Merge pull request #124 from tigregalis/iter-bidi-para

Replace use of `str::Lines` with `unicode-bidi`-based iterator

examples/editor-test/src/main.rs

@@ -1,7 +1,8 @@
 // SPDX-License-Identifier: MIT OR Apache-2.0
 
 use cosmic_text::{
-    Action, BorrowedWithFontSystem, Buffer, Color, Edit, Editor, FontSystem, Metrics, SwashCache,
+    Action, BidiParagraphs, BorrowedWithFontSystem, Buffer, Color, Edit, Editor, FontSystem,
+    Metrics, SwashCache,
 };
 use orbclient::{EventOption, Renderer, Window, WindowFlag};
 use std::{env, fs, process, time::Instant};
@@ -83,8 +84,7 @@ fn main() {
 
     let test_start = Instant::now();
 
-    //TODO: support bidi
-    for line in text.lines() {
+    for line in BidiParagraphs::new(&text) {
         log::debug!("Line {:?}", line);
 
         for grapheme in line.graphemes(true) {

src/bidi_para.rs

@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: MIT OR Apache-2.0
+
+use unicode_bidi::{bidi_class, BidiClass, BidiInfo, ParagraphInfo};
+
+/// An iterator over the paragraphs in the input text.
+/// It is equivalent to [`core::str::Lines`] but follows `unicode-bidi` behaviour.
+pub struct BidiParagraphs<'text> {
+    text: &'text str,
+    info: alloc::vec::IntoIter<ParagraphInfo>,
+}
+
+impl<'text> BidiParagraphs<'text> {
+    /// Create an iterator to split the input text into paragraphs
+    /// in accordance with `unicode-bidi` behaviour.
+    pub fn new(text: &'text str) -> Self {
+        let info = BidiInfo::new(text, None);
+        let info = info.paragraphs.into_iter();
+        Self { text, info }
+    }
+}
+
+impl<'text> Iterator for BidiParagraphs<'text> {
+    type Item = &'text str;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        let para = self.info.next()?;
+        let paragraph = &self.text[para.range];
+        // `para.range` includes the newline that splits the line, so remove it if present
+        let mut char_indices = paragraph.char_indices();
+        if let Some(i) = char_indices.next_back().and_then(|(i, c)| {
+            // `BidiClass::B` is a Paragraph_Separator (various newline characters)
+            (bidi_class(c) == BidiClass::B).then_some(i)
+        }) {
+            Some(&paragraph[0..i])
+        } else {
+            Some(paragraph)
+        }
+    }
+}

src/buffer.rs

@@ -9,8 +9,8 @@ use core::{cmp, fmt};
 use unicode_segmentation::UnicodeSegmentation;
 
 use crate::{
-    Attrs, AttrsList, BorrowedWithFontSystem, BufferLine, Color, FontSystem, LayoutGlyph,
-    LayoutLine, ShapeLine, Shaping, Wrap,
+    Attrs, AttrsList, BidiParagraphs, BorrowedWithFontSystem, BufferLine, Color, FontSystem,
+    LayoutGlyph, LayoutLine, ShapeLine, Shaping, Wrap,
 };
 
 /// Current cursor location
@@ -594,7 +594,7 @@ impl Buffer {
         shaping: Shaping,
     ) {
         self.lines.clear();
-        for line in text.lines() {
+        for line in BidiParagraphs::new(text) {
             self.lines.push(BufferLine::new(
                 line.to_string(),
                 AttrsList::new(attrs),

src/lib.rs

@@ -95,6 +95,9 @@ extern crate alloc;
 pub use self::attrs::*;
 mod attrs;
 
+pub use self::bidi_para::*;
+mod bidi_para;
+
 pub use self::buffer::*;
 mod buffer;