aboutsummaryrefslogtreecommitdiff
path: root/helix-core
diff options
context:
space:
mode:
Diffstat (limited to 'helix-core')
-rw-r--r--helix-core/src/doc_formatter.rs384
-rw-r--r--helix-core/src/doc_formatter/test.rs222
-rw-r--r--helix-core/src/graphemes.rs183
-rw-r--r--helix-core/src/lib.rs8
-rw-r--r--helix-core/src/movement.rs202
-rw-r--r--helix-core/src/position.rs428
-rw-r--r--helix-core/src/selection.rs32
-rw-r--r--helix-core/src/text_annotations.rs271
8 files changed, 1690 insertions, 40 deletions
diff --git a/helix-core/src/doc_formatter.rs b/helix-core/src/doc_formatter.rs
new file mode 100644
index 00000000..c7dc9081
--- /dev/null
+++ b/helix-core/src/doc_formatter.rs
@@ -0,0 +1,384 @@
+//! The `DocumentFormatter` forms the bridge between the raw document text
+//! and onscreen positioning. It yields the text graphemes as an iterator
+//! and traverses (part) of the document text. During that traversal it
+//! handles grapheme detection, softwrapping and annotations.
+//! It yields `FormattedGrapheme`s and their corresponding visual coordinates.
+//!
+//! As both virtual text and softwrapping can insert additional lines into the document
+//! it is generally not possible to find the start of the previous visual line.
+//! Instead the `DocumentFormatter` starts at the last "checkpoint" (usually a linebreak)
+//! called a "block" and the caller must advance it as needed.
+
+use std::borrow::Cow;
+use std::fmt::Debug;
+use std::mem::{replace, take};
+
+#[cfg(test)]
+mod test;
+
+use unicode_segmentation::{Graphemes, UnicodeSegmentation};
+
+use crate::graphemes::{Grapheme, GraphemeStr};
+use crate::syntax::Highlight;
+use crate::text_annotations::TextAnnotations;
+use crate::{Position, RopeGraphemes, RopeSlice};
+
+/// TODO make Highlight a u32 to reduce the size of this enum to a single word.
+#[derive(Debug, Clone, Copy)]
+pub enum GraphemeSource {
+ Document {
+ codepoints: u32,
+ },
+ /// Inline virtual text can not be highlighted with a `Highlight` iterator
+ /// because it's not part of the document. Instead the `Highlight`
+ /// is emitted right by the document formatter
+ VirtualText {
+ highlight: Option<Highlight>,
+ },
+}
+
+#[derive(Debug, Clone)]
+pub struct FormattedGrapheme<'a> {
+ pub grapheme: Grapheme<'a>,
+ pub source: GraphemeSource,
+}
+
+impl<'a> FormattedGrapheme<'a> {
+ pub fn new(
+ g: GraphemeStr<'a>,
+ visual_x: usize,
+ tab_width: u16,
+ source: GraphemeSource,
+ ) -> FormattedGrapheme<'a> {
+ FormattedGrapheme {
+ grapheme: Grapheme::new(g, visual_x, tab_width),
+ source,
+ }
+ }
+ /// Returns whether this grapheme is virtual inline text
+ pub fn is_virtual(&self) -> bool {
+ matches!(self.source, GraphemeSource::VirtualText { .. })
+ }
+
+ pub fn placeholder() -> Self {
+ FormattedGrapheme {
+ grapheme: Grapheme::Other { g: " ".into() },
+ source: GraphemeSource::Document { codepoints: 0 },
+ }
+ }
+
+ pub fn doc_chars(&self) -> usize {
+ match self.source {
+ GraphemeSource::Document { codepoints } => codepoints as usize,
+ GraphemeSource::VirtualText { .. } => 0,
+ }
+ }
+
+ pub fn is_whitespace(&self) -> bool {
+ self.grapheme.is_whitespace()
+ }
+
+ pub fn width(&self) -> usize {
+ self.grapheme.width()
+ }
+
+ pub fn is_word_boundary(&self) -> bool {
+ self.grapheme.is_word_boundary()
+ }
+}
+
+#[derive(Debug, Clone)]
+pub struct TextFormat {
+ pub soft_wrap: bool,
+ pub tab_width: u16,
+ pub max_wrap: u16,
+ pub max_indent_retain: u16,
+ pub wrap_indicator: Box<str>,
+ pub wrap_indicator_highlight: Option<Highlight>,
+ pub viewport_width: u16,
+}
+
+// test implementation is basically only used for testing or when softwrap is always disabled
+impl Default for TextFormat {
+ fn default() -> Self {
+ TextFormat {
+ soft_wrap: false,
+ tab_width: 4,
+ max_wrap: 3,
+ max_indent_retain: 4,
+ wrap_indicator: Box::from(" "),
+ viewport_width: 17,
+ wrap_indicator_highlight: None,
+ }
+ }
+}
+
+#[derive(Debug)]
+pub struct DocumentFormatter<'t> {
+ text_fmt: &'t TextFormat,
+ annotations: &'t TextAnnotations,
+
+ /// The visual position at the end of the last yielded word boundary
+ visual_pos: Position,
+ graphemes: RopeGraphemes<'t>,
+ /// The character pos of the `graphemes` iter used for inserting annotations
+ char_pos: usize,
+ /// The line pos of the `graphemes` iter used for inserting annotations
+ line_pos: usize,
+ exhausted: bool,
+
+ /// Line breaks to be reserved for virtual text
+ /// at the next line break
+ virtual_lines: usize,
+ inline_anntoation_graphemes: Option<(Graphemes<'t>, Option<Highlight>)>,
+
+ // softwrap specific
+ /// The indentation of the current line
+ /// Is set to `None` if the indentation level is not yet known
+ /// because no non-whitespace graphemes have been encountered yet
+ indent_level: Option<usize>,
+ /// In case a long word needs to be split a single grapheme might need to be wrapped
+ /// while the rest of the word stays on the same line
+ peeked_grapheme: Option<(FormattedGrapheme<'t>, usize)>,
+ /// A first-in first-out (fifo) buffer for the Graphemes of any given word
+ word_buf: Vec<FormattedGrapheme<'t>>,
+ /// The index of the next grapheme that will be yielded from the `word_buf`
+ word_i: usize,
+}
+
+impl<'t> DocumentFormatter<'t> {
+ /// Creates a new formatter at the last block before `char_idx`.
+ /// A block is a chunk which always ends with a linebreak.
+ /// This is usually just a normal line break.
+ /// However very long lines are always wrapped at constant intervals that can be cheaply calculated
+ /// to avoid pathological behaviour.
+ pub fn new_at_prev_checkpoint(
+ text: RopeSlice<'t>,
+ text_fmt: &'t TextFormat,
+ annotations: &'t TextAnnotations,
+ char_idx: usize,
+ ) -> (Self, usize) {
+ // TODO divide long lines into blocks to avoid bad performance for long lines
+ let block_line_idx = text.char_to_line(char_idx.min(text.len_chars()));
+ let block_char_idx = text.line_to_char(block_line_idx);
+ annotations.reset_pos(block_char_idx);
+ (
+ DocumentFormatter {
+ text_fmt,
+ annotations,
+ visual_pos: Position { row: 0, col: 0 },
+ graphemes: RopeGraphemes::new(text.slice(block_char_idx..)),
+ char_pos: block_char_idx,
+ exhausted: false,
+ virtual_lines: 0,
+ indent_level: None,
+ peeked_grapheme: None,
+ word_buf: Vec::with_capacity(64),
+ word_i: 0,
+ line_pos: block_line_idx,
+ inline_anntoation_graphemes: None,
+ },
+ block_char_idx,
+ )
+ }
+
+ fn next_inline_annotation_grapheme(&mut self) -> Option<(&'t str, Option<Highlight>)> {
+ loop {
+ if let Some(&mut (ref mut annotation, highlight)) =
+ self.inline_anntoation_graphemes.as_mut()
+ {
+ if let Some(grapheme) = annotation.next() {
+ return Some((grapheme, highlight));
+ }
+ }
+
+ if let Some((annotation, highlight)) =
+ self.annotations.next_inline_annotation_at(self.char_pos)
+ {
+ self.inline_anntoation_graphemes = Some((
+ UnicodeSegmentation::graphemes(&*annotation.text, true),
+ highlight,
+ ))
+ } else {
+ return None;
+ }
+ }
+ }
+
+ fn advance_grapheme(&mut self, col: usize) -> Option<FormattedGrapheme<'t>> {
+ let (grapheme, source) =
+ if let Some((grapheme, highlight)) = self.next_inline_annotation_grapheme() {
+ (grapheme.into(), GraphemeSource::VirtualText { highlight })
+ } else if let Some(grapheme) = self.graphemes.next() {
+ self.virtual_lines += self.annotations.annotation_lines_at(self.char_pos);
+ let codepoints = grapheme.len_chars() as u32;
+
+ let overlay = self.annotations.overlay_at(self.char_pos);
+ let grapheme = match overlay {
+ Some((overlay, _)) => overlay.grapheme.as_str().into(),
+ None => Cow::from(grapheme).into(),
+ };
+
+ self.char_pos += codepoints as usize;
+ (grapheme, GraphemeSource::Document { codepoints })
+ } else {
+ if self.exhausted {
+ return None;
+ }
+ self.exhausted = true;
+ // EOF grapheme is required for rendering
+ // and correct position computations
+ return Some(FormattedGrapheme {
+ grapheme: Grapheme::Other { g: " ".into() },
+ source: GraphemeSource::Document { codepoints: 0 },
+ });
+ };
+
+ let grapheme = FormattedGrapheme::new(grapheme, col, self.text_fmt.tab_width, source);
+
+ Some(grapheme)
+ }
+
+ /// Move a word to the next visual line
+ fn wrap_word(&mut self, virtual_lines_before_word: usize) -> usize {
+ // softwrap this word to the next line
+ let indent_carry_over = if let Some(indent) = self.indent_level {
+ if indent as u16 <= self.text_fmt.max_indent_retain {
+ indent as u16
+ } else {
+ 0
+ }
+ } else {
+ // ensure the indent stays 0
+ self.indent_level = Some(0);
+ 0
+ };
+
+ self.visual_pos.col = indent_carry_over as usize;
+ self.virtual_lines -= virtual_lines_before_word;
+ self.visual_pos.row += 1 + virtual_lines_before_word;
+ let mut i = 0;
+ let mut word_width = 0;
+ let wrap_indicator = UnicodeSegmentation::graphemes(&*self.text_fmt.wrap_indicator, true)
+ .map(|g| {
+ i += 1;
+ let grapheme = FormattedGrapheme::new(
+ g.into(),
+ self.visual_pos.col + word_width,
+ self.text_fmt.tab_width,
+ GraphemeSource::VirtualText {
+ highlight: self.text_fmt.wrap_indicator_highlight,
+ },
+ );
+ word_width += grapheme.width();
+ grapheme
+ });
+ self.word_buf.splice(0..0, wrap_indicator);
+
+ for grapheme in &mut self.word_buf[i..] {
+ let visual_x = self.visual_pos.col + word_width;
+ grapheme
+ .grapheme
+ .change_position(visual_x, self.text_fmt.tab_width);
+ word_width += grapheme.width();
+ }
+ word_width
+ }
+
+ fn advance_to_next_word(&mut self) {
+ self.word_buf.clear();
+ let mut word_width = 0;
+ let virtual_lines_before_word = self.virtual_lines;
+ let mut virtual_lines_before_grapheme = self.virtual_lines;
+
+ loop {
+ // softwrap word if necessary
+ if word_width + self.visual_pos.col >= self.text_fmt.viewport_width as usize {
+ // wrapping this word would move too much text to the next line
+ // split the word at the line end instead
+ if word_width > self.text_fmt.max_wrap as usize {
+ // Usually we stop accomulating graphemes as soon as softwrapping becomes necessary.
+ // However if the last grapheme is multiple columns wide it might extend beyond the EOL.
+ // The condition below ensures that this grapheme is not cutoff and instead wrapped to the next line
+ if word_width + self.visual_pos.col > self.text_fmt.viewport_width as usize {
+ self.peeked_grapheme = self.word_buf.pop().map(|grapheme| {
+ (grapheme, self.virtual_lines - virtual_lines_before_grapheme)
+ });
+ self.virtual_lines = virtual_lines_before_grapheme;
+ }
+ return;
+ }
+
+ word_width = self.wrap_word(virtual_lines_before_word);
+ }
+
+ virtual_lines_before_grapheme = self.virtual_lines;
+
+ let grapheme = if let Some((grapheme, virtual_lines)) = self.peeked_grapheme.take() {
+ self.virtual_lines += virtual_lines;
+ grapheme
+ } else if let Some(grapheme) = self.advance_grapheme(self.visual_pos.col + word_width) {
+ grapheme
+ } else {
+ return;
+ };
+
+ // Track indentation
+ if !grapheme.is_whitespace() && self.indent_level.is_none() {
+ self.indent_level = Some(self.visual_pos.col);
+ } else if grapheme.grapheme == Grapheme::Newline {
+ self.indent_level = None;
+ }
+
+ let is_word_boundary = grapheme.is_word_boundary();
+ word_width += grapheme.width();
+ self.word_buf.push(grapheme);
+
+ if is_word_boundary {
+ return;
+ }
+ }
+ }
+
+ /// returns the document line pos of the **next** grapheme that will be yielded
+ pub fn line_pos(&self) -> usize {
+ self.line_pos
+ }
+
+ /// returns the visual pos of the **next** grapheme that will be yielded
+ pub fn visual_pos(&self) -> Position {
+ self.visual_pos
+ }
+}
+
+impl<'t> Iterator for DocumentFormatter<'t> {
+ type Item = (FormattedGrapheme<'t>, Position);
+
+ fn next(&mut self) -> Option<Self::Item> {
+ let grapheme = if self.text_fmt.soft_wrap {
+ if self.word_i >= self.word_buf.len() {
+ self.advance_to_next_word();
+ self.word_i = 0;
+ }
+ let grapheme = replace(
+ self.word_buf.get_mut(self.word_i)?,
+ FormattedGrapheme::placeholder(),
+ );
+ self.word_i += 1;
+ grapheme
+ } else {
+ self.advance_grapheme(self.visual_pos.col)?
+ };
+
+ let pos = self.visual_pos;
+ if grapheme.grapheme == Grapheme::Newline {
+ self.visual_pos.row += 1;
+ self.visual_pos.row += take(&mut self.virtual_lines);
+ self.visual_pos.col = 0;
+ self.line_pos += 1;
+ } else {
+ self.visual_pos.col += grapheme.width();
+ }
+ Some((grapheme, pos))
+ }
+}
diff --git a/helix-core/src/doc_formatter/test.rs b/helix-core/src/doc_formatter/test.rs
new file mode 100644
index 00000000..e68b31fd
--- /dev/null
+++ b/helix-core/src/doc_formatter/test.rs
@@ -0,0 +1,222 @@
+use std::rc::Rc;
+
+use crate::doc_formatter::{DocumentFormatter, TextFormat};
+use crate::text_annotations::{InlineAnnotation, Overlay, TextAnnotations};
+
+impl TextFormat {
+ fn new_test(softwrap: bool) -> Self {
+ TextFormat {
+ soft_wrap: softwrap,
+ tab_width: 2,
+ max_wrap: 3,
+ max_indent_retain: 4,
+ wrap_indicator: ".".into(),
+ wrap_indicator_highlight: None,
+ // use a prime number to allow lining up too often with repeat
+ viewport_width: 17,
+ }
+ }
+}
+
+impl<'t> DocumentFormatter<'t> {
+ fn collect_to_str(&mut self) -> String {
+ use std::fmt::Write;
+ let mut res = String::new();
+ let viewport_width = self.text_fmt.viewport_width;
+ let mut line = 0;
+
+ for (grapheme, pos) in self {
+ if pos.row != line {
+ line += 1;
+ assert_eq!(pos.row, line);
+ write!(res, "\n{}", ".".repeat(pos.col)).unwrap();
+ assert!(
+ pos.col <= viewport_width as usize,
+ "softwrapped failed {}<={viewport_width}",
+ pos.col
+ );
+ }
+ write!(res, "{}", grapheme.grapheme).unwrap();
+ }
+
+ res
+ }
+}
+
+fn softwrap_text(text: &str) -> String {
+ DocumentFormatter::new_at_prev_checkpoint(
+ text.into(),
+ &TextFormat::new_test(true),
+ &TextAnnotations::default(),
+ 0,
+ )
+ .0
+ .collect_to_str()
+}
+
+#[test]
+fn basic_softwrap() {
+ assert_eq!(
+ softwrap_text(&"foo ".repeat(10)),
+ "foo foo foo foo \n.foo foo foo foo \n.foo foo "
+ );
+ assert_eq!(
+ softwrap_text(&"fooo ".repeat(10)),
+ "fooo fooo fooo \n.fooo fooo fooo \n.fooo fooo fooo \n.fooo "
+ );
+
+ // check that we don't wrap unnecessarily
+ assert_eq!(softwrap_text("\t\txxxx1xxxx2xx\n"), " xxxx1xxxx2xx \n ");
+}
+
+#[test]
+fn softwrap_indentation() {
+ assert_eq!(
+ softwrap_text("\t\tfoo1 foo2 foo3 foo4 foo5 foo6\n"),
+ " foo1 foo2 \n.....foo3 foo4 \n.....foo5 foo6 \n "
+ );
+ assert_eq!(
+ softwrap_text("\t\t\tfoo1 foo2 foo3 foo4 foo5 foo6\n"),
+ " foo1 foo2 \n.foo3 foo4 foo5 \n.foo6 \n "
+ );
+}
+
+#[test]
+fn long_word_softwrap() {
+ assert_eq!(
+ softwrap_text("\t\txxxx1xxxx2xxxx3xxxx4xxxx5xxxx6xxxx7xxxx8xxxx9xxx\n"),
+ " xxxx1xxxx2xxx\n.....x3xxxx4xxxx5\n.....xxxx6xxxx7xx\n.....xx8xxxx9xxx \n "
+ );
+ assert_eq!(
+ softwrap_text("xxxxxxxx1xxxx2xxx\n"),
+ "xxxxxxxx1xxxx2xxx\n. \n "
+ );
+ assert_eq!(
+ softwrap_text("\t\txxxx1xxxx 2xxxx3xxxx4xxxx5xxxx6xxxx7xxxx8xxxx9xxx\n"),
+ " xxxx1xxxx \n.....2xxxx3xxxx4x\n.....xxx5xxxx6xxx\n.....x7xxxx8xxxx9\n.....xxx \n "
+ );
+ assert_eq!(
+ softwrap_text("\t\txxxx1xxx 2xxxx3xxxx4xxxx5xxxx6xxxx7xxxx8xxxx9xxx\n"),
+ " xxxx1xxx 2xxx\n.....x3xxxx4xxxx5\n.....xxxx6xxxx7xx\n.....xx8xxxx9xxx \n "
+ );
+}
+
+fn overlay_text(text: &str, char_pos: usize, softwrap: bool, overlays: &[Overlay]) -> String {
+ DocumentFormatter::new_at_prev_checkpoint(
+ text.into(),
+ &TextFormat::new_test(softwrap),
+ TextAnnotations::default().add_overlay(overlays.into(), None),
+ char_pos,
+ )
+ .0
+ .collect_to_str()
+}
+
+#[test]
+fn overlay() {
+ assert_eq!(
+ overlay_text(
+ "foobar",
+ 0,
+ false,
+ &[
+ Overlay {
+ char_idx: 0,
+ grapheme: "X".into(),
+ },
+ Overlay {
+ char_idx: 2,
+ grapheme: "\t".into(),
+ },
+ ]
+ ),
+ "Xo bar "
+ );
+ assert_eq!(
+ overlay_text(
+ &"foo ".repeat(10),
+ 0,
+ true,
+ &[
+ Overlay {
+ char_idx: 2,
+ grapheme: "\t".into(),
+ },
+ Overlay {
+ char_idx: 5,
+ grapheme: "\t".into(),
+ },
+ Overlay {
+ char_idx: 16,
+ grapheme: "X".into(),
+ },
+ ]
+ ),
+ "fo f o foo \n.foo Xoo foo foo \n.foo foo foo "
+ );
+}
+
+fn annotate_text(text: &str, softwrap: bool, annotations: &[InlineAnnotation]) -> String {
+ DocumentFormatter::new_at_prev_checkpoint(
+ text.into(),
+ &TextFormat::new_test(softwrap),
+ TextAnnotations::default().add_inline_annotations(annotations.into(), None),
+ 0,
+ )
+ .0
+ .collect_to_str()
+}
+
+#[test]
+fn annotation() {
+ assert_eq!(
+ annotate_text(
+ "bar",
+ false,
+ &[InlineAnnotation {
+ char_idx: 0,
+ text: "foo".into(),
+ }]
+ ),
+ "foobar "
+ );
+ assert_eq!(
+ annotate_text(
+ &"foo ".repeat(10),
+ true,
+ &[InlineAnnotation {
+ char_idx: 0,
+ text: "foo ".into(),
+ }]
+ ),
+ "foo foo foo foo \n.foo foo foo foo \n.foo foo foo "
+ );
+}
+#[test]
+fn annotation_and_overlay() {
+ assert_eq!(
+ DocumentFormatter::new_at_prev_checkpoint(
+ "bbar".into(),
+ &TextFormat::new_test(false),
+ TextAnnotations::default()
+ .add_inline_annotations(
+ Rc::new([InlineAnnotation {
+ char_idx: 0,
+ text: "fooo".into(),
+ }]),
+ None
+ )
+ .add_overlay(
+ Rc::new([Overlay {
+ char_idx: 0,
+ grapheme: "\t".into(),
+ }]),
+ None
+ ),
+ 0,
+ )
+ .0
+ .collect_to_str(),
+ "fooo bar "
+ );
+}
diff --git a/helix-core/src/graphemes.rs b/helix-core/src/graphemes.rs
index 675f5750..15ef3eb0 100644
--- a/helix-core/src/graphemes.rs
+++ b/helix-core/src/graphemes.rs
@@ -5,7 +5,88 @@ use ropey::{iter::Chunks, str_utils::byte_to_char_idx, RopeSlice};
use unicode_segmentation::{GraphemeCursor, GraphemeIncomplete};
use unicode_width::UnicodeWidthStr;
-use std::fmt;
+use std::borrow::Cow;
+use std::fmt::{self, Debug, Display};
+use std::marker::PhantomData;
+use std::ops::Deref;
+use std::ptr::NonNull;
+use std::{slice, str};
+
+use crate::chars::{char_is_whitespace, char_is_word};
+use crate::LineEnding;
+
+#[inline]
+pub fn tab_width_at(visual_x: usize, tab_width: u16) -> usize {
+ tab_width as usize - (visual_x % tab_width as usize)
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum Grapheme<'a> {
+ Newline,
+ Tab { width: usize },
+ Other { g: GraphemeStr<'a> },
+}
+
+impl<'a> Grapheme<'a> {
+ pub fn new(g: GraphemeStr<'a>, visual_x: usize, tab_width: u16) -> Grapheme<'a> {
+ match g {
+ g if g == "\t" => Grapheme::Tab {
+ width: tab_width_at(visual_x, tab_width),
+ },
+ _ if LineEnding::from_str(&g).is_some() => Grapheme::Newline,
+ _ => Grapheme::Other { g },
+ }
+ }
+
+ pub fn change_position(&mut self, visual_x: usize, tab_width: u16) {
+ if let Grapheme::Tab { width } = self {
+ *width = tab_width_at(visual_x, tab_width)
+ }
+ }
+
+ /// Returns the a visual width of this grapheme,
+ #[inline]
+ pub fn width(&self) -> usize {
+ match *self {
+ // width is not cached because we are dealing with
+ // ASCII almost all the time which already has a fastpath
+ // it's okay to convert to u16 here because no codepoint has a width larger
+ // than 2 and graphemes are usually atmost two visible codepoints wide
+ Grapheme::Other { ref g } => grapheme_width(g),
+ Grapheme::Tab { width } => width,
+ Grapheme::Newline => 1,
+ }
+ }
+
+ pub fn is_whitespace(&self) -> bool {
+ !matches!(&self, Grapheme::Other { g } if !g.chars().all(char_is_whitespace))
+ }
+
+ // TODO currently word boundaries are used for softwrapping.
+ // This works best for programming languages and well for prose.
+ // This could however be improved in the future by considering unicode
+ // character classes but
+ pub fn is_word_boundary(&self) -> bool {
+ !matches!(&self, Grapheme::Other { g,.. } if g.chars().all(char_is_word))
+ }
+}
+
+impl Display for Grapheme<'_> {
+ fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+ match *self {
+ Grapheme::Newline => write!(f, " "),
+ Grapheme::Tab { width } => {
+ for _ in 0..width {
+ write!(f, " ")?;
+ }
+ Ok(())
+ }
+ Grapheme::Other { ref g } => {
+ write!(f, "{g}")
+ }
+ }
+ }
+}
#[must_use]
pub fn grapheme_width(g: &str) -> usize {
@@ -27,6 +108,8 @@ pub fn grapheme_width(g: &str) -> usize {
// We use max(1) here because all grapeheme clusters--even illformed
// ones--should have at least some width so they can be edited
// properly.
+ // TODO properly handle unicode width for all codepoints
+ // example of where unicode width is currently wrong: 🤦🏼‍♂️ (taken from https://hsivonen.fi/string-length/)
UnicodeWidthStr::width(g).max(1)
}
}
@@ -341,3 +424,101 @@ impl<'a> Iterator for RopeGraphemes<'a> {
}
}
}
+
+/// A highly compressed Cow<'a, str> that holds
+/// atmost u31::MAX bytes and is readonly
+pub struct GraphemeStr<'a> {
+ ptr: NonNull<u8>,
+ len: u32,
+ phantom: PhantomData<&'a str>,
+}
+
+impl GraphemeStr<'_> {
+ const MASK_OWNED: u32 = 1 << 31;
+
+ fn compute_len(&self) -> usize {
+ (self.len & !Self::MASK_OWNED) as usize
+ }
+}
+
+impl Deref for GraphemeStr<'_> {
+ type Target = str;
+ fn deref(&self) -> &Self::Target {
+ unsafe {
+ let bytes = slice::from_raw_parts(self.ptr.as_ptr(), self.compute_len());
+ str::from_utf8_unchecked(bytes)
+ }
+ }
+}
+
+impl Drop for GraphemeStr<'_> {
+ fn drop(&mut self) {
+ if self.len & Self::MASK_OWNED != 0 {
+ // free allocation
+ unsafe {
+ drop(Box::from_raw(slice::from_raw_parts_mut(
+ self.ptr.as_ptr(),
+ self.compute_len(),
+ )));
+ }
+ }
+ }
+}
+
+impl<'a> From<&'a str> for GraphemeStr<'a> {
+ fn from(g: &'a str) -> Self {
+ GraphemeStr {
+ ptr: unsafe { NonNull::new_unchecked(g.as_bytes().as_ptr() as *mut u8) },
+ len: i32::try_from(g.len()).unwrap() as u32,
+ phantom: PhantomData,
+ }
+ }
+}
+
+impl<'a> From<String> for GraphemeStr<'a> {
+ fn from(g: String) -> Self {
+ let len = g.len();
+ let ptr = Box::into_raw(g.into_bytes().into_boxed_slice()) as *mut u8;
+ GraphemeStr {
+ ptr: unsafe { NonNull::new_unchecked(ptr) },
+ len: i32::try_from(len).unwrap() as u32,
+ phantom: PhantomData,
+ }
+ }
+}
+
+impl<'a> From<Cow<'a, str>> for GraphemeStr<'a> {
+ fn from(g: Cow<'a, str>) -> Self {
+ match g {
+ Cow::Borrowed(g) => g.into(),
+ Cow::Owned(g) => g.into(),
+ }
+ }
+}
+
+impl<T: Deref<Target = str>> PartialEq<T> for GraphemeStr<'_> {
+ fn eq(&self, other: &T) -> bool {
+ self.deref() == other.deref()
+ }
+}
+impl PartialEq<str> for GraphemeStr<'_> {
+ fn eq(&self, other: &str) -> bool {
+ self.deref() == other
+ }
+}
+impl Eq for GraphemeStr<'_> {}
+impl Debug for GraphemeStr<'_> {
+ fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+ Debug::fmt(self.deref(), f)
+ }
+}
+impl Display for GraphemeStr<'_> {
+ fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+ Display::fmt(self.deref(), f)
+ }
+}
+impl Clone for GraphemeStr<'_> {
+ fn clone(&self) -> Self {
+ self.deref().to_owned().into()
+ }
+}
diff --git a/helix-core/src/lib.rs b/helix-core/src/lib.rs
index ee174e69..e3f862a6 100644
--- a/helix-core/src/lib.rs
+++ b/helix-core/src/lib.rs
@@ -6,6 +6,7 @@ pub mod comment;
pub mod config;
pub mod diagnostic;
pub mod diff;
+pub mod doc_formatter;
pub mod graphemes;
pub mod history;
pub mod increment;
@@ -24,6 +25,7 @@ pub mod shellwords;
pub mod surround;
pub mod syntax;
pub mod test;
+pub mod text_annotations;
pub mod textobject;
mod transaction;
pub mod wrap;
@@ -95,8 +97,12 @@ pub use {regex, tree_sitter};
pub use graphemes::RopeGraphemes;
pub use position::{
- coords_at_pos, pos_at_coords, pos_at_visual_coords, visual_coords_at_pos, Position,
+ char_idx_at_visual_offset, coords_at_pos, pos_at_coords, visual_offset_from_anchor,
+ visual_offset_from_block, Position,
};
+#[allow(deprecated)]
+pub use position::{pos_at_visual_coords, visual_coords_at_pos};
+
pub use selection::{Range, Selection};
pub use smallvec::{smallvec, SmallVec};
pub use syntax::Syntax;
diff --git a/helix-core/src/movement.rs b/helix-core/src/movement.rs
index 278375e8..11c12a6f 100644
--- a/helix-core/src/movement.rs
+++ b/helix-core/src/movement.rs
@@ -4,16 +4,19 @@ use ropey::iter::Chars;
use tree_sitter::{Node, QueryCursor};
use crate::{
+ char_idx_at_visual_offset,
chars::{categorize_char, char_is_line_ending, CharCategory},
+ doc_formatter::TextFormat,
graphemes::{
next_grapheme_boundary, nth_next_grapheme_boundary, nth_prev_grapheme_boundary,
prev_grapheme_boundary,
},
line_ending::rope_is_line_ending,
- pos_at_visual_coords,
+ position::char_idx_at_visual_block_offset,
syntax::LanguageConfiguration,
+ text_annotations::TextAnnotations,
textobject::TextObject,
- visual_coords_at_pos, Position, Range, RopeSlice,
+ visual_offset_from_block, Range, RopeSlice,
};
#[derive(Debug, Copy, Clone, PartialEq, Eq)]
@@ -34,7 +37,8 @@ pub fn move_horizontally(
dir: Direction,
count: usize,
behaviour: Movement,
- _: usize,
+ _: &TextFormat,
+ _: &mut TextAnnotations,
) -> Range {
let pos = range.cursor(slice);
@@ -48,35 +52,116 @@ pub fn move_horizontally(
range.put_cursor(slice, new_pos, behaviour == Movement::Extend)
}
+pub fn move_vertically_visual(
+ slice: RopeSlice,
+ range: Range,
+ dir: Direction,
+ count: usize,
+ behaviour: Movement,
+ text_fmt: &TextFormat,
+ annotations: &mut TextAnnotations,
+) -> Range {
+ if !text_fmt.soft_wrap {
+ move_vertically(slice, range, dir, count, behaviour, text_fmt, annotations);
+ }
+ annotations.clear_line_annotations();
+ let pos = range.cursor(slice);
+
+ // Compute the current position's 2d coordinates.
+ let (visual_pos, block_off) = visual_offset_from_block(slice, pos, pos, text_fmt, annotations);
+ let new_col = range
+ .old_visual_position
+ .map_or(visual_pos.col as u32, |(_, col)| col);
+
+ // Compute the new position.
+ let mut row_off = match dir {
+ Direction::Forward => count as isize,
+ Direction::Backward => -(count as isize),
+ };
+
+ // TODO how to handle inline annotations that span an entire visual line (very unlikely).
+
+ // Compute visual offset relative to block start to avoid trasversing the block twice
+ row_off += visual_pos.row as isize;
+ let new_pos = char_idx_at_visual_offset(
+ slice,
+ block_off,
+ row_off,
+ new_col as usize,
+ text_fmt,
+ annotations,
+ )
+ .0;
+
+ // Special-case to avoid moving to the end of the last non-empty line.
+ if behaviour == Movement::Extend && slice.line(slice.char_to_line(new_pos)).len_chars() == 0 {
+ return range;
+ }
+
+ let mut new_range = range.put_cursor(slice, new_pos, behaviour == Movement::Extend);
+ new_range.old_visual_position = Some((0, new_col));
+ new_range
+}
+
pub fn move_vertically(
slice: RopeSlice,
range: Range,
dir: Direction,
count: usize,
behaviour: Movement,
- tab_width: usize,
+ text_fmt: &TextFormat,
+ annotations: &mut TextAnnotations,
) -> Range {
+ annotations.clear_line_annotations();
let pos = range.cursor(slice);
+ let line_idx = slice.char_to_line(pos);
+ let line_start = slice.line_to_char(line_idx);
// Compute the current position's 2d coordinates.
- let Position { row, col } = visual_coords_at_pos(slice, pos, tab_width);
- let horiz = range.horiz.unwrap_or(col as u32);
+ let visual_pos = visual_offset_from_block(slice, line_start, pos, text_fmt, annotations).0;
+ let (mut new_row, new_col) = range
+ .old_visual_position
+ .map_or((visual_pos.row as u32, visual_pos.col as u32), |pos| pos);
+ new_row = new_row.max(visual_pos.row as u32);
+ let line_idx = slice.char_to_line(pos);
// Compute the new position.
- let new_row = match dir {
- Direction::Forward => (row + count).min(slice.len_lines().saturating_sub(1)),
- Direction::Backward => row.saturating_sub(count),
+ let mut new_line_idx = match dir {
+ Direction::Forward => line_idx.saturating_add(count),
+ Direction::Backward => line_idx.saturating_sub(count),
};
- let new_col = col.max(horiz as usize);
- let new_pos = pos_at_visual_coords(slice, Position::new(new_row, new_col), tab_width);
+
+ let line = if new_line_idx >= slice.len_lines() - 1 {
+ // there is no line terminator for the last line
+ // so the logic below is not necessary here
+ new_line_idx = slice.len_lines() - 1;
+ slice
+ } else {
+ // char_idx_at_visual_block_offset returns a one-past-the-end index
+ // in case it reaches the end of the slice
+ // to avoid moving to the nextline in that case the line terminator is removed from the line
+ let new_line_end = prev_grapheme_boundary(slice, slice.line_to_char(new_line_idx + 1));
+ slice.slice(..new_line_end)
+ };
+
+ let new_line_start = line.line_to_char(new_line_idx);
+
+ let (new_pos, _) = char_idx_at_visual_block_offset(
+ line,
+ new_line_start,
+ new_row as usize,
+ new_col as usize,
+ text_fmt,
+ annotations,
+ );
// Special-case to avoid moving to the end of the last non-empty line.
- if behaviour == Movement::Extend && slice.line(new_row).len_chars() == 0 {
+ if behaviour == Movement::Extend && slice.line(new_line_idx).len_chars() == 0 {
return range;
}
let mut new_range = range.put_cursor(slice, new_pos, behaviour == Movement::Extend);
- new_range.horiz = Some(horiz);
+ new_range.old_visual_position = Some((new_row, new_col));
new_range
}
@@ -473,7 +558,16 @@ mod test {
assert_eq!(
coords_at_pos(
slice,
- move_vertically(slice, range, Direction::Forward, 1, Movement::Move, 4).head
+ move_vertically_visual(
+ slice,
+ range,
+ Direction::Forward,
+ 1,
+ Movement::Move,
+ &TextFormat::default(),
+ &mut TextAnnotations::default(),
+ )
+ .head
),
(1, 3).into()
);
@@ -497,7 +591,15 @@ mod test {
];
for ((direction, amount), coordinates) in moves_and_expected_coordinates {
- range = move_horizontally(slice, range, direction, amount, Movement::Move, 0);
+ range = move_horizontally(
+ slice,
+ range,
+ direction,
+ amount,
+ Movement::Move,
+ &TextFormat::default(),
+ &mut TextAnnotations::default(),
+ );
assert_eq!(coords_at_pos(slice, range.head), coordinates.into())
}
}
@@ -523,7 +625,15 @@ mod test {
];
for ((direction, amount), coordinates) in moves_and_expected_coordinates {
- range = move_horizontally(slice, range, direction, amount, Movement::Move, 0);
+ range = move_horizontally(
+ slice,
+ range,
+ direction,
+ amount,
+ Movement::Move,
+ &TextFormat::default(),
+ &mut TextAnnotations::default(),
+ );
assert_eq!(coords_at_pos(slice, range.head), coordinates.into());
assert_eq!(range.head, range.anchor);
}
@@ -545,7 +655,15 @@ mod test {
];
for (direction, amount) in moves {
- range = move_horizontally(slice, range, direction, amount, Movement::Extend, 0);
+ range = move_horizontally(
+ slice,
+ range,
+ direction,
+ amount,
+ Movement::Extend,
+ &TextFormat::default(),
+ &mut TextAnnotations::default(),
+ );
assert_eq!(range.anchor, original_anchor);
}
}
@@ -569,7 +687,15 @@ mod test {
];
for ((direction, amount), coordinates) in moves_and_expected_coordinates {
- range = move_vertically(slice, range, direction, amount, Movement::Move, 4);
+ range = move_vertically_visual(
+ slice,
+ range,
+ direction,
+ amount,
+ Movement::Move,
+ &TextFormat::default(),
+ &mut TextAnnotations::default(),
+ );
assert_eq!(coords_at_pos(slice, range.head), coordinates.into());
assert_eq!(range.head, range.anchor);
}
@@ -603,8 +729,24 @@ mod test {
for ((axis, direction, amount), coordinates) in moves_and_expected_coordinates {
range = match axis {
- Axis::H => move_horizontally(slice, range, direction, amount, Movement::Move, 0),
- Axis::V => move_vertically(slice, range, direction, amount, Movement::Move, 4),
+ Axis::H => move_horizontally(
+ slice,
+ range,
+ direction,
+ amount,
+ Movement::Move,
+ &TextFormat::default(),
+ &mut TextAnnotations::default(),
+ ),
+ Axis::V => move_vertically_visual(
+ slice,
+ range,
+ direction,
+ amount,
+ Movement::Move,
+ &TextFormat::default(),
+ &mut TextAnnotations::default(),
+ ),
};
assert_eq!(coords_at_pos(slice, range.head), coordinates.into());
assert_eq!(range.head, range.anchor);
@@ -638,8 +780,24 @@ mod test {
for ((axis, direction, amount), coordinates) in moves_and_expected_coordinates {
range = match axis {
- Axis::H => move_horizontally(slice, range, direction, amount, Movement::Move, 0),
- Axis::V => move_vertically(slice, range, direction, amount, Movement::Move, 4),
+ Axis::H => move_horizontally(
+ slice,
+ range,
+ direction,
+ amount,
+ Movement::Move,
+ &TextFormat::default(),
+ &mut TextAnnotations::default(),
+ ),
+ Axis::V => move_vertically_visual(
+ slice,
+ range,
+ direction,
+ amount,
+ Movement::Move,
+ &TextFormat::default(),
+ &mut TextAnnotations::default(),
+ ),
};
assert_eq!(coords_at_pos(slice, range.head), coordinates.into());
assert_eq!(range.head, range.anchor);
diff --git a/helix-core/src/position.rs b/helix-core/src/position.rs
index f456eb98..7b8dc326 100644
--- a/helix-core/src/position.rs
+++ b/helix-core/src/position.rs
@@ -1,9 +1,11 @@
-use std::borrow::Cow;
+use std::{borrow::Cow, cmp::Ordering};
use crate::{
chars::char_is_line_ending,
+ doc_formatter::{DocumentFormatter, TextFormat},
graphemes::{ensure_grapheme_boundary_prev, grapheme_width, RopeGraphemes},
line_ending::line_end_char_index,
+ text_annotations::TextAnnotations,
RopeSlice,
};
@@ -73,6 +75,13 @@ pub fn coords_at_pos(text: RopeSlice, pos: usize) -> Position {
/// 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.
+///
+/// This function should be used very rarely. Usually `visual_offset_from_anchor`
+/// or `visual_offset_from_block` is preferable. However when you want to compute the
+/// actual visual row/column in the text (not what is actually shown on screen)
+/// then you should use this function. For example aligning text should ignore virtual
+/// text and softwrap.
+#[deprecated = "Doesn't account for softwrap or decorations, use visual_offset_from_anchor instead"]
pub fn visual_coords_at_pos(text: RopeSlice, pos: usize, tab_width: usize) -> Position {
let line = text.char_to_line(pos);
@@ -93,6 +102,82 @@ pub fn visual_coords_at_pos(text: RopeSlice, pos: usize, tab_width: usize) -> Po
Position::new(line, col)
}
+/// Returns the visual offset from the start of the first visual line
+/// in the block that contains anchor.
+/// Text is always wrapped at blocks, they usually correspond to
+/// actual line breaks but for very long lines
+/// softwrapping positions are estimated with an O(1) algorithm
+/// to ensure consistent performance for large lines (currently unimplemented)
+///
+/// Usualy you want to use `visual_offset_from_anchor` instead but this function
+/// can be useful (and faster) if
+/// * You already know the visual position of the block
+/// * You only care about the horizontal offset (column) and not the vertical offset (row)
+pub fn visual_offset_from_block(
+ text: RopeSlice,
+ anchor: usize,
+ pos: usize,
+ text_fmt: &TextFormat,
+ annotations: &TextAnnotations,
+) -> (Position, usize) {
+ let mut last_pos = Position::default();
+ let (formatter, block_start) =
+ DocumentFormatter::new_at_prev_checkpoint(text, text_fmt, annotations, anchor);
+ let mut char_pos = block_start;
+
+ for (grapheme, vpos) in formatter {
+ last_pos = vpos;
+ char_pos += grapheme.doc_chars();
+
+ if char_pos > pos {
+ return (last_pos, block_start);
+ }
+ }
+
+ (last_pos, block_start)
+}
+
+/// Returns the visual offset from the start of the visual line
+/// that contains anchor.
+pub fn visual_offset_from_anchor(
+ text: RopeSlice,
+ anchor: usize,
+ pos: usize,
+ text_fmt: &TextFormat,
+ annotations: &TextAnnotations,
+ max_rows: usize,
+) -> Option<(Position, usize)> {
+ let (formatter, block_start) =
+ DocumentFormatter::new_at_prev_checkpoint(text, text_fmt, annotations, anchor);
+ let mut char_pos = block_start;
+ let mut anchor_line = None;
+ let mut last_pos = Position::default();
+
+ for (grapheme, vpos) in formatter {
+ last_pos = vpos;
+ char_pos += grapheme.doc_chars();
+
+ if char_pos > anchor && anchor_line.is_none() {
+ anchor_line = Some(last_pos.row);
+ }
+ if char_pos > pos {
+ last_pos.row -= anchor_line.unwrap();
+ return Some((last_pos, block_start));
+ }
+
+ if let Some(anchor_line) = anchor_line {
+ if vpos.row >= anchor_line + max_rows {
+ return None;
+ }
+ }
+ }
+
+ let anchor_line = anchor_line.unwrap_or(last_pos.row);
+ last_pos.row -= anchor_line;
+
+ Some((last_pos, block_start))
+}
+
/// Convert (line, column) coordinates to a character index.
///
/// If the `line` coordinate is beyond the end of the file, the EOF
@@ -140,6 +225,11 @@ pub fn pos_at_coords(text: RopeSlice, coords: Position, limit_before_line_ending
/// If the `column` coordinate is past the end of the given line, the
/// line-end position (in this case, just before the line ending
/// character) will be returned.
+/// This function should be used very rarely. Usually `char_idx_at_visual_offset` is preferable.
+/// However when you want to compute a char position from the visual row/column in the text
+/// (not what is actually shown on screen) then you should use this function.
+/// For example aligning text should ignore virtual text and softwrap.
+#[deprecated = "Doesn't account for softwrap or decorations, use char_idx_at_visual_offset instead"]
pub fn pos_at_visual_coords(text: RopeSlice, coords: Position, tab_width: usize) -> usize {
let Position { mut row, col } = coords;
row = row.min(text.len_lines() - 1);
@@ -169,6 +259,120 @@ pub fn pos_at_visual_coords(text: RopeSlice, coords: Position, tab_width: usize)
line_start + col_char_offset
}
+/// Returns the char index on the visual line `row_offset` below the visual line of
+/// the provided char index `anchor` that is closest to the supplied visual `column`.
+///
+/// If the targeted visual line is entirely covered by virtual text the last
+/// char position before the virtual text and a virtual offset is returned instead.
+///
+/// If no (text) grapheme starts at exactly at the specified column the
+/// start of the grapheme to the left is returned. If there is no grapheme
+/// to the left (for example if the line starts with virtual text) then the positiong
+/// of the next grapheme to the right is returned.
+///
+/// If the `line` coordinate is beyond the end of the file, the EOF
+/// position will be returned.
+///
+/// If the `column` coordinate is past the end of the given line, the
+/// line-end position (in this case, just before the line ending
+/// character) will be returned.
+///
+/// # Returns
+///
+/// `(real_char_idx, virtual_lines)`
+///
+/// The nearest character idx "closest" (see above) to the specified visual offset
+/// on the visual line is returned if the visual line contains any text:
+/// If the visual line at the specified offset is a virtual line generated by a `LineAnnotation`
+/// the previous char_index is returned, together with the remaining vertical offset (`virtual_lines`)
+pub fn char_idx_at_visual_offset<'a>(
+ text: RopeSlice<'a>,
+ mut anchor: usize,
+ mut row_offset: isize,
+ column: usize,
+ text_fmt: &TextFormat,
+ annotations: &TextAnnotations,
+) -> (usize, usize) {
+ // convert row relative to visual line containing anchor to row relative to a block containing anchor (anchor may change)
+ loop {
+ let (visual_pos_in_block, block_char_offset) =
+ visual_offset_from_block(text, anchor, anchor, text_fmt, annotations);
+ row_offset += visual_pos_in_block.row as isize;
+ anchor = block_char_offset;
+ if row_offset >= 0 {
+ break;
+ }
+
+ if block_char_offset == 0 {
+ row_offset = 0;
+ break;
+ }
+ // the row_offset is negative so we need to look at the previous block
+ // set the anchor to the last char before the current block
+ // this char index is also always a line earlier so increase the row_offset by 1
+ anchor -= 1;
+ row_offset += 1;
+ }
+
+ char_idx_at_visual_block_offset(
+ text,
+ anchor,
+ row_offset as usize,
+ column,
+ text_fmt,
+ annotations,
+ )
+}
+
+/// This function behaves the same as `char_idx_at_visual_offset`, except that
+/// the vertical offset `row` is always computed relative to the block that contains `anchor`
+/// instead of the visual line that contains `anchor`.
+/// Usually `char_idx_at_visual_offset` is more useful but this function can be
+/// used in some situations as an optimization when `visual_offset_from_block` was used
+///
+/// # Returns
+///
+/// `(real_char_idx, virtual_lines)`
+///
+/// See `char_idx_at_visual_offset` for details
+pub fn char_idx_at_visual_block_offset(
+ text: RopeSlice,
+ anchor: usize,
+ row: usize,
+ column: usize,
+ text_fmt: &TextFormat,
+ annotations: &TextAnnotations,
+) -> (usize, usize) {
+ let (formatter, mut char_idx) =
+ DocumentFormatter::new_at_prev_checkpoint(text, text_fmt, annotations, anchor);
+ let mut last_char_idx = char_idx;
+ let mut last_char_idx_on_line = None;
+ let mut last_row = 0;
+ for (grapheme, grapheme_pos) in formatter {
+ match grapheme_pos.row.cmp(&row) {
+ Ordering::Equal => {
+ if grapheme_pos.col + grapheme.width() > column {
+ if !grapheme.is_virtual() {
+ return (char_idx, 0);
+ } else if let Some(char_idx) = last_char_idx_on_line {
+ return (char_idx, 0);
+ }
+ } else if !grapheme.is_virtual() {
+ last_char_idx_on_line = Some(char_idx)
+ }
+ }
+ Ordering::Greater => return (last_char_idx, row - last_row),
+ _ => (),
+ }
+
+ last_char_idx = char_idx;
+ last_row = grapheme_pos.row;
+ char_idx += grapheme.doc_chars();
+ }
+
+ (char_idx, 0)
+}
+
#[cfg(test)]
mod test {
use super::*;
@@ -228,6 +432,7 @@ mod test {
}
#[test]
+ #[allow(deprecated)]
fn test_visual_coords_at_pos() {
let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
let slice = text.slice(..);
@@ -276,6 +481,130 @@ mod test {
}
#[test]
+ fn test_visual_off_from_block() {
+ let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
+ let slice = text.slice(..);
+ let annot = TextAnnotations::default();
+ let text_fmt = TextFormat::default();
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 0, &text_fmt, &annot).0,
+ (0, 0).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 5, &text_fmt, &annot).0,
+ (0, 5).into()
+ ); // position on \n
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 6, &text_fmt, &annot).0,
+ (1, 0).into()
+ ); // position on w
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 7, &text_fmt, &annot).0,
+ (1, 1).into()
+ ); // position on o
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 10, &text_fmt, &annot).0,
+ (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_offset_from_block(slice, 0, 0, &text_fmt, &annot).0,
+ (0, 0).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 1, &text_fmt, &annot).0,
+ (0, 2).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 2, &text_fmt, &annot).0,
+ (0, 4).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 3, &text_fmt, &annot).0,
+ (0, 6).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 4, &text_fmt, &annot).0,
+ (0, 8).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 5, &text_fmt, &annot).0,
+ (0, 10).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 6, &text_fmt, &annot).0,
+ (1, 0).into()
+ );
+
+ // Test with grapheme clusters.
+ let text = Rope::from("a̐éö̲\r\n");
+ let slice = text.slice(..);
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 0, &text_fmt, &annot).0,
+ (0, 0).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 2, &text_fmt, &annot).0,
+ (0, 1).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 4, &text_fmt, &annot).0,
+ (0, 2).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 7, &text_fmt, &annot).0,
+ (0, 3).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 9, &text_fmt, &annot).0,
+ (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_offset_from_block(slice, 0, 0, &text_fmt, &annot).0,
+ (0, 0).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 2, &text_fmt, &annot).0,
+ (0, 2).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 3, &text_fmt, &annot).0,
+ (0, 3).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 5, &text_fmt, &annot).0,
+ (0, 5).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 6, &text_fmt, &annot).0,
+ (1, 0).into()
+ );
+
+ // Test with tabs.
+ let text = Rope::from("\tHello\n");
+ let slice = text.slice(..);
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 0, &text_fmt, &annot).0,
+ (0, 0).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 1, &text_fmt, &annot).0,
+ (0, 4).into()
+ );
+ assert_eq!(
+ visual_offset_from_block(slice, 0, 2, &text_fmt, &annot).0,
+ (0, 5).into()
+ );
+ }
+ #[test]
fn test_pos_at_coords() {
let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
let slice = text.slice(..);
@@ -341,6 +670,7 @@ mod test {
}
#[test]
+ #[allow(deprecated)]
fn test_pos_at_visual_coords() {
let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ");
let slice = text.slice(..);
@@ -405,4 +735,100 @@ mod test {
assert_eq!(pos_at_visual_coords(slice, (0, 10).into(), 4), 0);
assert_eq!(pos_at_visual_coords(slice, (10, 10).into(), 4), 0);
}
+
+ #[test]
+ fn test_char_idx_at_visual_row_offset() {
+ let text = Rope::from("ḧëḷḷö\nẅöṛḷḋ\nfoo");
+ let slice = text.slice(..);
+ let mut text_fmt = TextFormat::default();
+ for i in 0isize..3isize {
+ for j in -2isize..=2isize {
+ if !(0..3).contains(&(i + j)) {
+ continue;
+ }
+ println!("{i} {j}");
+ assert_eq!(
+ char_idx_at_visual_offset(
+ slice,
+ slice.line_to_char(i as usize),
+ j,
+ 3,
+ &text_fmt,
+ &TextAnnotations::default(),
+ )
+ .0,
+ slice.line_to_char((i + j) as usize) + 3
+ );
+ }
+ }
+
+ text_fmt.soft_wrap = true;
+ let mut softwrapped_text = "foo ".repeat(10);
+ softwrapped_text.push('\n');
+ let last_char = softwrapped_text.len() - 1;
+
+ let text = Rope::from(softwrapped_text.repeat(3));
+ let slice = text.slice(..);
+ assert_eq!(
+ char_idx_at_visual_offset(
+ slice,
+ last_char,
+ 0,
+ 0,
+ &text_fmt,
+ &TextAnnotations::default(),
+ )
+ .0,
+ 32
+ );
+ assert_eq!(
+ char_idx_at_visual_offset(
+ slice,
+ last_char,
+ -1,
+ 0,
+ &text_fmt,
+ &TextAnnotations::default(),
+ )
+ .0,
+ 16
+ );
+ assert_eq!(
+ char_idx_at_visual_offset(
+ slice,
+ last_char,
+ -2,
+ 0,
+ &text_fmt,
+ &TextAnnotations::default(),
+ )
+ .0,
+ 0
+ );
+ assert_eq!(
+ char_idx_at_visual_offset(
+ slice,
+ softwrapped_text.len() + last_char,
+ -2,
+ 0,
+ &text_fmt,
+ &TextAnnotations::default(),
+ )
+ .0,
+ softwrapped_text.len()
+ );
+
+ assert_eq!(
+ char_idx_at_visual_offset(
+ slice,
+ softwrapped_text.len() + last_char,
+ -5,
+ 0,
+ &text_fmt,
+ &TextAnnotations::default(),
+ )
+ .0,
+ 0
+ );
+ }
}
diff --git a/helix-core/src/selection.rs b/helix-core/src/selection.rs
index ffba46ab..7817618f 100644
--- a/helix-core/src/selection.rs
+++ b/helix-core/src/selection.rs
@@ -53,7 +53,9 @@ pub struct Range {
pub anchor: usize,
/// The head of the range, moved when extending.
pub head: usize,
- pub horiz: Option<u32>,
+ /// The previous visual offset (softwrapped lines and columns) from
+ /// the start of the line
+ pub old_visual_position: Option<(u32, u32)>,
}
impl Range {
@@ -61,7 +63,7 @@ impl Range {
Self {
anchor,
head,
- horiz: None,
+ old_visual_position: None,
}
}
@@ -127,7 +129,7 @@ impl Range {
Self {
anchor: self.head,
head: self.anchor,
- horiz: self.horiz,
+ old_visual_position: self.old_visual_position,
}
}
@@ -185,7 +187,7 @@ impl Range {
Self {
anchor,
head,
- horiz: None,
+ old_visual_position: None,
}
}
@@ -198,13 +200,13 @@ impl Range {
Self {
anchor: self.anchor.min(from),
head: self.head.max(to),
- horiz: None,
+ old_visual_position: None,
}
} else {
Self {
anchor: self.anchor.max(to),
head: self.head.min(from),
- horiz: None,
+ old_visual_position: None,
}
}
}
@@ -219,13 +221,13 @@ impl Range {
Range {
anchor: self.anchor.max(other.anchor),
head: self.head.min(other.head),
- horiz: None,
+ old_visual_position: None,
}
} else {
Range {
anchor: self.from().min(other.from()),
head: self.to().max(other.to()),
- horiz: None,
+ old_visual_position: None,
}
}
}
@@ -279,8 +281,8 @@ impl Range {
Range {
anchor: new_anchor,
head: new_head,
- horiz: if new_anchor == self.anchor {
- self.horiz
+ old_visual_position: if new_anchor == self.anchor {
+ self.old_visual_position
} else {
None
},
@@ -306,7 +308,7 @@ impl Range {
Range {
anchor: self.anchor,
head: next_grapheme_boundary(slice, self.head),
- horiz: self.horiz,
+ old_visual_position: self.old_visual_position,
}
} else {
*self
@@ -378,7 +380,7 @@ impl From<(usize, usize)> for Range {
Self {
anchor,
head,
- horiz: None,
+ old_visual_position: None,
}
}
}
@@ -482,7 +484,7 @@ impl Selection {
ranges: smallvec![Range {
anchor,
head,
- horiz: None
+ old_visual_position: None
}],
primary_index: 0,
}
@@ -566,9 +568,9 @@ impl Selection {
}
/// Takes a closure and maps each `Range` over the closure.
- pub fn transform<F>(mut self, f: F) -> Self
+ pub fn transform<F>(mut self, mut f: F) -> Self
where
- F: Fn(Range) -> Range,
+ F: FnMut(Range) -> Range,
{
for range in self.ranges.iter_mut() {
*range = f(*range)
diff --git a/helix-core/src/text_annotations.rs b/helix-core/src/text_annotations.rs
new file mode 100644
index 00000000..1956f6b5
--- /dev/null
+++ b/helix-core/src/text_annotations.rs
@@ -0,0 +1,271 @@
+use std::cell::Cell;
+use std::convert::identity;
+use std::ops::Range;
+use std::rc::Rc;
+
+use crate::syntax::Highlight;
+use crate::Tendril;
+
+/// An inline annotation is continuous text shown
+/// on the screen before the grapheme that starts at
+/// `char_idx`
+#[derive(Debug, Clone)]
+pub struct InlineAnnotation {
+ pub text: Tendril,
+ pub char_idx: usize,
+}
+
+/// Represents a **single Grapheme** that is part of the document
+/// that start at `char_idx` that will be replaced with
+/// a different `grapheme`.
+/// If `grapheme` contains multiple graphemes the text
+/// will render incorrectly.
+/// If you want to overlay multiple graphemes simply
+/// use multiple `Overlays`.
+///
+/// # Examples
+///
+/// The following examples are valid overlays for the following text:
+///
+/// `aX͎̊͢͜͝͡bc`
+///
+/// ```
+/// use helix_core::text_annotations::Overlay;
+///
+/// // replaces a
+/// Overlay {
+/// char_idx: 0,
+/// grapheme: "X".into(),
+/// };
+///
+/// // replaces X͎̊͢͜͝͡
+/// Overlay{
+/// char_idx: 1,
+/// grapheme: "\t".into(),
+/// };
+///
+/// // replaces b
+/// Overlay{
+/// char_idx: 6,
+/// grapheme: "X̢̢̟͖̲͌̋̇͑͝".into(),
+/// };
+/// ```
+///
+/// The following examples are invalid uses
+///
+/// ```
+/// use helix_core::text_annotations::Overlay;
+///
+/// // overlay is not aligned at grapheme boundary
+/// Overlay{
+/// char_idx: 3,
+/// grapheme: "x".into(),
+/// };
+///
+/// // overlay contains multiple graphemes
+/// Overlay{
+/// char_idx: 0,
+/// grapheme: "xy".into(),
+/// };
+/// ```
+#[derive(Debug, Clone)]
+pub struct Overlay {
+ pub char_idx: usize,
+ pub grapheme: Tendril,
+}
+
+/// Line annotations allow for virtual text between normal
+/// text lines. They cause `height` empty lines to be inserted
+/// below the document line that contains `anchor_char_idx`.
+///
+/// These lines can be filled with text in the rendering code
+/// as their contents have no effect beyond visual appearance.
+///
+/// To insert a line after a document line simply set
+/// `anchor_char_idx` to `doc.line_to_char(line_idx)`
+#[derive(Debug, Clone)]
+pub struct LineAnnotation {
+ pub anchor_char_idx: usize,
+ pub height: usize,
+}
+
+#[derive(Debug)]
+struct Layer<A, M> {
+ annotations: Rc<[A]>,
+ current_index: Cell<usize>,
+ metadata: M,
+}
+
+impl<A, M: Clone> Clone for Layer<A, M> {
+ fn clone(&self) -> Self {
+ Layer {
+ annotations: self.annotations.clone(),
+ current_index: self.current_index.clone(),
+ metadata: self.metadata.clone(),
+ }
+ }
+}
+
+impl<A, M> Layer<A, M> {
+ pub fn reset_pos(&self, char_idx: usize, get_char_idx: impl Fn(&A) -> usize) {
+ let new_index = self
+ .annotations
+ .binary_search_by_key(&char_idx, get_char_idx)
+ .unwrap_or_else(identity);
+
+ self.current_index.set(new_index);
+ }
+
+ pub fn consume(&self, char_idx: usize, get_char_idx: impl Fn(&A) -> usize) -> Option<&A> {
+ let annot = self.annotations.get(self.current_index.get())?;
+ debug_assert!(get_char_idx(annot) >= char_idx);
+ if get_char_idx(annot) == char_idx {
+ self.current_index.set(self.current_index.get() + 1);
+ Some(annot)
+ } else {
+ None
+ }
+ }
+}
+
+impl<A, M> From<(Rc<[A]>, M)> for Layer<A, M> {
+ fn from((annotations, metadata): (Rc<[A]>, M)) -> Layer<A, M> {
+ Layer {
+ annotations,
+ current_index: Cell::new(0),
+ metadata,
+ }
+ }
+}
+
+fn reset_pos<A, M>(layers: &[Layer<A, M>], pos: usize, get_pos: impl Fn(&A) -> usize) {
+ for layer in layers {
+ layer.reset_pos(pos, &get_pos)
+ }
+}
+
+/// Annotations that change that is displayed when the document is render.
+/// Also commonly called virtual text.
+#[derive(Default, Debug, Clone)]
+pub struct TextAnnotations {
+ inline_annotations: Vec<Layer<InlineAnnotation, Option<Highlight>>>,
+ overlays: Vec<Layer<Overlay, Option<Highlight>>>,
+ line_annotations: Vec<Layer<LineAnnotation, ()>>,
+}
+
+impl TextAnnotations {
+ /// Prepare the TextAnnotations for iteration starting at char_idx
+ pub fn reset_pos(&self, char_idx: usize) {
+ reset_pos(&self.inline_annotations, char_idx, |annot| annot.char_idx);
+ reset_pos(&self.overlays, char_idx, |annot| annot.char_idx);
+ reset_pos(&self.line_annotations, char_idx, |annot| {
+ annot.anchor_char_idx
+ });
+ }
+
+ pub fn collect_overlay_highlights(
+ &self,
+ char_range: Range<usize>,
+ ) -> Vec<(usize, Range<usize>)> {
+ let mut highlights = Vec::new();
+ self.reset_pos(char_range.start);
+ for char_idx in char_range {
+ if let Some((_, Some(highlight))) = self.overlay_at(char_idx) {
+ // we don't know the number of chars the original grapheme takes
+ // however it doesn't matter as highlight bounderies are automatically
+ // aligned to grapheme boundaries in the rendering code
+ highlights.push((highlight.0, char_idx..char_idx + 1))
+ }
+ }
+
+ highlights
+ }
+
+ /// Add new inline annotations.
+ ///
+ /// The annotations grapheme will be rendered with `highlight`
+ /// patched on top of `ui.text`.
+ ///
+ /// The annotations **must be sorted** by their `char_idx`.
+ /// Multiple annotations with the same `char_idx` are allowed,
+ /// they will be display in the order that they are present in the layer.
+ ///
+ /// If multiple layers contain annotations at the same position
+ /// the annotations that belong to the layers added first will be shown first.
+ pub fn add_inline_annotations(
+ &mut self,
+ layer: Rc<[InlineAnnotation]>,
+ highlight: Option<Highlight>,
+ ) -> &mut Self {
+ self.inline_annotations.push((layer, highlight).into());
+ self
+ }
+
+ /// Add new grapheme overlays.
+ ///
+ /// The overlayed grapheme will be rendered with `highlight`
+ /// patched on top of `ui.text`.
+ ///
+ /// The overlays **must be sorted** by their `char_idx`.
+ /// Multiple overlays with the same `char_idx` **are allowed**.
+ ///
+ /// If multiple layers contain overlay at the same position
+ /// the overlay from the layer added last will be show.
+ pub fn add_overlay(&mut self, layer: Rc<[Overlay]>, highlight: Option<Highlight>) -> &mut Self {
+ self.overlays.push((layer, highlight).into());
+ self
+ }
+
+ /// Add new annotation lines.
+ ///
+ /// The line annotations **must be sorted** by their `char_idx`.
+ /// Multiple line annotations with the same `char_idx` **are not allowed**.
+ pub fn add_line_annotation(&mut self, layer: Rc<[LineAnnotation]>) -> &mut Self {
+ self.line_annotations.push((layer, ()).into());
+ self
+ }
+
+ /// Removes all line annotations, useful for vertical motions
+ /// so that virtual text lines are automatically skipped.
+ pub fn clear_line_annotations(&mut self) {
+ self.line_annotations.clear();
+ }
+
+ pub(crate) fn next_inline_annotation_at(
+ &self,
+ char_idx: usize,
+ ) -> Option<(&InlineAnnotation, Option<Highlight>)> {
+ self.inline_annotations.iter().find_map(|layer| {
+ let annotation = layer.consume(char_idx, |annot| annot.char_idx)?;
+ Some((annotation, layer.metadata))
+ })
+ }
+
+ pub(crate) fn overlay_at(&self, char_idx: usize) -> Option<(&Overlay, Option<Highlight>)> {
+ let mut overlay = None;
+ for layer in &self.overlays {
+ while let Some(new_overlay) = layer.consume(char_idx, |annot| annot.char_idx) {
+ overlay = Some((new_overlay, layer.metadata));
+ }
+ }
+ overlay
+ }
+
+ pub(crate) fn annotation_lines_at(&self, char_idx: usize) -> usize {
+ self.line_annotations
+ .iter()
+ .map(|layer| {
+ let mut lines = 0;
+ while let Some(annot) = layer.annotations.get(layer.current_index.get()) {
+ if annot.anchor_char_idx == char_idx {
+ layer.current_index.set(layer.current_index.get() + 1);
+ lines += annot.height
+ } else {
+ break;
+ }
+ }
+ lines
+ })
+ .sum()
+ }
+}
generated by cgit v1.2.3-70-g09d2 (git 2.47.1) at 2024-12-23 09:09:17 +0000