Merge pull request #651 from zed-industries/refine-word-movement

Add sub-word movement and helper functions for finding word boundaries
This commit is contained in:
Max Brunsfeld
2022-03-22 09:09:03 -07:00
committed by GitHub
5 changed files with 330 additions and 181 deletions
+12 -3
View File
@@ -12,7 +12,7 @@ use gpui::{
Entity, ModelContext, ModelHandle,
};
use language::{Point, Subscription as BufferSubscription};
use std::{any::TypeId, ops::Range, sync::Arc};
use std::{any::TypeId, fmt::Debug, ops::Range, sync::Arc};
use sum_tree::{Bias, TreeMap};
use tab_map::TabMap;
use wrap_map::WrapMap;
@@ -414,9 +414,19 @@ impl DisplaySnapshot {
}
}
#[derive(Copy, Clone, Debug, Default, Eq, Ord, PartialOrd, PartialEq)]
#[derive(Copy, Clone, Default, Eq, Ord, PartialOrd, PartialEq)]
pub struct DisplayPoint(BlockPoint);
impl Debug for DisplayPoint {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
f.write_fmt(format_args!(
"DisplayPoint({}, {})",
self.row(),
self.column()
))
}
}
impl DisplayPoint {
pub fn new(row: u32, column: u32) -> Self {
Self(BlockPoint(Point::new(row, column)))
@@ -426,7 +436,6 @@ impl DisplayPoint {
Self::new(0, 0)
}
#[cfg(test)]
pub fn is_zero(&self) -> bool {
self.0.is_zero()
}
+6 -7
View File
@@ -3480,7 +3480,7 @@ impl Editor {
let mut selections = self.local_selections::<Point>(cx);
for selection in &mut selections {
let head = selection.head().to_display_point(&display_map);
let cursor = movement::prev_word_boundary(&display_map, head).to_point(&display_map);
let cursor = movement::previous_word_start(&display_map, head).to_point(&display_map);
selection.start = cursor.clone();
selection.end = cursor;
selection.reversed = false;
@@ -3498,7 +3498,7 @@ impl Editor {
let mut selections = self.local_selections::<Point>(cx);
for selection in &mut selections {
let head = selection.head().to_display_point(&display_map);
let cursor = movement::prev_word_boundary(&display_map, head).to_point(&display_map);
let cursor = movement::previous_word_start(&display_map, head).to_point(&display_map);
selection.set_head(cursor);
selection.goal = SelectionGoal::None;
}
@@ -3517,7 +3517,7 @@ impl Editor {
if selection.is_empty() {
let head = selection.head().to_display_point(&display_map);
let cursor =
movement::prev_word_boundary(&display_map, head).to_point(&display_map);
movement::previous_word_start(&display_map, head).to_point(&display_map);
selection.set_head(cursor);
selection.goal = SelectionGoal::None;
}
@@ -3536,7 +3536,7 @@ impl Editor {
let mut selections = self.local_selections::<Point>(cx);
for selection in &mut selections {
let head = selection.head().to_display_point(&display_map);
let cursor = movement::next_word_boundary(&display_map, head).to_point(&display_map);
let cursor = movement::next_word_end(&display_map, head).to_point(&display_map);
selection.start = cursor;
selection.end = cursor;
selection.reversed = false;
@@ -3554,7 +3554,7 @@ impl Editor {
let mut selections = self.local_selections::<Point>(cx);
for selection in &mut selections {
let head = selection.head().to_display_point(&display_map);
let cursor = movement::next_word_boundary(&display_map, head).to_point(&display_map);
let cursor = movement::next_word_end(&display_map, head).to_point(&display_map);
selection.set_head(cursor);
selection.goal = SelectionGoal::None;
}
@@ -3572,8 +3572,7 @@ impl Editor {
for selection in &mut selections {
if selection.is_empty() {
let head = selection.head().to_display_point(&display_map);
let cursor =
movement::next_word_boundary(&display_map, head).to_point(&display_map);
let cursor = movement::next_word_end(&display_map, head).to_point(&display_map);
selection.set_head(cursor);
selection.goal = SelectionGoal::None;
}
+309 -165
View File
@@ -132,68 +132,110 @@ pub fn line_end(
}
}
pub fn prev_word_boundary(map: &DisplaySnapshot, mut point: DisplayPoint) -> DisplayPoint {
let mut line_start = 0;
if point.row() > 0 {
if let Some(indent) = map.soft_wrap_indent(point.row() - 1) {
line_start = indent;
}
}
if point.column() == line_start {
if point.row() == 0 {
return DisplayPoint::new(0, 0);
} else {
let row = point.row() - 1;
point = map.clip_point(DisplayPoint::new(row, map.line_len(row)), Bias::Left);
}
}
let mut boundary = DisplayPoint::new(point.row(), 0);
let mut column = 0;
let mut prev_char_kind = CharKind::Newline;
for c in map.chars_at(DisplayPoint::new(point.row(), 0)) {
if column >= point.column() {
break;
}
let char_kind = char_kind(c);
if char_kind != prev_char_kind
&& char_kind != CharKind::Whitespace
&& char_kind != CharKind::Newline
{
*boundary.column_mut() = column;
}
prev_char_kind = char_kind;
column += c.len_utf8() as u32;
}
boundary
pub fn previous_word_start(map: &DisplaySnapshot, point: DisplayPoint) -> DisplayPoint {
find_preceding_boundary(map, point, |left, right| {
(char_kind(left) != char_kind(right) && !right.is_whitespace()) || left == '\n'
})
}
pub fn next_word_boundary(map: &DisplaySnapshot, mut point: DisplayPoint) -> DisplayPoint {
let mut prev_char_kind = None;
for c in map.chars_at(point) {
let char_kind = char_kind(c);
if let Some(prev_char_kind) = prev_char_kind {
if c == '\n' {
pub fn previous_subword_start(map: &DisplaySnapshot, point: DisplayPoint) -> DisplayPoint {
find_preceding_boundary(map, point, |left, right| {
let is_word_start = char_kind(left) != char_kind(right) && !right.is_whitespace();
let is_subword_start =
left == '_' && right != '_' || left.is_lowercase() && right.is_uppercase();
is_word_start || is_subword_start || left == '\n'
})
}
pub fn next_word_end(map: &DisplaySnapshot, point: DisplayPoint) -> DisplayPoint {
find_boundary(map, point, |left, right| {
(char_kind(left) != char_kind(right) && !left.is_whitespace()) || right == '\n'
})
}
pub fn next_subword_end(map: &DisplaySnapshot, point: DisplayPoint) -> DisplayPoint {
find_boundary(map, point, |left, right| {
let is_word_end = (char_kind(left) != char_kind(right)) && !left.is_whitespace();
let is_subword_end =
left != '_' && right == '_' || left.is_lowercase() && right.is_uppercase();
is_word_end || is_subword_end || right == '\n'
})
}
/// Scans for a boundary from the start of each line preceding the given end point until a boundary
/// is found, indicated by the given predicate returning true. The predicate is called with the
/// character to the left and right of the candidate boundary location, and will be called with `\n`
/// characters indicating the start or end of a line. If the predicate returns true multiple times
/// on a line, the *rightmost* boundary is returned.
pub fn find_preceding_boundary(
map: &DisplaySnapshot,
end: DisplayPoint,
mut is_boundary: impl FnMut(char, char) -> bool,
) -> DisplayPoint {
let mut point = end;
loop {
*point.column_mut() = 0;
if point.row() > 0 {
if let Some(indent) = map.soft_wrap_indent(point.row() - 1) {
*point.column_mut() = indent;
}
}
let mut boundary = None;
let mut prev_ch = if point.is_zero() { None } else { Some('\n') };
for ch in map.chars_at(point) {
if point >= end {
break;
}
if prev_char_kind != char_kind
&& prev_char_kind != CharKind::Whitespace
&& prev_char_kind != CharKind::Newline
{
if let Some(prev_ch) = prev_ch {
if is_boundary(prev_ch, ch) {
boundary = Some(point);
}
}
if ch == '\n' {
break;
}
prev_ch = Some(ch);
*point.column_mut() += ch.len_utf8() as u32;
}
if let Some(boundary) = boundary {
return boundary;
} else if point.row() == 0 {
return DisplayPoint::zero();
} else {
*point.row_mut() -= 1;
}
}
}
/// Scans for a boundary following the given start point until a boundary is found, indicated by the
/// given predicate returning true. The predicate is called with the character to the left and right
/// of the candidate boundary location, and will be called with `\n` characters indicating the start
/// or end of a line.
pub fn find_boundary(
map: &DisplaySnapshot,
mut point: DisplayPoint,
mut is_boundary: impl FnMut(char, char) -> bool,
) -> DisplayPoint {
let mut prev_ch = None;
for ch in map.chars_at(point) {
if let Some(prev_ch) = prev_ch {
if is_boundary(prev_ch, ch) {
break;
}
}
if c == '\n' {
if ch == '\n' {
*point.row_mut() += 1;
*point.column_mut() = 0;
} else {
*point.column_mut() += c.len_utf8() as u32;
*point.column_mut() += ch.len_utf8() as u32;
}
prev_char_kind = Some(char_kind);
prev_ch = Some(ch);
}
map.clip_point(point, Bias::Right)
}
@@ -228,6 +270,202 @@ mod tests {
use crate::{Buffer, DisplayMap, MultiBuffer};
use language::Point;
#[gpui::test]
fn test_previous_word_start(cx: &mut gpui::MutableAppContext) {
fn assert(marked_text: &str, cx: &mut gpui::MutableAppContext) {
let (snapshot, display_points) = marked_snapshot(marked_text, cx);
assert_eq!(
previous_word_start(&snapshot, display_points[1]),
display_points[0]
);
}
assert("\n| |lorem", cx);
assert("|\n| lorem", cx);
assert(" |lorem|", cx);
assert("| |lorem", cx);
assert(" |lor|em", cx);
assert("\nlorem\n| |ipsum", cx);
assert("\n\n|\n|", cx);
assert(" |lorem |ipsum", cx);
assert("lorem|-|ipsum", cx);
assert("lorem|-#$@|ipsum", cx);
assert("|lorem_|ipsum", cx);
assert(" |defγ|", cx);
assert(" |bcΔ|", cx);
assert(" ab|——|cd", cx);
}
#[gpui::test]
fn test_previous_subword_start(cx: &mut gpui::MutableAppContext) {
fn assert(marked_text: &str, cx: &mut gpui::MutableAppContext) {
let (snapshot, display_points) = marked_snapshot(marked_text, cx);
assert_eq!(
previous_subword_start(&snapshot, display_points[1]),
display_points[0]
);
}
// Subword boundaries are respected
assert("lorem_|ip|sum", cx);
assert("lorem_|ipsum|", cx);
assert("|lorem_|ipsum", cx);
assert("lorem_|ipsum_|dolor", cx);
assert("lorem|Ip|sum", cx);
assert("lorem|Ipsum|", cx);
// Word boundaries are still respected
assert("\n| |lorem", cx);
assert(" |lorem|", cx);
assert(" |lor|em", cx);
assert("\nlorem\n| |ipsum", cx);
assert("\n\n|\n|", cx);
assert(" |lorem |ipsum", cx);
assert("lorem|-|ipsum", cx);
assert("lorem|-#$@|ipsum", cx);
assert(" |defγ|", cx);
assert(" bc|Δ|", cx);
assert(" |bcδ|", cx);
assert(" ab|——|cd", cx);
}
#[gpui::test]
fn test_find_preceding_boundary(cx: &mut gpui::MutableAppContext) {
fn assert(
marked_text: &str,
cx: &mut gpui::MutableAppContext,
is_boundary: impl FnMut(char, char) -> bool,
) {
let (snapshot, display_points) = marked_snapshot(marked_text, cx);
assert_eq!(
find_preceding_boundary(&snapshot, display_points[1], is_boundary),
display_points[0]
);
}
assert("abc|def\ngh\nij|k", cx, |left, right| {
left == 'c' && right == 'd'
});
assert("abcdef\n|gh\nij|k", cx, |left, right| {
left == '\n' && right == 'g'
});
let mut line_count = 0;
assert("abcdef\n|gh\nij|k", cx, |left, _| {
if left == '\n' {
line_count += 1;
line_count == 2
} else {
false
}
});
}
#[gpui::test]
fn test_next_word_end(cx: &mut gpui::MutableAppContext) {
fn assert(marked_text: &str, cx: &mut gpui::MutableAppContext) {
let (snapshot, display_points) = marked_snapshot(marked_text, cx);
assert_eq!(
next_word_end(&snapshot, display_points[0]),
display_points[1]
);
}
assert("\n| lorem|", cx);
assert(" |lorem|", cx);
assert(" lor|em|", cx);
assert(" lorem| |\nipsum\n", cx);
assert("\n|\n|\n\n", cx);
assert("lorem| ipsum| ", cx);
assert("lorem|-|ipsum", cx);
assert("lorem|#$@-|ipsum", cx);
assert("lorem|_ipsum|", cx);
assert(" |bcΔ|", cx);
assert(" ab|——|cd", cx);
}
#[gpui::test]
fn test_next_subword_end(cx: &mut gpui::MutableAppContext) {
fn assert(marked_text: &str, cx: &mut gpui::MutableAppContext) {
let (snapshot, display_points) = marked_snapshot(marked_text, cx);
assert_eq!(
next_subword_end(&snapshot, display_points[0]),
display_points[1]
);
}
// Subword boundaries are respected
assert("lo|rem|_ipsum", cx);
assert("|lorem|_ipsum", cx);
assert("lorem|_ipsum|", cx);
assert("lorem|_ipsum|_dolor", cx);
assert("lo|rem|Ipsum", cx);
assert("lorem|Ipsum|Dolor", cx);
// Word boundaries are still respected
assert("\n| lorem|", cx);
assert(" |lorem|", cx);
assert(" lor|em|", cx);
assert(" lorem| |\nipsum\n", cx);
assert("\n|\n|\n\n", cx);
assert("lorem| ipsum| ", cx);
assert("lorem|-|ipsum", cx);
assert("lorem|#$@-|ipsum", cx);
assert("lorem|_ipsum|", cx);
assert(" |bc|Δ", cx);
assert(" ab|——|cd", cx);
}
#[gpui::test]
fn test_find_boundary(cx: &mut gpui::MutableAppContext) {
fn assert(
marked_text: &str,
cx: &mut gpui::MutableAppContext,
is_boundary: impl FnMut(char, char) -> bool,
) {
let (snapshot, display_points) = marked_snapshot(marked_text, cx);
assert_eq!(
find_boundary(&snapshot, display_points[0], is_boundary),
display_points[1]
);
}
assert("abc|def\ngh\nij|k", cx, |left, right| {
left == 'j' && right == 'k'
});
assert("ab|cdef\ngh\n|ijk", cx, |left, right| {
left == '\n' && right == 'i'
});
let mut line_count = 0;
assert("abc|def\ngh\n|ijk", cx, |left, _| {
if left == '\n' {
line_count += 1;
line_count == 2
} else {
false
}
});
}
#[gpui::test]
fn test_surrounding_word(cx: &mut gpui::MutableAppContext) {
fn assert(marked_text: &str, cx: &mut gpui::MutableAppContext) {
let (snapshot, display_points) = marked_snapshot(marked_text, cx);
assert_eq!(
surrounding_word(&snapshot, display_points[1]),
display_points[0]..display_points[2]
);
}
assert("||lorem| ipsum", cx);
assert("|lo|rem| ipsum", cx);
assert("|lorem|| ipsum", cx);
assert("lorem| | |ipsum", cx);
assert("lorem\n|||\nipsum", cx);
assert("lorem\n||ipsum|", cx);
assert("lorem,|| |ipsum", cx);
assert("|lorem||, ipsum", cx);
}
#[gpui::test]
fn test_move_up_and_down_with_excerpts(cx: &mut gpui::MutableAppContext) {
let family_id = cx.font_cache().load_family(&["Helvetica"]).unwrap();
@@ -297,8 +535,21 @@ mod tests {
);
}
#[gpui::test]
fn test_prev_next_word_boundary_multibyte(cx: &mut gpui::MutableAppContext) {
// Returns a snapshot from text containing '|' character markers with the markers removed, and DisplayPoints for each one.
fn marked_snapshot(
text: &str,
cx: &mut gpui::MutableAppContext,
) -> (DisplaySnapshot, Vec<DisplayPoint>) {
let mut marked_offsets = Vec::new();
let chunks = text.split('|');
let mut text = String::new();
for chunk in chunks {
text.push_str(chunk);
marked_offsets.push(text.len());
}
marked_offsets.pop();
let tab_size = 4;
let family_id = cx.font_cache().load_family(&["Helvetica"]).unwrap();
let font_id = cx
@@ -307,122 +558,15 @@ mod tests {
.unwrap();
let font_size = 14.0;
let buffer = MultiBuffer::build_simple("a bcΔ defγ hi—jk", cx);
let buffer = MultiBuffer::build_simple(&text, cx);
let display_map = cx
.add_model(|cx| DisplayMap::new(buffer, tab_size, font_id, font_size, None, 1, 1, cx));
let snapshot = display_map.update(cx, |map, cx| map.snapshot(cx));
assert_eq!(
prev_word_boundary(&snapshot, DisplayPoint::new(0, 12)),
DisplayPoint::new(0, 7)
);
assert_eq!(
prev_word_boundary(&snapshot, DisplayPoint::new(0, 7)),
DisplayPoint::new(0, 2)
);
assert_eq!(
prev_word_boundary(&snapshot, DisplayPoint::new(0, 6)),
DisplayPoint::new(0, 2)
);
assert_eq!(
prev_word_boundary(&snapshot, DisplayPoint::new(0, 2)),
DisplayPoint::new(0, 0)
);
assert_eq!(
prev_word_boundary(&snapshot, DisplayPoint::new(0, 1)),
DisplayPoint::new(0, 0)
);
let marked_display_points = marked_offsets
.into_iter()
.map(|offset| offset.to_display_point(&snapshot))
.collect();
assert_eq!(
next_word_boundary(&snapshot, DisplayPoint::new(0, 0)),
DisplayPoint::new(0, 1)
);
assert_eq!(
next_word_boundary(&snapshot, DisplayPoint::new(0, 1)),
DisplayPoint::new(0, 6)
);
assert_eq!(
next_word_boundary(&snapshot, DisplayPoint::new(0, 2)),
DisplayPoint::new(0, 6)
);
assert_eq!(
next_word_boundary(&snapshot, DisplayPoint::new(0, 6)),
DisplayPoint::new(0, 12)
);
assert_eq!(
next_word_boundary(&snapshot, DisplayPoint::new(0, 7)),
DisplayPoint::new(0, 12)
);
}
#[gpui::test]
fn test_surrounding_word(cx: &mut gpui::MutableAppContext) {
let tab_size = 4;
let family_id = cx.font_cache().load_family(&["Helvetica"]).unwrap();
let font_id = cx
.font_cache()
.select_font(family_id, &Default::default())
.unwrap();
let font_size = 14.0;
let buffer = MultiBuffer::build_simple("lorem ipsum dolor\n sit", cx);
let display_map = cx
.add_model(|cx| DisplayMap::new(buffer, tab_size, font_id, font_size, None, 1, 1, cx));
let snapshot = display_map.update(cx, |map, cx| map.snapshot(cx));
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(0, 0)),
DisplayPoint::new(0, 0)..DisplayPoint::new(0, 5),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(0, 2)),
DisplayPoint::new(0, 0)..DisplayPoint::new(0, 5),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(0, 5)),
DisplayPoint::new(0, 0)..DisplayPoint::new(0, 5),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(0, 6)),
DisplayPoint::new(0, 6)..DisplayPoint::new(0, 11),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(0, 7)),
DisplayPoint::new(0, 6)..DisplayPoint::new(0, 11),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(0, 11)),
DisplayPoint::new(0, 6)..DisplayPoint::new(0, 11),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(0, 13)),
DisplayPoint::new(0, 11)..DisplayPoint::new(0, 14),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(0, 14)),
DisplayPoint::new(0, 14)..DisplayPoint::new(0, 19),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(0, 17)),
DisplayPoint::new(0, 14)..DisplayPoint::new(0, 19),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(0, 19)),
DisplayPoint::new(0, 14)..DisplayPoint::new(0, 19),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(1, 0)),
DisplayPoint::new(1, 0)..DisplayPoint::new(1, 4),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(1, 1)),
DisplayPoint::new(1, 0)..DisplayPoint::new(1, 4),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(1, 6)),
DisplayPoint::new(1, 4)..DisplayPoint::new(1, 7),
);
assert_eq!(
surrounding_word(&snapshot, DisplayPoint::new(1, 7)),
DisplayPoint::new(1, 4)..DisplayPoint::new(1, 7),
);
(snapshot, marked_display_points)
}
}
+2 -2
View File
@@ -1407,7 +1407,7 @@ impl MultiBufferSnapshot {
);
for ch in prev_chars {
if Some(char_kind(ch)) == word_kind {
if Some(char_kind(ch)) == word_kind && ch != '\n' {
start -= ch.len_utf8();
} else {
break;
@@ -1415,7 +1415,7 @@ impl MultiBufferSnapshot {
}
for ch in next_chars {
if Some(char_kind(ch)) == word_kind {
if Some(char_kind(ch)) == word_kind && ch != '\n' {
end += ch.len_utf8();
} else {
break;
+1 -4
View File
@@ -271,7 +271,6 @@ pub(crate) struct DiagnosticEndpoint {
#[derive(Copy, Clone, Eq, PartialEq, PartialOrd, Ord, Debug)]
pub enum CharKind {
Newline,
Punctuation,
Whitespace,
Word,
@@ -2259,9 +2258,7 @@ pub fn contiguous_ranges(
}
pub fn char_kind(c: char) -> CharKind {
if c == '\n' {
CharKind::Newline
} else if c.is_whitespace() {
if c.is_whitespace() {
CharKind::Whitespace
} else if c.is_alphanumeric() || c == '_' {
CharKind::Word