Merge branch 'master' into debug

17 files changed, 296 insertions, 100 deletions

diff --git a/helix-core/src/auto_pairs.rs b/helix-core/src/auto_pairs.rs
index 9b901e9b..cc966852 100644
--- a/helix-core/src/auto_pairs.rs
+++ b/helix-core/src/auto_pairs.rs
@@ -1,3 +1,6 @@
+//! When typing the opening character of one of the possible pairs defined below,
+//! this module provides the functionality to insert the paired closing character.
+
 use crate::{Range, Rope, Selection, Tendril, Transaction};
 use smallvec::SmallVec;
 
diff --git a/helix-core/src/chars.rs b/helix-core/src/chars.rs
index 24133dd3..c8e5efbd 100644
--- a/helix-core/src/chars.rs
+++ b/helix-core/src/chars.rs
@@ -1,3 +1,5 @@
+//! Utility functions to categorize a `char`.
+
 use crate::LineEnding;
 
 #[derive(Debug, Eq, PartialEq)]
diff --git a/helix-core/src/comment.rs b/helix-core/src/comment.rs
index 3d8e1ce3..b22a95a6 100644
--- a/helix-core/src/comment.rs
+++ b/helix-core/src/comment.rs
@@ -1,3 +1,6 @@
+//! This module contains the functionality toggle comments on lines over the selection
+//! using the comment character defined in the user's `languages.toml`
+
 use crate::{
     find_first_non_whitespace_char, Change, Rope, RopeSlice, Selection, Tendril, Transaction,
 };
@@ -60,7 +63,7 @@ pub fn toggle_line_comments(doc: &Rope, selection: &Selection, token: Option<&st
     let token = token.unwrap_or("//");
     let comment = Tendril::from(format!("{} ", token));
 
-    let mut lines: Vec<usize> = Vec::new();
+    let mut lines: Vec<usize> = Vec::with_capacity(selection.len());
 
     let mut min_next_line = 0;
     for selection in selection {
diff --git a/helix-core/src/diagnostic.rs b/helix-core/src/diagnostic.rs
index e08a71e7..ad1ba16a 100644
--- a/helix-core/src/diagnostic.rs
+++ b/helix-core/src/diagnostic.rs
@@ -1,3 +1,6 @@
+//! LSP diagnostic utility types.
+
+/// Describes the severity level of a [`Diagnostic`].
 #[derive(Debug, Eq, PartialEq)]
 pub enum Severity {
     Error,
@@ -6,12 +9,14 @@ pub enum Severity {
     Hint,
 }
 
-#[derive(Debug)]
+/// A range of `char`s within the text.
+#[derive(Debug, Clone, Copy, PartialOrd, Ord, PartialEq, Eq)]
 pub struct Range {
     pub start: usize,
     pub end: usize,
 }
 
+/// Corresponds to [`lsp_types::Diagnostic`](https://docs.rs/lsp-types/0.91.0/lsp_types/struct.Diagnostic.html)
 #[derive(Debug)]
 pub struct Diagnostic {
     pub range: Range,
diff --git a/helix-core/src/graphemes.rs b/helix-core/src/graphemes.rs
index 0465fe51..c6398875 100644
--- a/helix-core/src/graphemes.rs
+++ b/helix-core/src/graphemes.rs
@@ -1,4 +1,6 @@
-// Based on https://github.com/cessen/led/blob/c4fa72405f510b7fd16052f90a598c429b3104a6/src/graphemes.rs
+//! Utility functions to traverse the unicode graphemes of a `Rope`'s text contents.
+//!
+//! Based on <https://github.com/cessen/led/blob/c4fa72405f510b7fd16052f90a598c429b3104a6/src/graphemes.rs>
 use ropey::{iter::Chunks, str_utils::byte_to_char_idx, RopeSlice};
 use unicode_segmentation::{GraphemeCursor, GraphemeIncomplete};
 use unicode_width::UnicodeWidthStr;
diff --git a/helix-core/src/history.rs b/helix-core/src/history.rs
index 67ded166..b53c01fe 100644
--- a/helix-core/src/history.rs
+++ b/helix-core/src/history.rs
@@ -4,48 +4,50 @@ use regex::Regex;
 use std::num::NonZeroUsize;
 use std::time::{Duration, Instant};
 
-// Stores the history of changes to a buffer.
-//
-// Currently the history is represented as a vector of revisions. The vector
-// always has at least one element: the empty root revision. Each revision
-// with the exception of the root has a parent revision, a [Transaction]
-// that can be applied to its parent to transition from the parent to itself,
-// and an inversion of that transaction to transition from the parent to its
-// latest child.
-//
-// When using `u` to undo a change, an inverse of the stored transaction will
-// be applied which will transition the buffer to the parent state.
-//
-// Each revision with the exception of the last in the vector also has a
-// last child revision. When using `U` to redo a change, the last child transaction
-// will be applied to the current state of the buffer.
-//
-// The current revision is the one currently displayed in the buffer.
-//
-// Commiting a new revision to the history will update the last child of the
-// current revision, and push a new revision to the end of the vector.
-//
-// Revisions are commited with a timestamp. :earlier and :later can be used
-// to jump to the closest revision to a moment in time relative to the timestamp
-// of the current revision plus (:later) or minus (:earlier) the duration
-// given to the command. If a single integer is given, the editor will instead
-// jump the given number of revisions in the vector.
-//
-// Limitations:
-//  * Changes in selections currently don't commit history changes. The selection
-//    will only be updated to the state after a commited buffer change.
-//  * The vector of history revisions is currently unbounded. This might
-//    cause the memory consumption to grow significantly large during long
-//    editing sessions.
-//  * Because delete transactions currently don't store the text that they
-//    delete, we also store an inversion of the transaction.
+/// Stores the history of changes to a buffer.
+///
+/// Currently the history is represented as a vector of revisions. The vector
+/// always has at least one element: the empty root revision. Each revision
+/// with the exception of the root has a parent revision, a [Transaction]
+/// that can be applied to its parent to transition from the parent to itself,
+/// and an inversion of that transaction to transition from the parent to its
+/// latest child.
+///
+/// When using `u` to undo a change, an inverse of the stored transaction will
+/// be applied which will transition the buffer to the parent state.
+///
+/// Each revision with the exception of the last in the vector also has a
+/// last child revision. When using `U` to redo a change, the last child transaction
+/// will be applied to the current state of the buffer.
+///
+/// The current revision is the one currently displayed in the buffer.
+///
+/// Commiting a new revision to the history will update the last child of the
+/// current revision, and push a new revision to the end of the vector.
+///
+/// Revisions are commited with a timestamp. :earlier and :later can be used
+/// to jump to the closest revision to a moment in time relative to the timestamp
+/// of the current revision plus (:later) or minus (:earlier) the duration
+/// given to the command. If a single integer is given, the editor will instead
+/// jump the given number of revisions in the vector.
+///
+/// Limitations:
+///  * Changes in selections currently don't commit history changes. The selection
+///    will only be updated to the state after a commited buffer change.
+///  * The vector of history revisions is currently unbounded. This might
+///    cause the memory consumption to grow significantly large during long
+///    editing sessions.
+///  * Because delete transactions currently don't store the text that they
+///    delete, we also store an inversion of the transaction.
+///
+/// Using time to navigate the history: <https://github.com/helix-editor/helix/pull/194>
 #[derive(Debug)]
 pub struct History {
     revisions: Vec<Revision>,
     current: usize,
 }
 
-// A single point in history. See [History] for more information.
+/// A single point in history. See [History] for more information.
 #[derive(Debug)]
 struct Revision {
     parent: usize,
@@ -111,6 +113,7 @@ impl History {
         self.current == 0
     }
 
+    /// Undo the last edit.
     pub fn undo(&mut self) -> Option<&Transaction> {
         if self.at_root() {
             return None;
@@ -121,6 +124,7 @@ impl History {
         Some(&current_revision.inversion)
     }
 
+    /// Redo the last edit.
     pub fn redo(&mut self) -> Option<&Transaction> {
         let current_revision = &self.revisions[self.current];
         let last_child = current_revision.last_child?;
@@ -147,8 +151,8 @@ impl History {
         }
     }
 
-    // List of nodes on the way from `n` to 'a`. Doesn`t include `a`.
-    // Includes `n` unless `a == n`. `a` must be an ancestor of `n`.
+    /// List of nodes on the way from `n` to 'a`. Doesn`t include `a`.
+    /// Includes `n` unless `a == n`. `a` must be an ancestor of `n`.
     fn path_up(&self, mut n: usize, a: usize) -> Vec<usize> {
         let mut path = Vec::new();
         while n != a {
@@ -158,6 +162,7 @@ impl History {
         path
     }
 
+    /// Create a [`Transaction`] that will jump to a specific revision in the history.
     fn jump_to(&mut self, to: usize) -> Vec<Transaction> {
         let lca = self.lowest_common_ancestor(self.current, to);
         let up = self.path_up(self.current, lca);
@@ -171,10 +176,12 @@ impl History {
         up_txns.chain(down_txns).collect()
     }
 
+    /// Creates a [`Transaction`] that will undo `delta` revisions.
     fn jump_backward(&mut self, delta: usize) -> Vec<Transaction> {
         self.jump_to(self.current.saturating_sub(delta))
     }
 
+    /// Creates a [`Transaction`] that will redo `delta` revisions.
     fn jump_forward(&mut self, delta: usize) -> Vec<Transaction> {
         self.jump_to(
             self.current
@@ -183,7 +190,7 @@ impl History {
         )
     }
 
-    // Helper for a binary search case below.
+    /// Helper for a binary search case below.
     fn revision_closer_to_instant(&self, i: usize, instant: Instant) -> usize {
         let dur_im1 = instant.duration_since(self.revisions[i - 1].timestamp);
         let dur_i = self.revisions[i].timestamp.duration_since(instant);
@@ -194,6 +201,8 @@ impl History {
         }
     }
 
+    /// Creates a [`Transaction`] that will match a revision created at around
+    /// `instant`.
     fn jump_instant(&mut self, instant: Instant) -> Vec<Transaction> {
         let search_result = self
             .revisions
@@ -209,6 +218,8 @@ impl History {
         self.jump_to(revision)
     }
 
+    /// Creates a [`Transaction`] that will match a revision created `duration` ago
+    /// from the timestamp of current revision.
     fn jump_duration_backward(&mut self, duration: Duration) -> Vec<Transaction> {
         match self.revisions[self.current].timestamp.checked_sub(duration) {
             Some(instant) => self.jump_instant(instant),
@@ -216,6 +227,8 @@ impl History {
         }
     }
 
+    /// Creates a [`Transaction`] that will match a revision created `duration` in
+    /// the future from the timestamp of the current revision.
     fn jump_duration_forward(&mut self, duration: Duration) -> Vec<Transaction> {
         match self.revisions[self.current].timestamp.checked_add(duration) {
             Some(instant) => self.jump_instant(instant),
@@ -223,6 +236,7 @@ impl History {
         }
     }
 
+    /// Creates an undo [`Transaction`].
     pub fn earlier(&mut self, uk: UndoKind) -> Vec<Transaction> {
         use UndoKind::*;
         match uk {
@@ -231,6 +245,7 @@ impl History {
         }
     }
 
+    /// Creates a redo [`Transaction`].
     pub fn later(&mut self, uk: UndoKind) -> Vec<Transaction> {
         use UndoKind::*;
         match uk {
@@ -240,13 +255,14 @@ impl History {
     }
 }
 
+/// Whether to undo by a number of edits or a duration of time.
 #[derive(Debug, PartialEq)]
 pub enum UndoKind {
     Steps(usize),
     TimePeriod(std::time::Duration),
 }
 
-// A subset of sytemd.time time span syntax units.
+/// A subset of sytemd.time time span syntax units.
 const TIME_UNITS: &[(&[&str], &str, u64)] = &[
     (&["seconds", "second", "sec", "s"], "seconds", 1),
     (&["minutes", "minute", "min", "m"], "minutes", 60),
@@ -254,11 +270,20 @@ const TIME_UNITS: &[(&[&str], &str, u64)] = &[
     (&["days", "day", "d"], "days", 24 * 60 * 60),
 ];
 
+/// Checks if the duration input can be turned into a valid duration. It must be a
+/// positive integer and denote the [unit of time.](`TIME_UNITS`)
+/// Examples of valid durations:
+///  * `5 sec`
+///  * `5 min`
+///  * `5 hr`
+///  * `5 days`
 static DURATION_VALIDATION_REGEX: Lazy<Regex> =
     Lazy::new(|| Regex::new(r"^(?:\d+\s*[a-z]+\s*)+$").unwrap());
 
+/// Captures both the number and unit as separate capture groups.
 static NUMBER_UNIT_REGEX: Lazy<Regex> = Lazy::new(|| Regex::new(r"(\d+)\s*([a-z]+)").unwrap());
 
+/// Parse a string (e.g. "5 sec") and try to convert it into a [`Duration`].
 fn parse_human_duration(s: &str) -> Result<Duration, String> {
     if !DURATION_VALIDATION_REGEX.is_match(s) {
         return Err("duration should be composed \
diff --git a/helix-core/src/indent.rs b/helix-core/src/indent.rs
index 1f32d038..df158363 100644
--- a/helix-core/src/indent.rs
+++ b/helix-core/src/indent.rs
@@ -464,6 +464,7 @@ where
                     unit: String::from("    "),
                 }),
                 indent_query: OnceCell::new(),
+                textobject_query: OnceCell::new(),
                 debugger: None,
             }],
         });
diff --git a/helix-core/src/lib.rs b/helix-core/src/lib.rs
index 0854eb04..f4284139 100644
--- a/helix-core/src/lib.rs
+++ b/helix-core/src/lib.rs
@@ -35,6 +35,7 @@ pub fn find_first_non_whitespace_char(line: RopeSlice) -> Option<usize> {
     line.chars().position(|ch| !ch.is_whitespace())
 }
 
+/// Find `.git` root.
 pub fn find_root(root: Option<&str>) -> Option<std::path::PathBuf> {
     let current_dir = std::env::current_dir().expect("unable to determine current directory");
 
@@ -193,7 +194,7 @@ pub use tendril::StrTendril as Tendril;
 pub use {regex, tree_sitter};
 
 pub use graphemes::RopeGraphemes;
-pub use position::{coords_at_pos, pos_at_coords, Position};
+pub use position::{coords_at_pos, pos_at_coords, visual_coords_at_pos, Position};
 pub use selection::{Range, Selection};
 pub use smallvec::{smallvec, SmallVec};
 pub use syntax::Syntax;
diff --git a/helix-core/src/line_ending.rs b/helix-core/src/line_ending.rs
index 18ea5f9f..3541305c 100644
--- a/helix-core/src/line_ending.rs
+++ b/helix-core/src/line_ending.rs
@@ -20,7 +20,7 @@ pub enum LineEnding {
 
 impl LineEnding {
     #[inline]
-    pub fn len_chars(&self) -> usize {
+    pub const fn len_chars(&self) -> usize {
         match self {
             Self::Crlf => 2,
             _ => 1,
@@ -28,7 +28,7 @@ impl LineEnding {
     }
 
     #[inline]
-    pub fn as_str(&self) -> &'static str {
+    pub const fn as_str(&self) -> &'static str {
         match self {
             Self::Crlf => "\u{000D}\u{000A}",
             Self::LF => "\u{000A}",
@@ -42,7 +42,7 @@ impl LineEnding {
     }
 
     #[inline]
-    pub fn from_char(ch: char) -> Option<LineEnding> {
+    pub const fn from_char(ch: char) -> Option<LineEnding> {
         match ch {
             '\u{000A}' => Some(LineEnding::LF),
             '\u{000B}' => Some(LineEnding::VT),
diff --git a/helix-core/src/movement.rs b/helix-core/src/movement.rs
index 5d080545..9e85bd21 100644
--- a/helix-core/src/movement.rs
+++ b/helix-core/src/movement.rs
@@ -53,6 +53,10 @@ pub fn move_vertically(
     let pos = range.cursor(slice);
 
     // Compute the current position's 2d coordinates.
+    // TODO: switch this to use `visual_coords_at_pos` rather than
+    // `coords_at_pos` as this will cause a jerky movement when the visual
+    // position does not match, like moving from a line with tabs/CJK to
+    // a line without
     let Position { row, col } = coords_at_pos(slice, pos);
     let horiz = range.horiz.unwrap_or(col as u32);
 
diff --git a/helix-core/src/object.rs b/helix-core/src/object.rs
index d9558dd8..717c5994 100644
--- a/helix-core/src/object.rs
+++ b/helix-core/src/object.rs
@@ -13,8 +13,13 @@ pub fn expand_selection(syntax: &Syntax, text: RopeSlice, selection: &Selection)
         let parent = match tree
             .root_node()
             .descendant_for_byte_range(from, to)
-            .and_then(|node| node.parent())
-        {
+            .and_then(|node| {
+                if node.child_count() == 0 || (node.start_byte() == from && node.end_byte() == to) {
+                    node.parent()
+                } else {
+                    Some(node)
+                }
+            }) {
             Some(parent) => parent,
             None => return range,
         };
diff --git a/helix-core/src/position.rs b/helix-core/src/position.rs
index 08a8aed5..c6018ce6 100644
--- a/helix-core/src/position.rs
+++ b/helix-core/src/position.rs
@@ -2,6 +2,7 @@ use crate::{
     chars::char_is_line_ending,
     graphemes::{ensure_grapheme_boundary_prev, RopeGraphemes},
     line_ending::line_end_char_index,
+    unicode::width::UnicodeWidthChar,
     RopeSlice,
 };
 
@@ -54,11 +55,8 @@ 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.
+/// column in `char` count which can be used for row:column display in
+/// status line. See [`visual_coords_at_pos`] for a visual one.
 pub fn coords_at_pos(text: RopeSlice, pos: usize) -> Position {
     let line = text.char_to_line(pos);
 
@@ -69,6 +67,28 @@ pub fn coords_at_pos(text: RopeSlice, pos: usize) -> Position {
     Position::new(line, col)
 }
 
+/// Convert a character index to (line, column) coordinates visually.
+///
+/// Takes \t, double-width characters (CJK) into account as well as text
+/// not in the document in the future.
+/// See [`coords_at_pos`] for an "objective" one.
+pub fn visual_coords_at_pos(text: RopeSlice, pos: usize, tab_width: 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 = text
+        .slice(line_start..pos)
+        .chars()
+        .flat_map(|c| match c {
+            '\t' => Some(tab_width),
+            c => UnicodeWidthChar::width(c),
+        })
+        .sum();
+
+    Position::new(line, col)
+}
+
 /// Convert (line, column) coordinates to a character index.
 ///
 /// If the `line` coordinate is beyond the end of the file, the EOF
@@ -130,7 +150,6 @@ mod test {
         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());
@@ -151,7 +170,6 @@ mod test {
         assert_eq!(coords_at_pos(slice, 9), (1, 0).into());
 
         // 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());
@@ -161,7 +179,6 @@ mod test {
         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());
@@ -170,6 +187,54 @@ mod test {
     }
 
     #[test]
+    fn test_visual_coords_at_pos() {
+        let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
+        let slice = text.slice(..);
+        assert_eq!(visual_coords_at_pos(slice, 0, 8), (0, 0).into());
+        assert_eq!(visual_coords_at_pos(slice, 5, 8), (0, 5).into()); // position on \n
+        assert_eq!(visual_coords_at_pos(slice, 6, 8), (1, 0).into()); // position on w
+        assert_eq!(visual_coords_at_pos(slice, 7, 8), (1, 1).into()); // position on o
+        assert_eq!(visual_coords_at_pos(slice, 10, 8), (1, 4).into()); // position on d
+
+        // Test with wide characters.
+        let text = Rope::from("今日はいい\n");
+        let slice = text.slice(..);
+        assert_eq!(visual_coords_at_pos(slice, 0, 8), (0, 0).into());
+        assert_eq!(visual_coords_at_pos(slice, 1, 8), (0, 2).into());
+        assert_eq!(visual_coords_at_pos(slice, 2, 8), (0, 4).into());
+        assert_eq!(visual_coords_at_pos(slice, 3, 8), (0, 6).into());
+        assert_eq!(visual_coords_at_pos(slice, 4, 8), (0, 8).into());
+        assert_eq!(visual_coords_at_pos(slice, 5, 8), (0, 10).into());
+        assert_eq!(visual_coords_at_pos(slice, 6, 8), (1, 0).into());
+
+        // Test with grapheme clusters.
+        let text = Rope::from("a̐éö̲\r\n");
+        let slice = text.slice(..);
+        assert_eq!(visual_coords_at_pos(slice, 0, 8), (0, 0).into());
+        assert_eq!(visual_coords_at_pos(slice, 2, 8), (0, 1).into());
+        assert_eq!(visual_coords_at_pos(slice, 4, 8), (0, 2).into());
+        assert_eq!(visual_coords_at_pos(slice, 7, 8), (0, 3).into());
+        assert_eq!(visual_coords_at_pos(slice, 9, 8), (1, 0).into());
+
+        // Test with wide-character grapheme clusters.
+        // TODO: account for cluster.
+        let text = Rope::from("किमपि\n");
+        let slice = text.slice(..);
+        assert_eq!(visual_coords_at_pos(slice, 0, 8), (0, 0).into());
+        assert_eq!(visual_coords_at_pos(slice, 2, 8), (0, 2).into());
+        assert_eq!(visual_coords_at_pos(slice, 3, 8), (0, 3).into());
+        assert_eq!(visual_coords_at_pos(slice, 5, 8), (0, 5).into());
+        assert_eq!(visual_coords_at_pos(slice, 6, 8), (1, 0).into());
+
+        // Test with tabs.
+        let text = Rope::from("\tHello\n");
+        let slice = text.slice(..);
+        assert_eq!(visual_coords_at_pos(slice, 0, 8), (0, 0).into());
+        assert_eq!(visual_coords_at_pos(slice, 1, 8), (0, 8).into());
+        assert_eq!(visual_coords_at_pos(slice, 2, 8), (0, 9).into());
+    }
+
+    #[test]
     fn test_pos_at_coords() {
         let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
         let slice = text.slice(..);
diff --git a/helix-core/src/register.rs b/helix-core/src/register.rs
index c3e6652e..c5444eb7 100644
--- a/helix-core/src/register.rs
+++ b/helix-core/src/register.rs
@@ -7,7 +7,7 @@ pub struct Register {
 }
 
 impl Register {
-    pub fn new(name: char) -> Self {
+    pub const fn new(name: char) -> Self {
         Self {
             name,
             values: Vec::new(),
@@ -18,7 +18,7 @@ impl Register {
         Self { name, values }
     }
 
-    pub fn name(&self) -> char {
+    pub const fn name(&self) -> char {
         self.name
     }
 
diff --git a/helix-core/src/selection.rs b/helix-core/src/selection.rs
index 18af4d08..f3b5d2c8 100644
--- a/helix-core/src/selection.rs
+++ b/helix-core/src/selection.rs
@@ -362,6 +362,11 @@ impl Selection {
 
     /// Adds a new range to the selection and makes it the primary range.
     pub fn remove(mut self, index: usize) -> Self {
+        assert!(
+            self.ranges.len() > 1,
+            "can't remove the last range from a selection!"
+        );
+
         self.ranges.remove(index);
         if index < self.primary_index || self.primary_index == self.ranges.len() {
             self.primary_index -= 1;
@@ -369,6 +374,12 @@ impl Selection {
         self
     }
 
+    /// Replace a range in the selection with a new range.
+    pub fn replace(mut self, index: usize, range: Range) -> Self {
+        self.ranges[index] = range;
+        self.normalize()
+    }
+
     /// Map selections over a set of changes. Useful for adjusting the selection position after
     /// applying changes to a document.
     pub fn map(self, changes: &ChangeSet) -> Self {
diff --git a/helix-core/src/syntax.rs b/helix-core/src/syntax.rs
index 441802a5..18504c21 100644
--- a/helix-core/src/syntax.rs
+++ b/helix-core/src/syntax.rs
@@ -49,7 +49,7 @@ pub struct Configuration {
 #[serde(rename_all = "kebab-case")]
 pub struct LanguageConfiguration {
     #[serde(rename = "name")]
-    pub(crate) language_id: String,
+    pub language_id: String,
     pub scope: String,           // source.rust
     pub file_types: Vec<String>, // filename ends_with? <Gemfile, rb, etc>
     pub roots: Vec<String>,      // these indicate project roots <.git, Cargo.toml>
@@ -76,6 +76,8 @@ pub struct LanguageConfiguration {
 
     #[serde(skip)]
     pub(crate) indent_query: OnceCell<Option<IndentQuery>>,
+    #[serde(skip)]
+    pub(crate) textobject_query: OnceCell<Option<TextObjectQuery>>,
     #[serde(skip_serializing_if = "Option::is_none")]
     pub debugger: Option<DebugAdapterConfig>,
 }
@@ -160,6 +162,32 @@ pub struct IndentQuery {
     pub outdent: HashSet<String>,
 }
 
+#[derive(Debug)]
+pub struct TextObjectQuery {
+    pub query: Query,
+}
+
+impl TextObjectQuery {
+    /// Run the query on the given node and return sub nodes which match given
+    /// capture ("function.inside", "class.around", etc).
+    pub fn capture_nodes<'a>(
+        &'a self,
+        capture_name: &str,
+        node: Node<'a>,
+        slice: RopeSlice<'a>,
+        cursor: &'a mut QueryCursor,
+    ) -> Option<impl Iterator<Item = Node<'a>>> {
+        let capture_idx = self.query.capture_index_for_name(capture_name)?;
+        let captures = cursor.captures(&self.query, node, RopeProvider(slice));
+
+        captures
+            .filter_map(move |(mat, idx)| {
+                (mat.captures[idx].index == capture_idx).then(|| mat.captures[idx].node)
+            })
+            .into()
+    }
+}
+
 fn load_runtime_file(language: &str, filename: &str) -> Result<String, std::io::Error> {
     let path = crate::RUNTIME_DIR
         .join("queries")
@@ -208,13 +236,14 @@ impl LanguageConfiguration {
         // highlights_query += "\n(ERROR) @error";
 
         let injections_query = read_query(&language, "injections.scm");
-
         let locals_query = read_query(&language, "locals.scm");
 
         if highlights_query.is_empty() {
             None
         } else {
-            let language = get_language(&crate::RUNTIME_DIR, &self.language_id).ok()?;
+            let language = get_language(&crate::RUNTIME_DIR, &self.language_id)
+                .map_err(|e| log::info!("{}", e))
+                .ok()?;
             let config = HighlightConfiguration::new(
                 language,
                 &highlights_query,
@@ -258,6 +287,18 @@ impl LanguageConfiguration {
             .as_ref()
     }
 
+    pub fn textobject_query(&self) -> Option<&TextObjectQuery> {
+        self.textobject_query
+            .get_or_init(|| -> Option<TextObjectQuery> {
+                let lang_name = self.language_id.to_ascii_lowercase();
+                let query_text = read_query(&lang_name, "textobjects.scm");
+                let lang = self.highlight_config.get()?.as_ref()?.language;
+                let query = Query::new(lang, &query_text).ok()?;
+                Some(TextObjectQuery { query })
+            })
+            .as_ref()
+    }
+
     pub fn scope(&self) -> &str {
         &self.scope
     }
@@ -451,7 +492,7 @@ impl Syntax {
 
     /// Iterate over the highlighted regions for a given slice of source code.
     pub fn highlight_iter<'a>(
-        &self,
+        &'a self,
         source: RopeSlice<'a>,
         range: Option<std::ops::Range<usize>>,
         cancellation_flag: Option<&'a AtomicUsize>,
@@ -466,11 +507,10 @@ impl Syntax {
             let highlighter = &mut ts_parser.borrow_mut();
             highlighter.cursors.pop().unwrap_or_else(QueryCursor::new)
         });
-        let tree_ref = unsafe { mem::transmute::<_, &'static Tree>(self.tree()) };
+        let tree_ref = self.tree();
         let cursor_ref = unsafe { mem::transmute::<_, &'static mut QueryCursor>(&mut cursor) };
-        let query_ref = unsafe { mem::transmute::<_, &'static Query>(&self.config.query) };
-        let config_ref =
-            unsafe { mem::transmute::<_, &'static HighlightConfiguration>(self.config.as_ref()) };
+        let query_ref = &self.config.query;
+        let config_ref = self.config.as_ref();
 
         // if reusing cursors & no range this resets to whole range
         cursor_ref.set_byte_range(range.clone().unwrap_or(0..usize::MAX));
@@ -582,39 +622,7 @@ impl LanguageLayer {
                     self.tree.as_ref(),
                 )
                 .ok_or(Error::Cancelled)?;
-            // unsafe { syntax.parser.set_cancellation_flag(None) };
-            // let mut cursor = syntax.cursors.pop().unwrap_or_else(QueryCursor::new);
-
-            // Process combined injections. (ERB, EJS, etc https://github.com/tree-sitter/tree-sitter/pull/526)
-            // if let Some(combined_injections_query) = &config.combined_injections_query {
-            //     let mut injections_by_pattern_index =
-            //         vec![(None, Vec::new(), false); combined_injections_query.pattern_count()];
-            //     let matches =
-            //         cursor.matches(combined_injections_query, tree.root_node(), RopeProvider(source));
-            //     for mat in matches {
-            //         let entry = &mut injections_by_pattern_index[mat.pattern_index];
-            //         let (language_name, content_node, include_children) =
-            //             injection_for_match(config, combined_injections_query, &mat, source);
-            //         if language_name.is_some() {
-            //             entry.0 = language_name;
-            //         }
-            //         if let Some(content_node) = content_node {
-            //             entry.1.push(content_node);
-            //         }
-            //         entry.2 = include_children;
-            //     }
-            //     for (lang_name, content_nodes, includes_children) in injections_by_pattern_index {
-            //         if let (Some(lang_name), false) = (lang_name, content_nodes.is_empty()) {
-            //             if let Some(next_config) = (injection_callback)(lang_name) {
-            //                 let ranges =
-            //                     Self::intersect_ranges(&ranges, &content_nodes, includes_children);
-            //                 if !ranges.is_empty() {
-            //                     queue.push((next_config, depth + 1, ranges));
-            //                 }
-            //             }
-            //         }
-            //     }
-            // }
+
             self.tree = Some(tree)
         }
         Ok(())
diff --git a/helix-core/src/textobject.rs b/helix-core/src/textobject.rs
index b965f6df..975ed115 100644
--- a/helix-core/src/textobject.rs
+++ b/helix-core/src/textobject.rs
@@ -1,9 +1,13 @@
+use std::fmt::Display;
+
 use ropey::RopeSlice;
+use tree_sitter::{Node, QueryCursor};
 
 use crate::chars::{categorize_char, char_is_whitespace, CharCategory};
 use crate::graphemes::next_grapheme_boundary;
 use crate::movement::Direction;
 use crate::surround;
+use crate::syntax::LanguageConfiguration;
 use crate::Range;
 
 fn find_word_boundary(slice: RopeSlice, mut pos: usize, direction: Direction) -> usize {
@@ -51,6 +55,15 @@ pub enum TextObject {
     Inside,
 }
 
+impl Display for TextObject {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.write_str(match self {
+            Self::Around => "around",
+            Self::Inside => "inside",
+        })
+    }
+}
+
 // count doesn't do anything yet
 pub fn textobject_word(
     slice: RopeSlice,
@@ -108,6 +121,44 @@ pub fn textobject_surround(
         .unwrap_or(range)
 }
 
+/// Transform the given range to select text objects based on tree-sitter.
+/// `object_name` is a query capture base name like "function", "class", etc.
+/// `slice_tree` is the tree-sitter node corresponding to given text slice.
+pub fn textobject_treesitter(
+    slice: RopeSlice,
+    range: Range,
+    textobject: TextObject,
+    object_name: &str,
+    slice_tree: Node,
+    lang_config: &LanguageConfiguration,
+    _count: usize,
+) -> Range {
+    let get_range = move || -> Option<Range> {
+        let byte_pos = slice.char_to_byte(range.cursor(slice));
+
+        let capture_name = format!("{}.{}", object_name, textobject); // eg. function.inner
+        let mut cursor = QueryCursor::new();
+        let node = lang_config
+            .textobject_query()?
+            .capture_nodes(&capture_name, slice_tree, slice, &mut cursor)?
+            .filter(|node| node.byte_range().contains(&byte_pos))
+            .min_by_key(|node| node.byte_range().len())?;
+
+        let len = slice.len_bytes();
+        let start_byte = node.start_byte();
+        let end_byte = node.end_byte();
+        if start_byte >= len || end_byte >= len {
+            return None;
+        }
+
+        let start_char = slice.byte_to_char(start_byte);
+        let end_char = slice.byte_to_char(end_byte);
+
+        Some(Range::new(start_char, end_char))
+    };
+    get_range().unwrap_or(range)
+}
+
 #[cfg(test)]
 mod test {
     use super::TextObject::*;
diff --git a/helix-core/src/transaction.rs b/helix-core/src/transaction.rs
index d682f058..dfc18fbe 100644
--- a/helix-core/src/transaction.rs
+++ b/helix-core/src/transaction.rs
@@ -132,6 +132,9 @@ impl ChangeSet {
         if self.changes.is_empty() {
             return other;
         }
+        if other.changes.is_empty() {
+            return self;
+        }
 
         let len = self.changes.len();
 
@@ -465,6 +468,13 @@ impl Transaction {
         }
     }
 
+    pub fn compose(mut self, other: Self) -> Self {
+        self.changes = self.changes.compose(other.changes);
+        // Other selection takes precedence
+        self.selection = other.selection;
+        self
+    }
+
     pub fn with_selection(mut self, selection: Selection) -> Self {
         self.selection = Some(selection);
         self