Automatically add spacing between CJK and Latin characters (#2334)

2 files changed, 26 insertions, 3 deletions

diff --git a/crates/typst-library/src/text/mod.rs b/crates/typst-library/src/text/mod.rs
index 748d71ad..4d2f5c63 100644
--- a/crates/typst-library/src/text/mod.rs
+++ b/crates/typst-library/src/text/mod.rs
@@ -19,6 +19,7 @@ pub use self::shift::*;
 use rustybuzz::Tag;
 use ttf_parser::Rect;
 use typst::diag::{bail, error, SourceResult};
+use typst::eval::Never;
 use typst::font::{Font, FontStretch, FontStyle, FontWeight, VerticalFontMetric};
 
 use crate::layout::ParElem;
@@ -224,6 +225,17 @@ pub struct TextElem {
     #[default(Rel::one())]
     pub spacing: Rel<Length>,
 
+    /// Whether to automatically insert spacing between CJK and Latin characters.
+    ///
+    /// ```example
+    /// #set text(cjk-latin-spacing: auto)
+    /// 第4章介绍了基本的API。
+    ///
+    /// #set text(cjk-latin-spacing: none)
+    /// 第4章介绍了基本的API。
+    /// ```
+    pub cjk_latin_spacing: Smart<Option<Never>>,
+
     /// An amount to shift the text baseline by.
     ///
     /// ```example
diff --git a/crates/typst-library/src/text/shaping.rs b/crates/typst-library/src/text/shaping.rs
index df30a9c8..1b7d9e09 100644
--- a/crates/typst-library/src/text/shaping.rs
+++ b/crates/typst-library/src/text/shaping.rs
@@ -100,9 +100,7 @@ impl ShapedGlyph {
     }
 
     pub fn is_cjk_script(&self) -> bool {
-        use Script::*;
-        // U+30FC: Katakana-Hiragana Prolonged Sound Mark
-        matches!(self.c.script(), Hiragana | Katakana | Han) || self.c == '\u{30FC}'
+        char_is_cjk_script(self.c)
     }
 
     pub fn is_cjk_punctuation(&self) -> bool {
@@ -151,6 +149,13 @@ impl ShapedGlyph {
         matches!(self.c, '\u{30FB}')
     }
 
+    /// Whether the glyph is a western letter or number.
+    pub fn is_letter_or_number(&self) -> bool {
+        matches!(self.c.script(), Script::Latin | Script::Greek | Script::Cyrillic)
+            || matches!(self.c, '#' | '$' | '%' | '&')
+            || self.c.is_ascii_digit()
+    }
+
     pub fn base_adjustability(&self, gb_style: bool) -> Adjustability {
         let width = self.x_advance;
         if self.is_space() {
@@ -958,6 +963,12 @@ fn language(styles: StyleChain) -> rustybuzz::Language {
     rustybuzz::Language::from_str(&bcp).unwrap()
 }
 
+pub fn char_is_cjk_script(c: char) -> bool {
+    use Script::*;
+    // U+30FC: Katakana-Hiragana Prolonged Sound Mark
+    matches!(c.script(), Hiragana | Katakana | Han) || c == '\u{30FC}'
+}
+
 /// Returns true if all glyphs in `glyphs` have ranges within the range `range`.
 #[cfg(debug_assertions)]
 fn assert_all_glyphs_in_range(glyphs: &[ShapedGlyph], text: &str, range: Range<usize>) {