };
-        self.cursor -= 1;
-        for (nn, nc) in chars {
-            if get_general_category(nc) != cat {
-                break;
-            }
-            i = nn;
-            self.cursor -= 1;
-        }
-        self.line.drain(i..);
-        self.completion = (self.completion_fn)(&self.line);
         self.exit_selection();
+        self.completion = (self.completion_fn)(&self.line);
     }
 
     pub fn clear(&mut self) {
@@ -363,7 +454,9 @@ impl Component for Prompt {
         (
             Some(Position::new(
                 area.y as usize + line,
-                area.x as usize + self.prompt.len() + self.cursor,
+                area.x as usize
+                    + self.prompt.len()
+                    + UnicodeWidthStr::width(&self.line[..self.cursor]),
             )),
             CursorKind::Block,
         )