Merge e02d7683c6 into fed3edcab7

This was previously highlighted as `punctuation` because the capture
applied to the whole `(color_value)` node rather than the `"#"` child
node specifically.

Cargo.lock

@@ -1552,6 +1552,7 @@ dependencies = [
  "helix-event",
  "helix-loader",
  "helix-lsp",
+ "helix-parsec",
  "helix-stdx",
  "helix-tui",
  "helix-vcs",

book/src/generated/typable-cmd.md

@@ -16,6 +16,8 @@
 | `:write-buffer-close`, `:wbc` | Write changes to disk and closes the buffer. Accepts an optional path (:write-buffer-close some/path.txt) |
 | `:write-buffer-close!`, `:wbc!` | Force write changes to disk creating necessary subdirectories and closes the buffer. Accepts an optional path (:write-buffer-close! some/path.txt) |
 | `:new`, `:n` | Create a new scratch buffer. |
+| `:goto-mark` | Go to the selection saved in a register. Register can be provided as argument or selected register else ^ will be used |
+| `:register-mark` | Save current selection into a register. Register can be provided as argument or selected register else ^ will be used |
 | `:format`, `:fmt` | Format the file using an external formatter or language server. |
 | `:indent-style` | Set the indentation style for editing. ('t' for tabs or 1-16 for number of spaces.) |
 | `:line-ending` | Set the document's default line ending. Options: crlf, lf. |

helix-core/src/selection.rs

@@ -12,10 +12,11 @@ use crate::{
     tree_sitter::Node,
     Assoc, ChangeSet, RopeSlice,
 };
+use helix_parsec::{seq, take_until, Parser};
 use helix_stdx::range::is_subset;
 use helix_stdx::rope::{self, RopeSliceExt};
 use smallvec::{smallvec, SmallVec};
-use std::{borrow::Cow, iter, slice};
+use std::{borrow::Cow, fmt::Display, iter, slice};
 
 /// A single selection range.
 ///
@@ -392,6 +393,34 @@ impl Range {
     }
 }
 
+impl Display for Range {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "({},{})", self.anchor, self.head)
+    }
+}
+
+impl TryFrom<&str> for Range {
+    type Error = String;
+
+    fn try_from(value: &str) -> Result<Self, Self::Error> {
+        let parser = seq!(
+            "(",
+            take_until(|c| c == ','),
+            ",",
+            take_until(|c| c == ')'),
+            ")"
+        );
+        match parser.parse(value) {
+            Ok((_tail, (_, anchor, _, head, _))) => Ok(Self {
+                anchor: anchor.parse::<usize>().map_err(|e| e.to_string())?,
+                head: head.parse::<usize>().map_err(|e| e.to_string())?,
+                old_visual_position: None,
+            }),
+            Err(e) => Err(e.to_string()),
+        }
+    }
+}
+
 impl From<(usize, usize)> for Range {
     fn from((anchor, head): (usize, usize)) -> Self {
         Self {
@@ -884,6 +913,54 @@ mod test {
     use super::*;
     use crate::Rope;
 
+    #[test]
+    fn parse_range() -> Result<(), String> {
+        // sometimes we want Ok, sometimes we want Err, but we never want a panic
+        assert_eq!(
+            Range::try_from("(0,28)"),
+            Ok(Range {
+                anchor: 0,
+                head: 28,
+                old_visual_position: None
+            })
+        );
+        assert_eq!(
+            Range::try_from("(3456789,123456789)"),
+            Ok(Range {
+                anchor: 3456789,
+                head: 123456789,
+                old_visual_position: None
+            })
+        );
+        assert_eq!(Range::try_from("(,)"), Err("(,)".to_string()));
+        assert_eq!(
+            Range::try_from("(asdf,asdf)"),
+            Err("invalid digit found in string".to_string())
+        );
+        assert_eq!(Range::try_from("()"), Err("()".to_string()));
+        assert_eq!(
+            Range::try_from("(-4,ALSK)"),
+            Err("invalid digit found in string".to_string())
+        );
+        assert_eq!(
+            Range::try_from("(⦡⓼␀⍆ⴉ├⺶⍄⾨,⦡⓼␀⍆ⴉ├⺶⍄⾨)"),
+            Err("invalid digit found in string".to_string())
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn display_range() {
+        assert_eq!(
+            Range {
+                anchor: 72,
+                head: 28,
+                old_visual_position: None,
+            }
+            .to_string(),
+            "(72,28)".to_string(),
+        );
+    }
     #[test]
     #[should_panic]
     fn test_new_empty() {

helix-stdx/src/rope.rs

@@ -135,7 +135,9 @@ pub trait RopeSliceExt<'a>: Sized {
     /// let graphemes: Vec<_> = text.graphemes().collect();
     /// assert_eq!(graphemes.as_slice(), &["😶‍🌫️", "🏴‍☠️", "🖼️"]);
     /// ```
-    fn graphemes(self) -> RopeGraphemes<'a>;
+    fn graphemes(self) -> RopeGraphemes<'a> {
+        self.graphemes_at(0)
+    }
     /// Returns an iterator over the grapheme clusters in the slice, reversed.
     ///
     /// The returned iterator starts at the end of the slice and ends at the beginning of the
@@ -150,7 +152,127 @@ pub trait RopeSliceExt<'a>: Sized {
     /// let graphemes: Vec<_> = text.graphemes_rev().collect();
     /// assert_eq!(graphemes.as_slice(), &["🖼️", "🏴‍☠️", "😶‍🌫️"]);
     /// ```
-    fn graphemes_rev(self) -> RevRopeGraphemes<'a>;
+    fn graphemes_rev(self) -> RopeGraphemes<'a>;
+    /// Returns an iterator over the grapheme clusters in the slice at the given byte index.
+    ///
+    /// # Example
+    ///
+    /// ```
+    /// # use ropey::Rope;
+    /// # use helix_stdx::rope::RopeSliceExt;
+    /// let text = Rope::from_str("😶‍🌫️🏴‍☠️🖼️");
+    /// // 14 is the byte index of the pirate flag's starting cluster boundary.
+    /// let graphemes: Vec<_> = text.slice(..).graphemes_at(14).collect();
+    /// assert_eq!(graphemes.as_slice(), &["🏴‍☠️", "🖼️"]);
+    /// // 27 is the byte index of the pirate flag's ending cluster boundary.
+    /// let graphemes: Vec<_> = text.slice(..).graphemes_at(27).reversed().collect();
+    /// assert_eq!(graphemes.as_slice(), &["🏴‍☠️", "😶‍🌫️"]);
+    /// ```
+    fn graphemes_at(self, byte_idx: usize) -> RopeGraphemes<'a>;
+    /// Returns an iterator over the grapheme clusters in a rope and the byte index where each
+    /// grapheme cluster starts.
+    ///
+    /// # Example
+    ///
+    /// ```
+    /// # use ropey::Rope;
+    /// # use helix_stdx::rope::RopeSliceExt;
+    /// let text = Rope::from_str("😶‍🌫️🏴‍☠️🖼️");
+    /// let slice = text.slice(..);
+    /// let graphemes: Vec<_> = slice.grapheme_indices_at(0).collect();
+    /// assert_eq!(
+    ///   graphemes.as_slice(),
+    ///   &[(0, "😶‍🌫️".into()), (14, "🏴‍☠️".into()), (27, "🖼️".into())]
+    /// );
+    /// let graphemes: Vec<_> = slice.grapheme_indices_at(slice.len_bytes()).reversed().collect();
+    /// assert_eq!(
+    ///   graphemes.as_slice(),
+    ///   &[(27, "🖼️".into()), (14, "🏴‍☠️".into()), (0, "😶‍🌫️".into())]
+    /// );
+    /// ```
+    fn grapheme_indices_at(self, byte_idx: usize) -> RopeGraphemeIndices<'a>;
+    /// Finds the byte index of the next grapheme boundary after `byte_idx`.
+    ///
+    /// If the byte index lies on the last grapheme cluster in the slice then this function
+    /// returns `RopeSlice::len_bytes`.
+    ///
+    /// # Example
+    ///
+    /// ```
+    /// # use ropey::Rope;
+    /// # use helix_stdx::rope::RopeSliceExt;
+    /// let text = Rope::from_str("😶‍🌫️🏴‍☠️🖼️");
+    /// let slice = text.slice(..);
+    /// let mut byte_idx = 0;
+    /// assert_eq!(slice.graphemes_at(byte_idx).next(), Some("😶‍🌫️".into()));
+    /// byte_idx = slice.next_grapheme_boundary(byte_idx);
+    /// assert_eq!(slice.graphemes_at(byte_idx).next(), Some("🏴‍☠️".into()));
+    ///
+    /// // If `byte_idx` does not lie on a character or grapheme boundary then this function is
+    /// // functionally the same as `ceil_grapheme_boundary`.
+    /// assert_eq!(slice.next_grapheme_boundary(byte_idx - 1), byte_idx);
+    /// assert_eq!(slice.next_grapheme_boundary(byte_idx - 2), byte_idx);
+    /// assert_eq!(slice.next_grapheme_boundary(byte_idx + 1), slice.next_grapheme_boundary(byte_idx));
+    /// assert_eq!(slice.next_grapheme_boundary(byte_idx + 2), slice.next_grapheme_boundary(byte_idx));
+    ///
+    /// byte_idx = slice.next_grapheme_boundary(byte_idx);
+    /// assert_eq!(slice.graphemes_at(byte_idx).next(), Some("🖼️".into()));
+    /// byte_idx = slice.next_grapheme_boundary(byte_idx);
+    /// assert_eq!(slice.graphemes_at(byte_idx).next(), None);
+    /// assert_eq!(byte_idx, slice.len_bytes());
+    /// ```
+    fn next_grapheme_boundary(self, byte_idx: usize) -> usize {
+        self.nth_next_grapheme_boundary(byte_idx, 1)
+    }
+    /// Finds the byte index of the `n`th grapheme cluster after the given `byte_idx`.
+    ///
+    /// If there are fewer than `n` grapheme clusters after `byte_idx` in the rope then this
+    /// function returns `RopeSlice::len_bytes`.
+    ///
+    /// This is functionally equivalent to calling `next_grapheme_boundary` `n` times but is more
+    /// efficient.
+    fn nth_next_grapheme_boundary(self, byte_idx: usize, n: usize) -> usize;
+    /// Finds the byte index of the previous grapheme boundary before `byte_idx`.
+    ///
+    /// If the byte index lies on the first grapheme cluster in the slice then this function
+    /// returns zero.
+    ///
+    /// # Example
+    ///
+    /// ```
+    /// # use ropey::Rope;
+    /// # use helix_stdx::rope::RopeSliceExt;
+    /// let text = Rope::from_str("😶‍🌫️🏴‍☠️🖼️");
+    /// let slice = text.slice(..);
+    /// let mut byte_idx = text.len_bytes();
+    /// assert_eq!(slice.graphemes_at(byte_idx).prev(), Some("🖼️".into()));
+    /// byte_idx = slice.prev_grapheme_boundary(byte_idx);
+    /// assert_eq!(slice.graphemes_at(byte_idx).prev(), Some("🏴‍☠️".into()));
+    ///
+    /// // If `byte_idx` does not lie on a character or grapheme boundary then this function is
+    /// // functionally the same as `floor_grapheme_boundary`.
+    /// assert_eq!(slice.prev_grapheme_boundary(byte_idx + 1), byte_idx);
+    /// assert_eq!(slice.prev_grapheme_boundary(byte_idx + 2), byte_idx);
+    /// assert_eq!(slice.prev_grapheme_boundary(byte_idx - 1), slice.prev_grapheme_boundary(byte_idx));
+    /// assert_eq!(slice.prev_grapheme_boundary(byte_idx - 2), slice.prev_grapheme_boundary(byte_idx));
+    ///
+    /// byte_idx = slice.prev_grapheme_boundary(byte_idx);
+    /// assert_eq!(slice.graphemes_at(byte_idx).prev(), Some("😶‍🌫️".into()));
+    /// byte_idx = slice.prev_grapheme_boundary(byte_idx);
+    /// assert_eq!(slice.graphemes_at(byte_idx).prev(), None);
+    /// assert_eq!(byte_idx, 0);
+    /// ```
+    fn prev_grapheme_boundary(self, byte_idx: usize) -> usize {
+        self.nth_prev_grapheme_boundary(byte_idx, 1)
+    }
+    /// Finds the byte index of the `n`th grapheme cluster before the given `byte_idx`.
+    ///
+    /// If there are fewer than `n` grapheme clusters before `byte_idx` in the rope then this
+    /// function returns zero.
+    ///
+    /// This is functionally equivalent to calling `prev_grapheme_boundary` `n` times but is more
+    /// efficient.
+    fn nth_prev_grapheme_boundary(self, byte_idx: usize, n: usize) -> usize;
 }
 
 impl<'a> RopeSliceExt<'a> for RopeSlice<'a> {
@@ -335,31 +457,111 @@ impl<'a> RopeSliceExt<'a> for RopeSlice<'a> {
         }
     }
 
-    fn graphemes(self) -> RopeGraphemes<'a> {
-        let mut chunks = self.chunks();
-        let first_chunk = chunks.next().unwrap_or("");
+    fn graphemes_rev(self) -> RopeGraphemes<'a> {
+        self.graphemes_at(self.len_bytes()).reversed()
+    }
+
+    fn graphemes_at(self, byte_idx: usize) -> RopeGraphemes<'a> {
+        // Bounds check
+        assert!(byte_idx <= self.len_bytes());
+
+        let (mut chunks, chunk_byte_idx, _, _) = self.chunks_at_byte(byte_idx);
+        let current_chunk = chunks.next().unwrap_or("");
+
         RopeGraphemes {
             text: self,
             chunks,
-            cur_chunk: first_chunk,
-            cur_chunk_start: 0,
-            cursor: GraphemeCursor::new(0, self.len_bytes(), true),
+            current_chunk,
+            chunk_byte_idx,
+            cursor: GraphemeCursor::new(byte_idx, self.len_bytes(), true),
+            is_reversed: false,
         }
     }
 
-    fn graphemes_rev(self) -> RevRopeGraphemes<'a> {
-        let (mut chunks, mut cur_chunk_start, _, _) = self.chunks_at_byte(self.len_bytes());
-        chunks.reverse();
-        let first_chunk = chunks.next().unwrap_or("");
-        cur_chunk_start -= first_chunk.len();
-        RevRopeGraphemes {
-            text: self,
-            chunks,
-            cur_chunk: first_chunk,
-            cur_chunk_start,
-            cursor: GraphemeCursor::new(self.len_bytes(), self.len_bytes(), true),
+    fn grapheme_indices_at(self, byte_idx: usize) -> RopeGraphemeIndices<'a> {
+        // Bounds check
+        assert!(byte_idx <= self.len_bytes());
+        RopeGraphemeIndices {
+            front_offset: byte_idx,
+            iter: self.graphemes_at(byte_idx),
+            is_reversed: false,
         }
     }
+
+    fn nth_next_grapheme_boundary(self, mut byte_idx: usize, n: usize) -> usize {
+        // Bounds check
+        assert!(byte_idx <= self.len_bytes());
+
+        byte_idx = self.floor_char_boundary(byte_idx);
+
+        // Get the chunk with our byte index in it.
+        let (mut chunk, mut chunk_byte_idx, _, _) = self.chunk_at_byte(byte_idx);
+
+        // Set up the grapheme cursor.
+        let mut gc = GraphemeCursor::new(byte_idx, self.len_bytes(), true);
+
+        // Find the nth next grapheme cluster boundary.
+        for _ in 0..n {
+            loop {
+                match gc.next_boundary(chunk, chunk_byte_idx) {
+                    Ok(None) => return self.len_bytes(),
+                    Ok(Some(boundary)) => {
+                        byte_idx = boundary;
+                        break;
+                    }
+                    Err(GraphemeIncomplete::NextChunk) => {
+                        chunk_byte_idx += chunk.len();
+                        let (a, _, _, _) = self.chunk_at_byte(chunk_byte_idx);
+                        chunk = a;
+                    }
+                    Err(GraphemeIncomplete::PreContext(n)) => {
+                        let ctx_chunk = self.chunk_at_byte(n - 1).0;
+                        gc.provide_context(ctx_chunk, n - ctx_chunk.len());
+                    }
+                    _ => unreachable!(),
+                }
+            }
+        }
+
+        byte_idx
+    }
+
+    fn nth_prev_grapheme_boundary(self, mut byte_idx: usize, n: usize) -> usize {
+        // Bounds check
+        assert!(byte_idx <= self.len_bytes());
+
+        byte_idx = self.ceil_char_boundary(byte_idx);
+
+        // Get the chunk with our byte index in it.
+        let (mut chunk, mut chunk_byte_idx, _, _) = self.chunk_at_byte(byte_idx);
+
+        // Set up the grapheme cursor.
+        let mut gc = GraphemeCursor::new(byte_idx, self.len_bytes(), true);
+
+        for _ in 0..n {
+            loop {
+                match gc.prev_boundary(chunk, chunk_byte_idx) {
+                    Ok(None) => return 0,
+                    Ok(Some(boundary)) => {
+                        byte_idx = boundary;
+                        break;
+                    }
+                    Err(GraphemeIncomplete::PrevChunk) => {
+                        let (a, b, _, _) = self.chunk_at_byte(chunk_byte_idx - 1);
+                        chunk = a;
+                        chunk_byte_idx = b;
+                    }
+                    Err(GraphemeIncomplete::PreContext(n)) => {
+                        let ctx_chunk = self.chunk_at_byte(n - 1).0;
+                        gc.provide_context(ctx_chunk, n - ctx_chunk.len());
+                    }
+                    _ => unreachable!(),
+                }
+            }
+        }
+
+        byte_idx
+    }
 }
 
 // copied from std
@@ -370,13 +572,19 @@ const fn is_utf8_char_boundary(b: u8) -> bool {
 }
 
 /// An iterator over the graphemes of a `RopeSlice`.
+///
+/// This iterator is cursor-like: rather than implementing DoubleEndedIterator it can be reversed
+/// like a cursor. This style matches `Bytes` and `Chars` iterator types in Ropey and is more
+/// natural and useful for wrapping `GraphemeCursor`.
 #[derive(Clone)]
 pub struct RopeGraphemes<'a> {
     text: RopeSlice<'a>,
     chunks: Chunks<'a>,
-    cur_chunk: &'a str,
-    cur_chunk_start: usize,
+    current_chunk: &'a str,
+    /// Byte index of the start of the current chunk.
+    chunk_byte_idx: usize,
     cursor: GraphemeCursor,
+    is_reversed: bool,
 }
 
 impl fmt::Debug for RopeGraphemes<'_> {
@@ -384,112 +592,178 @@ impl fmt::Debug for RopeGraphemes<'_> {
         f.debug_struct("RopeGraphemes")
             .field("text", &self.text)
             .field("chunks", &self.chunks)
-            .field("cur_chunk", &self.cur_chunk)
-            .field("cur_chunk_start", &self.cur_chunk_start)
+            .field("current_chunk", &self.current_chunk)
+            .field("chunk_byte_idx", &self.chunk_byte_idx)
             // .field("cursor", &self.cursor)
+            .field("is_reversed", &self.is_reversed)
             .finish()
     }
 }
 
+impl<'a> RopeGraphemes<'a> {
+    #[allow(clippy::should_implement_trait)]
+    pub fn next(&mut self) -> Option<RopeSlice<'a>> {
+        if self.is_reversed {
+            self.prev_impl()
+        } else {
+            self.next_impl()
+        }
+    }
+
+    pub fn prev(&mut self) -> Option<RopeSlice<'a>> {
+        if self.is_reversed {
+            self.next_impl()
+        } else {
+            self.prev_impl()
+        }
+    }
+
+    pub fn reverse(&mut self) {
+        self.is_reversed = !self.is_reversed;
+    }
+
+    #[must_use]
+    pub fn reversed(mut self) -> Self {
+        self.reverse();
+        self
+    }
+
+    fn next_impl(&mut self) -> Option<RopeSlice<'a>> {
+        let a = self.cursor.cur_cursor();
+        let b;
+        loop {
+            match self
+                .cursor
+                .next_boundary(self.current_chunk, self.chunk_byte_idx)
+            {
+                Ok(None) => return None,
+                Ok(Some(boundary)) => {
+                    b = boundary;
+                    break;
+                }
+                Err(GraphemeIncomplete::NextChunk) => {
+                    self.chunk_byte_idx += self.current_chunk.len();
+                    self.current_chunk = self.chunks.next().unwrap_or("");
+                }
+                Err(GraphemeIncomplete::PreContext(idx)) => {
+                    let (chunk, byte_idx, _, _) = self.text.chunk_at_byte(idx.saturating_sub(1));
+                    self.cursor.provide_context(chunk, byte_idx);
+                }
+                _ => unreachable!(),
+            }
+        }
+
+        if a < self.chunk_byte_idx {
+            Some(self.text.byte_slice(a..b))
+        } else {
+            let a2 = a - self.chunk_byte_idx;
+            let b2 = b - self.chunk_byte_idx;
+            Some((&self.current_chunk[a2..b2]).into())
+        }
+    }
+
+    fn prev_impl(&mut self) -> Option<RopeSlice<'a>> {
+        let a = self.cursor.cur_cursor();
+        let b;
+        loop {
+            match self
+                .cursor
+                .prev_boundary(self.current_chunk, self.chunk_byte_idx)
+            {
+                Ok(None) => return None,
+                Ok(Some(boundary)) => {
+                    b = boundary;
+                    break;
+                }
+                Err(GraphemeIncomplete::PrevChunk) => {
+                    self.current_chunk = self.chunks.prev().unwrap_or("");
+                    self.chunk_byte_idx -= self.current_chunk.len();
+                }
+                Err(GraphemeIncomplete::PreContext(idx)) => {
+                    let (chunk, byte_idx, _, _) = self.text.chunk_at_byte(idx.saturating_sub(1));
+                    self.cursor.provide_context(chunk, byte_idx);
+                }
+                _ => unreachable!(),
+            }
+        }
+
+        if a >= self.chunk_byte_idx + self.current_chunk.len() {
+            Some(self.text.byte_slice(b..a))
+        } else {
+            let a2 = a - self.chunk_byte_idx;
+            let b2 = b - self.chunk_byte_idx;
+            Some((&self.current_chunk[b2..a2]).into())
+        }
+    }
+}
+
 impl<'a> Iterator for RopeGraphemes<'a> {
     type Item = RopeSlice<'a>;
 
     fn next(&mut self) -> Option<Self::Item> {
-        let a = self.cursor.cur_cursor();
-        let b;
-        loop {
-            match self
-                .cursor
-                .next_boundary(self.cur_chunk, self.cur_chunk_start)
-            {
-                Ok(None) => {
-                    return None;
-                }
-                Ok(Some(n)) => {
-                    b = n;
-                    break;
-                }
-                Err(GraphemeIncomplete::NextChunk) => {
-                    self.cur_chunk_start += self.cur_chunk.len();
-                    self.cur_chunk = self.chunks.next().unwrap_or("");
-                }
-                Err(GraphemeIncomplete::PreContext(idx)) => {
-                    let (chunk, byte_idx, _, _) = self.text.chunk_at_byte(idx.saturating_sub(1));
-                    self.cursor.provide_context(chunk, byte_idx);
-                }
-                _ => unreachable!(),
-            }
+        RopeGraphemes::next(self)
     }
+}
 
-        if a < self.cur_chunk_start {
-            Some(self.text.byte_slice(a..b))
+/// An iterator over the grapheme clusters in a rope and the byte indices where each grapheme
+/// cluster starts.
+///
+/// This iterator wraps `RopeGraphemes` and is also cursor-like. Use `reverse` or `reversed` to
+/// toggle the direction of the iterator. See [RopeGraphemes].
+#[derive(Debug, Clone)]
+pub struct RopeGraphemeIndices<'a> {
+    front_offset: usize,
+    iter: RopeGraphemes<'a>,
+    is_reversed: bool,
+}
+
+impl<'a> RopeGraphemeIndices<'a> {
+    #[allow(clippy::should_implement_trait)]
+    pub fn next(&mut self) -> Option<(usize, RopeSlice<'a>)> {
+        if self.is_reversed {
+            self.prev_impl()
         } else {
-            let a2 = a - self.cur_chunk_start;
-            let b2 = b - self.cur_chunk_start;
-            Some((&self.cur_chunk[a2..b2]).into())
+            self.next_impl()
         }
     }
+
+    pub fn prev(&mut self) -> Option<(usize, RopeSlice<'a>)> {
+        if self.is_reversed {
+            self.next_impl()
+        } else {
+            self.prev_impl()
+        }
+    }
+
+    pub fn reverse(&mut self) {
+        self.is_reversed = !self.is_reversed;
+    }
+
+    #[must_use]
+    pub fn reversed(mut self) -> Self {
+        self.reverse();
+        self
+    }
+
+    fn next_impl(&mut self) -> Option<(usize, RopeSlice<'a>)> {
+        let slice = self.iter.next()?;
+        let idx = self.front_offset;
+        self.front_offset += slice.len_bytes();
+        Some((idx, slice))
+    }
+
+    fn prev_impl(&mut self) -> Option<(usize, RopeSlice<'a>)> {
+        let slice = self.iter.prev()?;
+        self.front_offset -= slice.len_bytes();
+        Some((self.front_offset, slice))
+    }
 }
 
-/// An iterator over the graphemes of a `RopeSlice` in reverse.
-#[derive(Clone)]
-pub struct RevRopeGraphemes<'a> {
-    text: RopeSlice<'a>,
-    chunks: Chunks<'a>,
-    cur_chunk: &'a str,
-    cur_chunk_start: usize,
-    cursor: GraphemeCursor,
-}
-
-impl fmt::Debug for RevRopeGraphemes<'_> {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        f.debug_struct("RevRopeGraphemes")
-            .field("text", &self.text)
-            .field("chunks", &self.chunks)
-            .field("cur_chunk", &self.cur_chunk)
-            .field("cur_chunk_start", &self.cur_chunk_start)
-            // .field("cursor", &self.cursor)
-            .finish()
-    }
-}
-
-impl<'a> Iterator for RevRopeGraphemes<'a> {
-    type Item = RopeSlice<'a>;
+impl<'a> Iterator for RopeGraphemeIndices<'a> {
+    type Item = (usize, RopeSlice<'a>);
 
     fn next(&mut self) -> Option<Self::Item> {
-        let a = self.cursor.cur_cursor();
-        let b;
-        loop {
-            match self
-                .cursor
-                .prev_boundary(self.cur_chunk, self.cur_chunk_start)
-            {
-                Ok(None) => {
-                    return None;
-                }
-                Ok(Some(n)) => {
-                    b = n;
-                    break;
-                }
-                Err(GraphemeIncomplete::PrevChunk) => {
-                    self.cur_chunk = self.chunks.next().unwrap_or("");
-                    self.cur_chunk_start -= self.cur_chunk.len();
-                }
-                Err(GraphemeIncomplete::PreContext(idx)) => {
-                    let (chunk, byte_idx, _, _) = self.text.chunk_at_byte(idx.saturating_sub(1));
-                    self.cursor.provide_context(chunk, byte_idx);
-                }
-                _ => unreachable!(),
-            }
-        }
-
-        if a >= self.cur_chunk_start + self.cur_chunk.len() {
-            Some(self.text.byte_slice(b..a))
-        } else {
-            let a2 = a - self.cur_chunk_start;
-            let b2 = b - self.cur_chunk_start;
-            Some((&self.cur_chunk[b2..a2]).into())
-        }
+        RopeGraphemeIndices::next(self)
     }
 }
 

helix-term/Cargo.toml

@@ -49,16 +49,32 @@ helix-lsp = { path = "../helix-lsp" }
 helix-dap = { path = "../helix-dap" }
 helix-vcs = { path = "../helix-vcs" }
 helix-loader = { path = "../helix-loader" }
+helix-parsec = { path = "../helix-parsec" }
 
 anyhow = "1"
 once_cell = "1.21"
 
-tokio = { version = "1", features = ["rt", "rt-multi-thread", "io-util", "io-std", "time", "process", "macros", "fs", "parking_lot"] }
-tui = { path = "../helix-tui", package = "helix-tui", default-features = false, features = ["crossterm"] }
+tokio = { version = "1", features = [
+  "rt",
+  "rt-multi-thread",
+  "io-util",
+  "io-std",
+  "time",
+  "process",
+  "macros",
+  "fs",
+  "parking_lot",
+] }
+tui = { path = "../helix-tui", package = "helix-tui", default-features = false, features = [
+  "crossterm",
+] }
 crossterm = { version = "0.28", features = ["event-stream"] }
 signal-hook = "0.3"
 tokio-stream = "0.1"
-futures-util = { version = "0.3", features = ["std", "async-await"], default-features = false }
+futures-util = { version = "0.3", features = [
+  "std",
+  "async-await",
+], default-features = false }
 arc-swap = { version = "1.7.1" }
 termini = "1"
 indexmap = "2.9"
@@ -96,7 +112,11 @@ signal-hook-tokio = { version = "0.3", features = ["futures-v0_3"] }
 libc = "0.2.172"
 
 [target.'cfg(target_os = "macos")'.dependencies]
-crossterm = { version = "0.28", features = ["event-stream", "use-dev-tty", "libc"] }
+crossterm = { version = "0.28", features = [
+  "event-stream",
+  "use-dev-tty",
+  "libc",
+] }
 
 [build-dependencies]
 helix-loader = { path = "../helix-loader" }

helix-term/src/commands.rs

@@ -5,6 +5,7 @@ pub(crate) mod typed;
 pub use dap::*;
 use futures_util::FutureExt;
 use helix_event::status;
+use helix_parsec::{seq, take_until, Parser};
 use helix_stdx::{
     path::{self, find_paths},
     rope::{self, RopeSliceExt},
@@ -47,6 +48,7 @@ use helix_view::{
     info::Info,
     input::KeyEvent,
     keyboard::KeyCode,
+    register::RegisterValues,
     theme::Style,
     tree,
     view::View,
@@ -74,6 +76,7 @@ use std::{
     future::Future,
     io::Read,
     num::NonZeroUsize,
+    str::FromStr,
 };
 
 use std::{
@@ -6635,6 +6638,10 @@ fn extend_to_word(cx: &mut Context) {
     jump_to_word(cx, Movement::Extend)
 }
 
+fn read_from_register(editor: &mut Editor, reg: char) -> Option<RegisterValues> {
+    editor.registers.read(reg, &*editor)
+}
+
 fn jump_to_label(cx: &mut Context, labels: Vec<Range>, behaviour: Movement) {
     let doc = doc!(cx.editor);
     let alphabet = &cx.editor.config().jump_label_alphabet;

helix-term/src/commands/typed.rs

@@ -479,6 +479,155 @@ fn new_file(cx: &mut compositor::Context, _args: Args, event: PromptEvent) -> an
     Ok(())
 }
 
+fn register_mark(
+    cx: &mut compositor::Context,
+    args: Args,
+    event: PromptEvent,
+) -> anyhow::Result<()> {
+    if event != PromptEvent::Validate {
+        return Ok(());
+    };
+    let register_name: char = args
+        .first()
+        .map_or_else(|| cx.editor.selected_register, |s| s.chars().next())
+        .unwrap_or('^');
+
+    let (view, doc) = current!(cx.editor);
+
+    let ranges_str = doc
+        .selection(view.id)
+        .ranges()
+        .iter()
+        .map(|r| r.to_string())
+        .collect::<Vec<String>>();
+
+    // we have to take because of cell
+    let history = doc.history.take();
+    let current_history_point = history.current_revision();
+    doc.history.replace(history);
+
+    // doc_id so we know which doc to switch to
+    // current_history_point so we can apply changes
+    // to our selection when we restore it.
+    // the rest of the elements are just the stringified ranges
+    let mut register_val = vec![
+        format!("{}", doc.id()),
+        format!("{}", current_history_point),
+    ];
+    register_val.extend(ranges_str);
+
+    cx.editor.registers.write(register_name, register_val)?;
+
+    cx.editor
+        .set_status(format!("Saved selection bookmark to [{}]", register_name));
+    Ok(())
+}
+
+fn parse_mark_register_contents(
+    registers_vals: Option<RegisterValues>,
+) -> anyhow::Result<(DocumentId, usize, Selection)> {
+    match registers_vals {
+        Some(rv) => {
+            let mut rv_iter = rv.into_iter();
+
+            let Some(doc_id) = rv_iter
+                .next()
+                .map(|c| c.into_owned())
+                .and_then(|s| s.try_into().ok())
+            else {
+                return Err(anyhow!("Register did not contain valid document id"));
+            };
+            let Some(history_rev) = rv_iter
+                .next()
+                .map(|c| c.into_owned())
+                .and_then(|s| s.parse().ok())
+            else {
+                return Err(anyhow!("Register did not contain valid revision number"));
+            };
+
+            let Ok(ranges) = rv_iter
+                .map(|tup| {
+                    let s = tup.into_owned();
+                    let range_parser = seq!(
+                        "(",
+                        take_until(|c| c == ','),
+                        ",",
+                        take_until(|c| c == ')'),
+                        ")"
+                    );
+                    let Ok((_tail, (_lparen, anchor_str, _comma, head_str, _rparen))) =
+                        range_parser.parse(&s)
+                    else {
+                        return Err(format!("Could not parse range from string: {}", s));
+                    };
+
+                    let Ok(anchor) = <usize as FromStr>::from_str(anchor_str) else {
+                        return Err(format!("Could not parse range from string: {}", s));
+                    };
+                    let Ok(head) = <usize as FromStr>::from_str(head_str) else {
+                        return Err(format!("Could not parse range from string: {}", s));
+                    };
+
+                    Ok(Range {
+                        anchor,
+                        head,
+                        old_visual_position: None,
+                    })
+                })
+                // reverse the iterators so the first range will end up as the primary when we push them
+                .rev()
+                .collect::<Result<Vec<Range>, String>>()
+            else {
+                return Err(anyhow!("Some ranges in the register failed to parse!"));
+            };
+
+            let mut ranges_iter = ranges.into_iter();
+
+            let last_range = ranges_iter.next().unwrap(); // safe since there is always at least one range
+            let mut selection = Selection::from(last_range);
+            for r in ranges_iter {
+                selection = selection.push(r);
+            }
+
+            Ok((doc_id, history_rev, selection))
+        }
+        None => Err(anyhow!("Register was empty")),
+    }
+}
+
+fn goto_mark(cx: &mut compositor::Context, args: Args, event: PromptEvent) -> anyhow::Result<()> {
+    if event != PromptEvent::Validate {
+        return Ok(());
+    };
+    let register_name: char = args
+        .first()
+        .map_or_else(|| cx.editor.selected_register, |s| s.chars().next())
+        .unwrap_or('^');
+
+    let scrolloff = cx.editor.config().scrolloff;
+    // use some helper functions to avoid making the borrow checker angry
+    let registers_vals = read_from_register(cx.editor, register_name);
+    let (doc_id, history_rev, mut selection) = parse_mark_register_contents(registers_vals)?;
+
+    cx.editor.switch(doc_id, Action::Replace);
+
+    let (view, doc) = current!(cx.editor);
+    let history = doc.history.take();
+    let revisions_to_apply = history.changes_since(history_rev);
+    doc.history.replace(history);
+
+    selection = match revisions_to_apply {
+        Some(t) => selection.map(t.changes()),
+        None => selection,
+    };
+
+    doc.set_selection(view.id, selection);
+
+    view.ensure_cursor_in_view(doc, scrolloff);
+
+    Ok(())
+}
+
 fn format(cx: &mut compositor::Context, _args: Args, event: PromptEvent) -> anyhow::Result<()> {
     if event != PromptEvent::Validate {
         return Ok(());
@@ -2758,6 +2907,28 @@ pub const TYPABLE_COMMAND_LIST: &[TypableCommand] = &[
             ..Signature::DEFAULT
         },
     },
+    TypableCommand {
+        name: "goto-mark",
+        aliases: &[],
+        doc: "Go to the selection saved in a register. Register can be provided as argument or selected register else ^ will be used",
+        fun: goto_mark,
+        completer: CommandCompleter::positional(&[completers::register]),
+        signature: Signature {
+            positionals: (0, Some(1)),
+            ..Signature::DEFAULT
+        },
+    },
+    TypableCommand {
+        name: "register-mark",
+        aliases: &[],
+        doc: "Save current selection into a register. Register can be provided as argument or selected register else ^ will be used",
+        fun: register_mark,
+        completer: CommandCompleter::positional(&[completers::register]),
+        signature: Signature {
+            positionals: (0, Some(1)),
+            ..Signature::DEFAULT
+        },
+    },
     TypableCommand {
         name: "format",
         aliases: &["fmt"],

helix-term/src/keymap/default.rs

@@ -336,6 +336,7 @@ pub fn default() -> HashMap<Mode, KeyTrie> {
 
         "C-a" => increment,
         "C-x" => decrement,
+
     });
     let mut select = normal.clone();
     select.merge_nodes(keymap!({ "Select mode"

helix-term/tests/test/movement.rs

@@ -65,6 +65,43 @@ async fn insert_to_normal_mode_cursor_position() -> anyhow::Result<()> {
     Ok(())
 }
 
+#[tokio::test(flavor = "multi_thread")]
+async fn bookmark() -> anyhow::Result<()> {
+    // add a mark and then immediately paste it out
+    test((
+        indoc! {"\
+            #[|Lorem]# 
+            ipsum 
+            #(|Lorem)# 
+            ipsum 
+            #(|Lorem)# 
+            ipsum 
+            #(|Lorem)# 
+            ipsum 
+            #(|Lorem)# 
+            ipsum"
+        },
+        // make a mark, make changes to the doc, colapse selection by going to end of doc
+        // then resore mark and see the selection is still good
+        ":register-mark<space>1<ret>casdf<esc>ge:goto-mark<space>1<ret>",
+        indoc! {"\
+            #[|asdf]# 
+            ipsum 
+            #(|asdf)# 
+            ipsum 
+            #(|asdf)# 
+            ipsum 
+            #(|asdf)# 
+            ipsum 
+            #(|asdf)# 
+            ipsum"
+        },
+    ))
+    .await?;
+
+    Ok(())
+}
+
 #[tokio::test(flavor = "multi_thread")]
 async fn surround_by_character() -> anyhow::Result<()> {
     // Only pairs matching the passed character count

helix-view/src/lib.rs

@@ -19,7 +19,7 @@ pub mod theme;
 pub mod tree;
 pub mod view;
 
-use std::num::NonZeroUsize;
+use std::num::{NonZeroUsize, ParseIntError};
 
 // uses NonZeroUsize so Option<DocumentId> use a byte rather than two
 #[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash, Debug)]
@@ -32,6 +32,21 @@ impl Default for DocumentId {
     }
 }
 
+impl TryFrom<&str> for DocumentId {
+    type Error = ParseIntError;
+
+    fn try_from(value: &str) -> Result<Self, Self::Error> {
+        Ok(Self(value.parse::<NonZeroUsize>()?))
+    }
+}
+
+impl TryFrom<String> for DocumentId {
+    type Error = ParseIntError;
+
+    fn try_from(value: String) -> Result<Self, Self::Error> {
+        Ok(Self(value.parse::<NonZeroUsize>()?))
+    }
+}
 impl std::fmt::Display for DocumentId {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         f.write_fmt(format_args!("{}", self.0))

runtime/queries/css/highlights.scm

@@ -65,7 +65,7 @@
 ] @punctuation
 
 (string_value) @string
-((color_value) "#") @string.special
+(color_value "#" @string.special)
 (color_value) @string.special
 
 (integer_value) @constant.numeric.integer