Communicate with language servers in terms of UTF-16 coordinates

This required indexing UTF-16 positions in `Rope`. We tried opting
into the UTF-8 experimental support but it didn't seem to work
correctly and the standard is UTF-16 anyway.

Co-Authored-By: Nathan Sobo <nathan@zed.dev>
This commit is contained in:
Antonio Scandurra 2021-10-29 18:31:21 +02:00
parent 9c1b01521a
commit 6212ebad9b
11 changed files with 564 additions and 221 deletions

View file

@ -1,7 +1,7 @@
[package]
name = "buffer"
version = "0.1.0"
edition = "2018"
edition = "2021"
[features]
test-support = ["rand", "seahash"]

View file

@ -1,6 +1,7 @@
mod anchor;
mod operation_queue;
mod point;
mod point_utf16;
#[cfg(any(test, feature = "test-support"))]
pub mod random_char_iter;
pub mod rope;
@ -13,8 +14,10 @@ use anyhow::{anyhow, Result};
use clock::ReplicaId;
use operation_queue::OperationQueue;
pub use point::*;
pub use point_utf16::*;
#[cfg(any(test, feature = "test-support"))]
pub use random_char_iter::*;
use rope::TextDimension;
pub use rope::{Chunks, Rope, TextSummary};
use rpc::proto;
pub use selection::*;
@ -309,41 +312,34 @@ impl UndoMap {
}
}
struct Edits<'a, F: FnMut(&FragmentSummary) -> bool> {
visible_text: &'a Rope,
deleted_text: &'a Rope,
cursor: Option<FilterCursor<'a, F, Fragment, FragmentTextSummary>>,
struct Edits<'a, D: TextDimension<'a>, F: FnMut(&FragmentSummary) -> bool> {
visible_cursor: rope::Cursor<'a>,
deleted_cursor: rope::Cursor<'a>,
fragments_cursor: Option<FilterCursor<'a, F, Fragment, FragmentTextSummary>>,
undos: &'a UndoMap,
since: clock::Global,
old_offset: usize,
new_offset: usize,
old_point: Point,
new_point: Point,
old_end: D,
new_end: D,
}
#[derive(Clone, Debug, Default, Eq, PartialEq)]
pub struct Edit {
pub old_bytes: Range<usize>,
pub new_bytes: Range<usize>,
pub old_lines: Range<Point>,
pub new_lines: Range<Point>,
pub struct Edit<D> {
pub old: Range<D>,
pub new: Range<D>,
}
impl Edit {
pub fn delta(&self) -> isize {
self.inserted_bytes() as isize - self.deleted_bytes() as isize
}
pub fn deleted_bytes(&self) -> usize {
self.old_bytes.end - self.old_bytes.start
}
pub fn inserted_bytes(&self) -> usize {
self.new_bytes.end - self.new_bytes.start
}
pub fn deleted_lines(&self) -> Point {
self.old_lines.end - self.old_lines.start
impl<D1, D2> Edit<(D1, D2)> {
pub fn flatten(self) -> (Edit<D1>, Edit<D2>) {
(
Edit {
old: self.old.start.0..self.old.end.0,
new: self.new.start.0..self.new.end.0,
},
Edit {
old: self.old.start.1..self.old.end.1,
new: self.new.start.1..self.new.end.1,
},
)
}
}
@ -1369,7 +1365,10 @@ impl Buffer {
})
}
pub fn edits_since<'a>(&'a self, since: clock::Global) -> impl 'a + Iterator<Item = Edit> {
pub fn edits_since<'a, D>(&'a self, since: clock::Global) -> impl 'a + Iterator<Item = Edit<D>>
where
D: 'a + TextDimension<'a> + Ord,
{
self.content().edits_since(since)
}
}
@ -1589,11 +1588,11 @@ impl Snapshot {
}
pub fn to_offset(&self, point: Point) -> usize {
self.visible_text.to_offset(point)
self.visible_text.point_to_offset(point)
}
pub fn to_point(&self, offset: usize) -> Point {
self.visible_text.to_point(offset)
self.visible_text.offset_to_point(offset)
}
pub fn anchor_before<T: ToOffset>(&self, position: T) -> Anchor {
@ -1604,7 +1603,10 @@ impl Snapshot {
self.content().anchor_at(position, Bias::Right)
}
pub fn edits_since<'a>(&'a self, since: clock::Global) -> impl 'a + Iterator<Item = Edit> {
pub fn edits_since<'a, D>(&'a self, since: clock::Global) -> impl 'a + Iterator<Item = Edit<D>>
where
D: 'a + TextDimension<'a> + Ord,
{
self.content().edits_since(since)
}
@ -1756,7 +1758,7 @@ impl<'a> Content<'a> {
} else {
0
};
summary += rope_cursor.summary(cursor.start().1 + overshoot);
summary += rope_cursor.summary::<TextSummary>(cursor.start().1 + overshoot);
(summary.clone(), value)
})
}
@ -1785,7 +1787,7 @@ impl<'a> Content<'a> {
} else {
0
};
summary += rope_cursor.summary(cursor.start().1 + overshoot);
summary += rope_cursor.summary::<TextSummary>(cursor.start().1 + overshoot);
let start_summary = summary.clone();
cursor.seek_forward(&VersionedFullOffset::Offset(*end_offset), *end_bias, &cx);
@ -1794,7 +1796,7 @@ impl<'a> Content<'a> {
} else {
0
};
summary += rope_cursor.summary(cursor.start().1 + overshoot);
summary += rope_cursor.summary::<TextSummary>(cursor.start().1 + overshoot);
let end_summary = summary.clone();
(start_summary..end_summary, value)
@ -1921,6 +1923,10 @@ impl<'a> Content<'a> {
self.visible_text.clip_point(point, bias)
}
pub fn clip_point_utf16(&self, point: PointUtf16, bias: Bias) -> PointUtf16 {
self.visible_text.clip_point_utf16(point, bias)
}
fn point_for_offset(&self, offset: usize) -> Result<Point> {
if offset <= self.len() {
Ok(self.text_summary_for_range(0..offset).lines)
@ -1930,9 +1936,12 @@ impl<'a> Content<'a> {
}
// TODO: take a reference to clock::Global.
pub fn edits_since(&self, since: clock::Global) -> impl 'a + Iterator<Item = Edit> {
pub fn edits_since<D>(&self, since: clock::Global) -> impl 'a + Iterator<Item = Edit<D>>
where
D: 'a + TextDimension<'a> + Ord,
{
let since_2 = since.clone();
let cursor = if since == *self.version {
let fragments_cursor = if since == *self.version {
None
} else {
Some(self.fragments.filter(
@ -1942,15 +1951,13 @@ impl<'a> Content<'a> {
};
Edits {
visible_text: &self.visible_text,
deleted_text: &self.deleted_text,
cursor,
visible_cursor: self.visible_text.cursor(0),
deleted_cursor: self.deleted_text.cursor(0),
fragments_cursor,
undos: &self.undo_map,
since,
old_offset: 0,
new_offset: 0,
old_point: Point::zero(),
new_point: Point::zero(),
old_end: Default::default(),
new_end: Default::default(),
}
}
}
@ -2008,70 +2015,61 @@ impl<'a> RopeBuilder<'a> {
}
}
impl<'a, F: FnMut(&FragmentSummary) -> bool> Iterator for Edits<'a, F> {
type Item = Edit;
impl<'a, D: TextDimension<'a> + Ord, F: FnMut(&FragmentSummary) -> bool> Iterator
for Edits<'a, D, F>
{
type Item = Edit<D>;
fn next(&mut self) -> Option<Self::Item> {
let mut change: Option<Edit> = None;
let cursor = self.cursor.as_mut()?;
let mut pending_edit: Option<Edit<D>> = None;
let cursor = self.fragments_cursor.as_mut()?;
while let Some(fragment) = cursor.item() {
let bytes = cursor.start().visible - self.new_offset;
let lines = self.visible_text.to_point(cursor.start().visible) - self.new_point;
self.old_offset += bytes;
self.old_point += &lines;
self.new_offset += bytes;
self.new_point += &lines;
let summary = self.visible_cursor.summary(cursor.start().visible);
self.old_end.add_assign(&summary);
self.new_end.add_assign(&summary);
if pending_edit
.as_ref()
.map_or(false, |change| change.new.end < self.new_end)
{
break;
}
if !fragment.was_visible(&self.since, &self.undos) && fragment.visible {
let fragment_lines =
self.visible_text.to_point(self.new_offset + fragment.len) - self.new_point;
if let Some(ref mut change) = change {
if change.new_bytes.end == self.new_offset {
change.new_bytes.end += fragment.len;
change.new_lines.end += fragment_lines;
} else {
break;
}
let fragment_summary = self.visible_cursor.summary(cursor.end(&None).visible);
let mut new_end = self.new_end.clone();
new_end.add_assign(&fragment_summary);
if let Some(pending_edit) = pending_edit.as_mut() {
pending_edit.new.end = new_end.clone();
} else {
change = Some(Edit {
old_bytes: self.old_offset..self.old_offset,
new_bytes: self.new_offset..self.new_offset + fragment.len,
old_lines: self.old_point..self.old_point,
new_lines: self.new_point..self.new_point + fragment_lines,
pending_edit = Some(Edit {
old: self.old_end.clone()..self.old_end.clone(),
new: self.new_end.clone()..new_end.clone(),
});
}
self.new_offset += fragment.len;
self.new_point += &fragment_lines;
self.new_end = new_end;
} else if fragment.was_visible(&self.since, &self.undos) && !fragment.visible {
let deleted_start = cursor.start().deleted;
let fragment_lines = self.deleted_text.to_point(deleted_start + fragment.len)
- self.deleted_text.to_point(deleted_start);
if let Some(ref mut change) = change {
if change.new_bytes.end == self.new_offset {
change.old_bytes.end += fragment.len;
change.old_lines.end += &fragment_lines;
} else {
break;
}
self.deleted_cursor.seek_forward(cursor.start().deleted);
let fragment_summary = self.deleted_cursor.summary(cursor.end(&None).deleted);
let mut old_end = self.old_end.clone();
old_end.add_assign(&fragment_summary);
if let Some(pending_edit) = pending_edit.as_mut() {
pending_edit.old.end = old_end.clone();
} else {
change = Some(Edit {
old_bytes: self.old_offset..self.old_offset + fragment.len,
new_bytes: self.new_offset..self.new_offset,
old_lines: self.old_point..self.old_point + &fragment_lines,
new_lines: self.new_point..self.new_point,
pending_edit = Some(Edit {
old: self.old_end.clone()..old_end.clone(),
new: self.new_end.clone()..self.new_end.clone(),
});
}
self.old_offset += fragment.len;
self.old_point += &fragment_lines;
self.old_end = old_end;
}
cursor.next(&None);
}
change
pending_edit
}
}
@ -2531,7 +2529,13 @@ pub trait ToOffset {
impl ToOffset for Point {
fn to_offset<'a>(&self, content: impl Into<Content<'a>>) -> usize {
content.into().visible_text.to_offset(*self)
content.into().visible_text.point_to_offset(*self)
}
}
impl ToOffset for PointUtf16 {
fn to_offset<'a>(&self, content: impl Into<Content<'a>>) -> usize {
content.into().visible_text.point_utf16_to_offset(*self)
}
}
@ -2566,7 +2570,7 @@ impl ToPoint for Anchor {
impl ToPoint for usize {
fn to_point<'a>(&self, content: impl Into<Content<'a>>) -> Point {
content.into().visible_text.to_point(*self)
content.into().visible_text.offset_to_point(*self)
}
}

View file

@ -32,6 +32,14 @@ impl<'a> Add<&'a Self> for Point {
type Output = Point;
fn add(self, other: &'a Self) -> Self::Output {
self + *other
}
}
impl Add for Point {
type Output = Point;
fn add(self, other: Self) -> Self::Output {
if other.row == 0 {
Point::new(self.row, self.column + other.column)
} else {
@ -40,25 +48,11 @@ impl<'a> Add<&'a Self> for Point {
}
}
impl Add for Point {
type Output = Point;
fn add(self, other: Self) -> Self::Output {
self + &other
}
}
impl<'a> Sub<&'a Self> for Point {
type Output = Point;
fn sub(self, other: &'a Self) -> Self::Output {
debug_assert!(*other <= self);
if self.row == other.row {
Point::new(0, self.column - other.column)
} else {
Point::new(self.row - other.row, self.column)
}
self - *other
}
}
@ -66,7 +60,13 @@ impl Sub for Point {
type Output = Point;
fn sub(self, other: Self) -> Self::Output {
self - &other
debug_assert!(other <= self);
if self.row == other.row {
Point::new(0, self.column - other.column)
} else {
Point::new(self.row - other.row, self.column)
}
}
}

View file

@ -0,0 +1,111 @@
use std::{
cmp::Ordering,
ops::{Add, AddAssign, Sub},
};
#[derive(Clone, Copy, Default, Eq, PartialEq, Debug, Hash)]
pub struct PointUtf16 {
pub row: u32,
pub column: u32,
}
impl PointUtf16 {
pub const MAX: Self = Self {
row: u32::MAX,
column: u32::MAX,
};
pub fn new(row: u32, column: u32) -> Self {
PointUtf16 { row, column }
}
pub fn zero() -> Self {
PointUtf16::new(0, 0)
}
pub fn is_zero(&self) -> bool {
self.row == 0 && self.column == 0
}
}
impl<'a> Add<&'a Self> for PointUtf16 {
type Output = PointUtf16;
fn add(self, other: &'a Self) -> Self::Output {
self + *other
}
}
impl Add for PointUtf16 {
type Output = PointUtf16;
fn add(self, other: Self) -> Self::Output {
if other.row == 0 {
PointUtf16::new(self.row, self.column + other.column)
} else {
PointUtf16::new(self.row + other.row, other.column)
}
}
}
impl<'a> Sub<&'a Self> for PointUtf16 {
type Output = PointUtf16;
fn sub(self, other: &'a Self) -> Self::Output {
self - *other
}
}
impl Sub for PointUtf16 {
type Output = PointUtf16;
fn sub(self, other: Self) -> Self::Output {
debug_assert!(other <= self);
if self.row == other.row {
PointUtf16::new(0, self.column - other.column)
} else {
PointUtf16::new(self.row - other.row, self.column)
}
}
}
impl<'a> AddAssign<&'a Self> for PointUtf16 {
fn add_assign(&mut self, other: &'a Self) {
*self += *other;
}
}
impl AddAssign<Self> for PointUtf16 {
fn add_assign(&mut self, other: Self) {
if other.row == 0 {
self.column += other.column;
} else {
self.row += other.row;
self.column = other.column;
}
}
}
impl PartialOrd for PointUtf16 {
fn partial_cmp(&self, other: &PointUtf16) -> Option<Ordering> {
Some(self.cmp(other))
}
}
impl Ord for PointUtf16 {
#[cfg(target_pointer_width = "64")]
fn cmp(&self, other: &PointUtf16) -> Ordering {
let a = (self.row as usize) << 32 | self.column as usize;
let b = (other.row as usize) << 32 | other.column as usize;
a.cmp(&b)
}
#[cfg(target_pointer_width = "32")]
fn cmp(&self, other: &PointUtf16) -> Ordering {
match self.row.cmp(&other.row) {
Ordering::Equal => self.column.cmp(&other.column),
comparison @ _ => comparison,
}
}
}

View file

@ -1,8 +1,10 @@
use crate::PointUtf16;
use super::Point;
use arrayvec::ArrayString;
use smallvec::SmallVec;
use std::{cmp, ops::Range, str};
use sum_tree::{Bias, SumTree};
use sum_tree::{Bias, Dimension, SumTree};
#[cfg(test)]
const CHUNK_BASE: usize = 6;
@ -136,7 +138,7 @@ impl Rope {
Chunks::new(self, range, true)
}
pub fn to_point(&self, offset: usize) -> Point {
pub fn offset_to_point(&self, offset: usize) -> Point {
assert!(offset <= self.summary().bytes);
let mut cursor = self.chunks.cursor::<(usize, Point)>();
cursor.seek(&offset, Bias::Left, &());
@ -144,15 +146,40 @@ impl Rope {
cursor.start().1
+ cursor
.item()
.map_or(Point::zero(), |chunk| chunk.to_point(overshoot))
.map_or(Point::zero(), |chunk| chunk.offset_to_point(overshoot))
}
pub fn to_offset(&self, point: Point) -> usize {
pub fn offset_to_point_utf16(&self, offset: usize) -> PointUtf16 {
assert!(offset <= self.summary().bytes);
let mut cursor = self.chunks.cursor::<(usize, PointUtf16)>();
cursor.seek(&offset, Bias::Left, &());
let overshoot = offset - cursor.start().0;
cursor.start().1
+ cursor.item().map_or(PointUtf16::zero(), |chunk| {
chunk.offset_to_point_utf16(overshoot)
})
}
pub fn point_to_offset(&self, point: Point) -> usize {
assert!(point <= self.summary().lines);
let mut cursor = self.chunks.cursor::<(Point, usize)>();
cursor.seek(&point, Bias::Left, &());
let overshoot = point - cursor.start().0;
cursor.start().1 + cursor.item().map_or(0, |chunk| chunk.to_offset(overshoot))
cursor.start().1
+ cursor
.item()
.map_or(0, |chunk| chunk.point_to_offset(overshoot))
}
pub fn point_utf16_to_offset(&self, point: PointUtf16) -> usize {
assert!(point <= self.summary().lines_utf16);
let mut cursor = self.chunks.cursor::<(PointUtf16, usize)>();
cursor.seek(&point, Bias::Left, &());
let overshoot = point - cursor.start().0;
cursor.start().1
+ cursor
.item()
.map_or(0, |chunk| chunk.point_utf16_to_offset(overshoot))
}
pub fn clip_offset(&self, mut offset: usize, bias: Bias) -> usize {
@ -188,6 +215,17 @@ impl Rope {
self.summary().lines
}
}
pub fn clip_point_utf16(&self, point: PointUtf16, bias: Bias) -> PointUtf16 {
let mut cursor = self.chunks.cursor::<PointUtf16>();
cursor.seek(&point, Bias::Right, &());
if let Some(chunk) = cursor.item() {
let overshoot = point - cursor.start();
*cursor.start() + chunk.clip_point_utf16(overshoot, bias)
} else {
self.summary().lines_utf16
}
}
}
impl<'a> From<&'a str> for Rope {
@ -258,22 +296,24 @@ impl<'a> Cursor<'a> {
slice
}
pub fn summary(&mut self, end_offset: usize) -> TextSummary {
pub fn summary<D: TextDimension<'a>>(&mut self, end_offset: usize) -> D {
debug_assert!(end_offset >= self.offset);
let mut summary = TextSummary::default();
let mut summary = D::default();
if let Some(start_chunk) = self.chunks.item() {
let start_ix = self.offset - self.chunks.start();
let end_ix = cmp::min(end_offset, self.chunks.end(&())) - self.chunks.start();
summary = TextSummary::from(&start_chunk.0[start_ix..end_ix]);
summary.add_assign(&D::from_summary(&TextSummary::from(
&start_chunk.0[start_ix..end_ix],
)));
}
if end_offset > self.chunks.end(&()) {
self.chunks.next(&());
summary += &self.chunks.summary(&end_offset, Bias::Right, &());
summary.add_assign(&self.chunks.summary(&end_offset, Bias::Right, &()));
if let Some(end_chunk) = self.chunks.item() {
let end_ix = end_offset - self.chunks.start();
summary += TextSummary::from(&end_chunk.0[..end_ix]);
summary.add_assign(&D::from_summary(&TextSummary::from(&end_chunk.0[..end_ix])));
}
}
@ -375,7 +415,7 @@ impl<'a> Iterator for Chunks<'a> {
struct Chunk(ArrayString<{ 2 * CHUNK_BASE }>);
impl Chunk {
fn to_point(&self, target: usize) -> Point {
fn offset_to_point(&self, target: usize) -> Point {
let mut offset = 0;
let mut point = Point::new(0, 0);
for ch in self.0.chars() {
@ -394,7 +434,26 @@ impl Chunk {
point
}
fn to_offset(&self, target: Point) -> usize {
fn offset_to_point_utf16(&self, target: usize) -> PointUtf16 {
let mut offset = 0;
let mut point = PointUtf16::new(0, 0);
for ch in self.0.chars() {
if offset >= target {
break;
}
if ch == '\n' {
point.row += 1;
point.column = 0;
} else {
point.column += ch.len_utf16() as u32;
}
offset += ch.len_utf8();
}
point
}
fn point_to_offset(&self, target: Point) -> usize {
let mut offset = 0;
let mut point = Point::new(0, 0);
for ch in self.0.chars() {
@ -416,6 +475,28 @@ impl Chunk {
offset
}
fn point_utf16_to_offset(&self, target: PointUtf16) -> usize {
let mut offset = 0;
let mut point = PointUtf16::new(0, 0);
for ch in self.0.chars() {
if point >= target {
if point > target {
panic!("point {:?} is inside of character {:?}", target, ch);
}
break;
}
if ch == '\n' {
point.row += 1;
point.column = 0;
} else {
point.column += ch.len_utf16() as u32;
}
offset += ch.len_utf8();
}
offset
}
fn clip_point(&self, target: Point, bias: Bias) -> Point {
for (row, line) in self.0.split('\n').enumerate() {
if row == target.row as usize {
@ -431,6 +512,23 @@ impl Chunk {
}
unreachable!()
}
fn clip_point_utf16(&self, target: PointUtf16, bias: Bias) -> PointUtf16 {
for (row, line) in self.0.split('\n').enumerate() {
if row == target.row as usize {
let mut code_units = line.encode_utf16();
let mut column = code_units.by_ref().take(target.column as usize).count();
if char::decode_utf16(code_units).next().transpose().is_err() {
match bias {
Bias::Left => column -= 1,
Bias::Right => column += 1,
}
}
return PointUtf16::new(row as u32, column as u32);
}
}
unreachable!()
}
}
impl sum_tree::Item for Chunk {
@ -445,6 +543,7 @@ impl sum_tree::Item for Chunk {
pub struct TextSummary {
pub bytes: usize,
pub lines: Point,
pub lines_utf16: PointUtf16,
pub first_line_chars: u32,
pub last_line_chars: u32,
pub longest_row: u32,
@ -454,17 +553,19 @@ pub struct TextSummary {
impl<'a> From<&'a str> for TextSummary {
fn from(text: &'a str) -> Self {
let mut lines = Point::new(0, 0);
let mut lines_utf16 = PointUtf16::new(0, 0);
let mut first_line_chars = 0;
let mut last_line_chars = 0;
let mut longest_row = 0;
let mut longest_row_chars = 0;
for c in text.chars() {
if c == '\n' {
lines.row += 1;
lines.column = 0;
lines += Point::new(1, 0);
lines_utf16 += PointUtf16::new(1, 0);
last_line_chars = 0;
} else {
lines.column += c.len_utf8() as u32;
lines_utf16.column += c.len_utf16() as u32;
last_line_chars += 1;
}
@ -481,6 +582,7 @@ impl<'a> From<&'a str> for TextSummary {
TextSummary {
bytes: text.len(),
lines,
lines_utf16,
first_line_chars,
last_line_chars,
longest_row,
@ -520,7 +622,8 @@ impl<'a> std::ops::AddAssign<&'a Self> for TextSummary {
}
self.bytes += other.bytes;
self.lines += &other.lines;
self.lines += other.lines;
self.lines_utf16 += other.lines_utf16;
}
}
@ -530,15 +633,77 @@ impl std::ops::AddAssign<Self> for TextSummary {
}
}
pub trait TextDimension<'a>: Dimension<'a, TextSummary> {
fn from_summary(summary: &TextSummary) -> Self;
fn add_assign(&mut self, other: &Self);
}
impl<'a, D1: TextDimension<'a>, D2: TextDimension<'a>> TextDimension<'a> for (D1, D2) {
fn from_summary(summary: &TextSummary) -> Self {
(D1::from_summary(summary), D2::from_summary(summary))
}
fn add_assign(&mut self, other: &Self) {
self.0.add_assign(&other.0);
self.1.add_assign(&other.1);
}
}
impl<'a> TextDimension<'a> for TextSummary {
fn from_summary(summary: &TextSummary) -> Self {
summary.clone()
}
fn add_assign(&mut self, other: &Self) {
*self += other;
}
}
impl<'a> sum_tree::Dimension<'a, TextSummary> for usize {
fn add_summary(&mut self, summary: &'a TextSummary, _: &()) {
*self += summary.bytes;
}
}
impl<'a> TextDimension<'a> for usize {
fn from_summary(summary: &TextSummary) -> Self {
summary.bytes
}
fn add_assign(&mut self, other: &Self) {
*self += other;
}
}
impl<'a> sum_tree::Dimension<'a, TextSummary> for Point {
fn add_summary(&mut self, summary: &'a TextSummary, _: &()) {
*self += &summary.lines;
*self += summary.lines;
}
}
impl<'a> TextDimension<'a> for Point {
fn from_summary(summary: &TextSummary) -> Self {
summary.lines
}
fn add_assign(&mut self, other: &Self) {
*self += other;
}
}
impl<'a> sum_tree::Dimension<'a, TextSummary> for PointUtf16 {
fn add_summary(&mut self, summary: &'a TextSummary, _: &()) {
*self += summary.lines_utf16;
}
}
impl<'a> TextDimension<'a> for PointUtf16 {
fn from_summary(summary: &TextSummary) -> Self {
summary.lines_utf16
}
fn add_assign(&mut self, other: &Self) {
*self += other;
}
}
@ -577,6 +742,41 @@ mod tests {
assert_eq!(rope.text(), text);
}
#[test]
fn test_clip() {
let rope = Rope::from("🧘");
assert_eq!(rope.clip_offset(1, Bias::Left), 0);
assert_eq!(rope.clip_offset(1, Bias::Right), 4);
assert_eq!(rope.clip_offset(5, Bias::Right), 4);
assert_eq!(
rope.clip_point(Point::new(0, 1), Bias::Left),
Point::new(0, 0)
);
assert_eq!(
rope.clip_point(Point::new(0, 1), Bias::Right),
Point::new(0, 4)
);
assert_eq!(
rope.clip_point(Point::new(0, 5), Bias::Right),
Point::new(0, 4)
);
assert_eq!(
rope.clip_point_utf16(PointUtf16::new(0, 1), Bias::Left),
PointUtf16::new(0, 0)
);
assert_eq!(
rope.clip_point_utf16(PointUtf16::new(0, 1), Bias::Right),
PointUtf16::new(0, 2)
);
assert_eq!(
rope.clip_point_utf16(PointUtf16::new(0, 3), Bias::Right),
PointUtf16::new(0, 2)
);
}
#[gpui::test(iterations = 100)]
fn test_random(mut rng: StdRng) {
let operations = env::var("OPERATIONS")
@ -624,14 +824,33 @@ mod tests {
}
let mut point = Point::new(0, 0);
let mut point_utf16 = PointUtf16::new(0, 0);
for (ix, ch) in expected.char_indices().chain(Some((expected.len(), '\0'))) {
assert_eq!(actual.to_point(ix), point, "to_point({})", ix);
assert_eq!(actual.to_offset(point), ix, "to_offset({:?})", point);
assert_eq!(actual.offset_to_point(ix), point, "offset_to_point({})", ix);
assert_eq!(
actual.offset_to_point_utf16(ix),
point_utf16,
"offset_to_point_utf16({})",
ix
);
assert_eq!(
actual.point_to_offset(point),
ix,
"point_to_offset({:?})",
point
);
assert_eq!(
actual.point_utf16_to_offset(point_utf16),
ix,
"point_utf16_to_offset({:?})",
point_utf16
);
if ch == '\n' {
point.row += 1;
point.column = 0
point += Point::new(1, 0);
point_utf16 += PointUtf16::new(1, 0);
} else {
point.column += ch.len_utf8() as u32;
point_utf16.column += ch.len_utf16() as u32;
}
}
@ -639,7 +858,7 @@ mod tests {
let end_ix = clip_offset(&expected, rng.gen_range(0..=expected.len()), Right);
let start_ix = clip_offset(&expected, rng.gen_range(0..=end_ix), Left);
assert_eq!(
actual.cursor(start_ix).summary(end_ix),
actual.cursor(start_ix).summary::<TextSummary>(end_ix),
TextSummary::from(&expected[start_ix..end_ix])
);
}

View file

@ -78,7 +78,7 @@ fn test_random_edits(mut rng: StdRng) {
for mut old_buffer in buffer_versions {
let edits = buffer
.edits_since(old_buffer.version.clone())
.edits_since::<usize>(old_buffer.version.clone())
.collect::<Vec<_>>();
log::info!(
@ -88,12 +88,12 @@ fn test_random_edits(mut rng: StdRng) {
edits,
);
let mut delta = 0_isize;
for edit in edits {
let old_start = (edit.old_bytes.start as isize + delta) as usize;
let new_text: String = buffer.text_for_range(edit.new_bytes.clone()).collect();
old_buffer.edit(Some(old_start..old_start + edit.deleted_bytes()), new_text);
delta += edit.delta();
let new_text: String = buffer.text_for_range(edit.new.clone()).collect();
old_buffer.edit(
Some(edit.new.start..edit.new.start + edit.old.len()),
new_text,
);
}
assert_eq!(old_buffer.text(), buffer.text());
}
@ -123,6 +123,7 @@ fn test_text_summary_for_range() {
TextSummary {
bytes: 2,
lines: Point::new(1, 0),
lines_utf16: PointUtf16::new(1, 0),
first_line_chars: 1,
last_line_chars: 0,
longest_row: 0,
@ -134,6 +135,7 @@ fn test_text_summary_for_range() {
TextSummary {
bytes: 11,
lines: Point::new(3, 0),
lines_utf16: PointUtf16::new(3, 0),
first_line_chars: 1,
last_line_chars: 0,
longest_row: 2,
@ -145,6 +147,7 @@ fn test_text_summary_for_range() {
TextSummary {
bytes: 20,
lines: Point::new(4, 1),
lines_utf16: PointUtf16::new(4, 1),
first_line_chars: 2,
last_line_chars: 1,
longest_row: 3,
@ -156,6 +159,7 @@ fn test_text_summary_for_range() {
TextSummary {
bytes: 22,
lines: Point::new(4, 3),
lines_utf16: PointUtf16::new(4, 3),
first_line_chars: 2,
last_line_chars: 3,
longest_row: 3,
@ -167,6 +171,7 @@ fn test_text_summary_for_range() {
TextSummary {
bytes: 15,
lines: Point::new(2, 3),
lines_utf16: PointUtf16::new(2, 3),
first_line_chars: 4,
last_line_chars: 3,
longest_row: 1,

View file

@ -1,6 +1,7 @@
use gpui::{AppContext, ModelHandle};
use language::{
Anchor, AnchorRangeExt, Buffer, HighlightId, HighlightedChunk, Point, TextSummary, ToOffset,
Anchor, AnchorRangeExt, Buffer, HighlightId, HighlightedChunk, Point, PointUtf16, TextSummary,
ToOffset,
};
use parking_lot::Mutex;
use std::{
@ -112,9 +113,8 @@ impl<'a> FoldMapWriter<'a> {
let fold = Fold(buffer.anchor_after(range.start)..buffer.anchor_before(range.end));
folds.push(fold);
edits.push(buffer::Edit {
old_bytes: range.clone(),
new_bytes: range.clone(),
..Default::default()
old: range.clone(),
new: range,
});
}
}
@ -157,9 +157,8 @@ impl<'a> FoldMapWriter<'a> {
while let Some(fold) = folds_cursor.item() {
let offset_range = fold.0.start.to_offset(&buffer)..fold.0.end.to_offset(&buffer);
edits.push(buffer::Edit {
old_bytes: offset_range.clone(),
new_bytes: offset_range,
..Default::default()
old: offset_range.clone(),
new: offset_range,
});
fold_ixs_to_delete.push(*folds_cursor.start());
folds_cursor.next(&buffer);
@ -288,7 +287,11 @@ impl FoldMap {
}
}
fn apply_edits(&self, buffer_edits: Vec<buffer::Edit>, cx: &AppContext) -> Vec<FoldEdit> {
fn apply_edits(
&self,
buffer_edits: Vec<buffer::Edit<usize>>,
cx: &AppContext,
) -> Vec<FoldEdit> {
let buffer = self.buffer.read(cx).snapshot();
let mut buffer_edits_iter = buffer_edits.iter().cloned().peekable();
@ -298,28 +301,28 @@ impl FoldMap {
cursor.seek(&0, Bias::Right, &());
while let Some(mut edit) = buffer_edits_iter.next() {
new_transforms.push_tree(cursor.slice(&edit.old_bytes.start, Bias::Left, &()), &());
edit.new_bytes.start -= edit.old_bytes.start - cursor.start();
edit.old_bytes.start = *cursor.start();
new_transforms.push_tree(cursor.slice(&edit.old.start, Bias::Left, &()), &());
edit.new.start -= edit.old.start - cursor.start();
edit.old.start = *cursor.start();
cursor.seek(&edit.old_bytes.end, Bias::Right, &());
cursor.seek(&edit.old.end, Bias::Right, &());
cursor.next(&());
let mut delta = edit.delta();
let mut delta = edit.new.len() as isize - edit.old.len() as isize;
loop {
edit.old_bytes.end = *cursor.start();
edit.old.end = *cursor.start();
if let Some(next_edit) = buffer_edits_iter.peek() {
if next_edit.old_bytes.start > edit.old_bytes.end {
if next_edit.old.start > edit.old.end {
break;
}
let next_edit = buffer_edits_iter.next().unwrap();
delta += next_edit.delta();
delta += next_edit.new.len() as isize - next_edit.old.len() as isize;
if next_edit.old_bytes.end >= edit.old_bytes.end {
edit.old_bytes.end = next_edit.old_bytes.end;
cursor.seek(&edit.old_bytes.end, Bias::Right, &());
if next_edit.old.end >= edit.old.end {
edit.old.end = next_edit.old.end;
cursor.seek(&edit.old.end, Bias::Right, &());
cursor.next(&());
}
} else {
@ -327,10 +330,9 @@ impl FoldMap {
}
}
edit.new_bytes.end =
((edit.new_bytes.start + edit.deleted_bytes()) as isize + delta) as usize;
edit.new.end = ((edit.new.start + edit.old.len()) as isize + delta) as usize;
let anchor = buffer.anchor_before(edit.new_bytes.start);
let anchor = buffer.anchor_before(edit.new.start);
let mut folds_cursor = self.folds.cursor::<Fold>();
folds_cursor.seek(&Fold(anchor..Anchor::max()), Bias::Left, &buffer);
@ -346,10 +348,7 @@ impl FoldMap {
})
.peekable();
while folds
.peek()
.map_or(false, |fold| fold.start < edit.new_bytes.end)
{
while folds.peek().map_or(false, |fold| fold.start < edit.new.end) {
let mut fold = folds.next().unwrap();
let sum = new_transforms.summary();
@ -382,13 +381,15 @@ impl FoldMap {
if fold.end > fold.start {
let output_text = "";
let chars = output_text.chars().count() as u32;
let lines = super::Point::new(0, output_text.len() as u32);
let lines = Point::new(0, output_text.len() as u32);
let lines_utf16 = PointUtf16::new(0, output_text.encode_utf16().count() as u32);
new_transforms.push(
Transform {
summary: TransformSummary {
output: TextSummary {
bytes: output_text.len(),
lines,
lines_utf16,
first_line_chars: chars,
last_line_chars: chars,
longest_row: 0,
@ -404,9 +405,8 @@ impl FoldMap {
}
let sum = new_transforms.summary();
if sum.input.bytes < edit.new_bytes.end {
let text_summary =
buffer.text_summary_for_range(sum.input.bytes..edit.new_bytes.end);
if sum.input.bytes < edit.new.end {
let text_summary = buffer.text_summary_for_range(sum.input.bytes..edit.new.end);
new_transforms.push(
Transform {
summary: TransformSummary {
@ -443,35 +443,35 @@ impl FoldMap {
let mut new_transforms = new_transforms.cursor::<(usize, FoldOffset)>();
for mut edit in buffer_edits {
old_transforms.seek(&edit.old_bytes.start, Bias::Left, &());
old_transforms.seek(&edit.old.start, Bias::Left, &());
if old_transforms.item().map_or(false, |t| t.is_fold()) {
edit.old_bytes.start = old_transforms.start().0;
edit.old.start = old_transforms.start().0;
}
let old_start =
old_transforms.start().1 .0 + (edit.old_bytes.start - old_transforms.start().0);
old_transforms.start().1 .0 + (edit.old.start - old_transforms.start().0);
old_transforms.seek_forward(&edit.old_bytes.end, Bias::Right, &());
old_transforms.seek_forward(&edit.old.end, Bias::Right, &());
if old_transforms.item().map_or(false, |t| t.is_fold()) {
old_transforms.next(&());
edit.old_bytes.end = old_transforms.start().0;
edit.old.end = old_transforms.start().0;
}
let old_end =
old_transforms.start().1 .0 + (edit.old_bytes.end - old_transforms.start().0);
old_transforms.start().1 .0 + (edit.old.end - old_transforms.start().0);
new_transforms.seek(&edit.new_bytes.start, Bias::Left, &());
new_transforms.seek(&edit.new.start, Bias::Left, &());
if new_transforms.item().map_or(false, |t| t.is_fold()) {
edit.new_bytes.start = new_transforms.start().0;
edit.new.start = new_transforms.start().0;
}
let new_start =
new_transforms.start().1 .0 + (edit.new_bytes.start - new_transforms.start().0);
new_transforms.start().1 .0 + (edit.new.start - new_transforms.start().0);
new_transforms.seek_forward(&edit.new_bytes.end, Bias::Right, &());
new_transforms.seek_forward(&edit.new.end, Bias::Right, &());
if new_transforms.item().map_or(false, |t| t.is_fold()) {
new_transforms.next(&());
edit.new_bytes.end = new_transforms.start().0;
edit.new.end = new_transforms.start().0;
}
let new_end =
new_transforms.start().1 .0 + (edit.new_bytes.end - new_transforms.start().0);
new_transforms.start().1 .0 + (edit.new.end - new_transforms.start().0);
fold_edits.push(FoldEdit {
old_bytes: FoldOffset(old_start)..FoldOffset(old_end),
@ -748,22 +748,22 @@ where
)
}
fn consolidate_buffer_edits(edits: &mut Vec<buffer::Edit>) {
fn consolidate_buffer_edits(edits: &mut Vec<buffer::Edit<usize>>) {
edits.sort_unstable_by(|a, b| {
a.old_bytes
a.old
.start
.cmp(&b.old_bytes.start)
.then_with(|| b.old_bytes.end.cmp(&a.old_bytes.end))
.cmp(&b.old.start)
.then_with(|| b.old.end.cmp(&a.old.end))
});
let mut i = 1;
while i < edits.len() {
let edit = edits[i].clone();
let prev_edit = &mut edits[i - 1];
if prev_edit.old_bytes.end >= edit.old_bytes.start {
prev_edit.old_bytes.end = prev_edit.old_bytes.end.max(edit.old_bytes.end);
prev_edit.new_bytes.start = prev_edit.new_bytes.start.min(edit.new_bytes.start);
prev_edit.new_bytes.end = prev_edit.new_bytes.end.max(edit.new_bytes.end);
if prev_edit.old.end >= edit.old.start {
prev_edit.old.end = prev_edit.old.end.max(edit.old.end);
prev_edit.new.start = prev_edit.new.start.min(edit.new.start);
prev_edit.new.end = prev_edit.new.end.max(edit.new.end);
edits.remove(i);
continue;
}
@ -1343,7 +1343,9 @@ mod tests {
let start_version = buffer.version.clone();
let edit_count = rng.gen_range(1..=5);
buffer.randomly_edit(&mut rng, edit_count);
buffer.edits_since(start_version).collect::<Vec<_>>()
buffer
.edits_since::<Point>(start_version)
.collect::<Vec<_>>()
});
log::info!("editing {:?}", edits);
}

View file

@ -359,13 +359,19 @@ impl Buffer {
),
content_changes: snapshot
.buffer_snapshot
.edits_since(
.edits_since::<(PointUtf16, usize)>(
prev_snapshot.buffer_snapshot.version().clone(),
)
.map(|edit| {
// TODO: Use UTF-16 positions.
let edit_start = edit.new_lines.start;
let edit_end = edit_start + edit.deleted_lines();
let edit_start = edit.new.start.0;
let edit_end = edit_start
+ (edit.old.end.0 - edit.old.start.0);
let new_text = snapshot
.buffer_snapshot
.text_for_range(
edit.new.start.1..edit.new.end.1,
)
.collect();
lsp::TextDocumentContentChangeEvent {
range: Some(lsp::Range::new(
lsp::Position::new(
@ -378,10 +384,7 @@ impl Buffer {
),
)),
range_length: None,
text: snapshot
.buffer_snapshot
.text_for_range(edit.new_bytes)
.collect(),
text: new_text,
}
})
.collect(),
@ -613,22 +616,17 @@ impl Buffer {
}
fn interpolate_tree(&self, tree: &mut SyntaxTree) {
let mut delta = 0_isize;
for edit in self.edits_since(tree.version.clone()) {
let start_offset = (edit.old_bytes.start as isize + delta) as usize;
let start_point = self.as_rope().to_point(start_offset);
for edit in self.edits_since::<(usize, Point)>(tree.version.clone()) {
let (bytes, lines) = edit.flatten();
tree.tree.edit(&InputEdit {
start_byte: start_offset,
old_end_byte: start_offset + edit.deleted_bytes(),
new_end_byte: start_offset + edit.inserted_bytes(),
start_position: start_point.to_ts_point(),
old_end_position: (start_point + edit.deleted_lines()).to_ts_point(),
new_end_position: self
.as_rope()
.to_point(start_offset + edit.inserted_bytes())
start_byte: bytes.new.start,
old_end_byte: bytes.new.start + bytes.old.len(),
new_end_byte: bytes.new.end,
start_position: lines.new.start.to_ts_point(),
old_end_position: (lines.new.start + (lines.old.end - lines.old.start))
.to_ts_point(),
new_end_position: lines.new.end.to_ts_point(),
});
delta += edit.inserted_bytes() as isize - edit.deleted_bytes() as isize;
}
tree.version = self.version();
}
@ -673,21 +671,22 @@ impl Buffer {
diagnostics.sort_unstable_by_key(|d| (d.range.start, d.range.end));
self.diagnostics = {
let mut edits_since_save = content.edits_since(self.saved_version.clone()).peekable();
let mut last_edit_old_end = Point::zero();
let mut last_edit_new_end = Point::zero();
let mut edits_since_save = content
.edits_since::<PointUtf16>(self.saved_version.clone())
.peekable();
let mut last_edit_old_end = PointUtf16::zero();
let mut last_edit_new_end = PointUtf16::zero();
content.anchor_range_multimap(
Bias::Left,
Bias::Right,
diagnostics.into_iter().filter_map(|diagnostic| {
// TODO: Use UTF-16 positions.
let mut start = Point::new(
let mut start = PointUtf16::new(
diagnostic.range.start.line,
diagnostic.range.start.character,
);
let mut end =
Point::new(diagnostic.range.end.line, diagnostic.range.end.character);
PointUtf16::new(diagnostic.range.end.line, diagnostic.range.end.character);
let severity = diagnostic.severity.unwrap_or(DiagnosticSeverity::ERROR);
if diagnostic
@ -696,11 +695,11 @@ impl Buffer {
.map_or(false, |source| disk_based_sources.contains(source))
{
while let Some(edit) = edits_since_save.peek() {
if edit.old_lines.end <= start {
last_edit_old_end = edit.old_lines.end;
last_edit_new_end = edit.new_lines.end;
if edit.old.end <= start {
last_edit_old_end = edit.old.end;
last_edit_new_end = edit.new.end;
edits_since_save.next();
} else if edit.old_lines.start <= end && edit.old_lines.end >= start {
} else if edit.old.start <= end && edit.old.end >= start {
return None;
} else {
break;
@ -711,8 +710,8 @@ impl Buffer {
end = last_edit_new_end + (end - last_edit_old_end);
}
let range =
content.clip_point(start, Bias::Left)..content.clip_point(end, Bias::Right);
let range = content.clip_point_utf16(start, Bias::Left)
..content.clip_point_utf16(end, Bias::Right);
Some((range, (severity, diagnostic.message)))
}),
)
@ -1223,7 +1222,7 @@ impl Buffer {
was_dirty: bool,
cx: &mut ModelContext<Self>,
) {
if self.edits_since(old_version).next().is_none() {
if self.edits_since::<usize>(old_version).next().is_none() {
return;
}

View file

@ -13,7 +13,7 @@ anyhow = "1.0"
async-pipe = { git = "https://github.com/routerify/async-pipe-rs", rev = "feeb77e83142a9ff837d0767652ae41bfc5d8e47", optional = true }
futures = "0.3"
log = "0.4"
lsp-types = { version = "0.91", features = ["proposed"] }
lsp-types = "0.91"
parking_lot = "0.11"
postage = { version = "0.4.1", features = ["futures-traits"] }
serde = { version = "1.0", features = ["derive"] }

View file

@ -226,7 +226,6 @@ impl LanguageServer {
experimental: Some(json!({
"serverStatusNotification": true,
})),
offset_encoding: Some(vec!["utf-8".to_string()]),
..Default::default()
},
trace: Default::default(),

View file

@ -537,6 +537,10 @@ where
self.cursor.start()
}
pub fn end(&self, cx: &<T::Summary as Summary>::Context) -> D {
self.cursor.end(cx)
}
pub fn item(&self) -> Option<&'a T> {
self.cursor.item()
}