Make chinese justification less bad

4 files changed, 34 insertions, 19 deletions

diff --git a/src/library/text/par.rs b/src/library/text/par.rs
index 67357135..1695e010 100644
--- a/src/library/text/par.rs
+++ b/src/library/text/par.rs
@@ -310,15 +310,15 @@ impl<'a> Line<'a> {
         self.items().nth(index)
     }
 
-    // How many spaces the line contains.
-    fn spaces(&self) -> usize {
-        let mut spaces = 0;
+    // How many justifiable glyphs the line contains.
+    fn justifiables(&self) -> usize {
+        let mut count = 0;
         for item in self.items() {
             if let ParItem::Text(shaped) = item {
-                spaces += shaped.spaces();
+                count += shaped.justifiables();
             }
         }
-        spaces
+        count
     }
 
     /// How much of the line is stretchable spaces.
@@ -509,11 +509,11 @@ fn linebreak_optimized<'a>(
     // Cost parameters.
     const HYPH_COST: Cost = 0.5;
     const CONSECUTIVE_DASH_COST: Cost = 30.0;
-    const MAX_COST: Cost = 10_000.0;
+    const MAX_COST: Cost = 1_000_000.0;
     const MIN_COST: Cost = -MAX_COST;
     const MIN_RATIO: f64 = -0.15;
 
-    // Density parameters.
+    let em = styles.get(TextNode::SIZE);
     let justify = styles.get(ParNode::JUSTIFY);
 
     // Dynamic programming table.
@@ -537,11 +537,15 @@ fn linebreak_optimized<'a>(
 
             // Determine how much the line's spaces would need to be stretched
             // to make it the desired width.
-            let mut ratio = (width - attempt.size.x) / attempt.stretch();
+            let delta = width - attempt.size.x;
+            let mut ratio = delta / attempt.stretch();
             if ratio.is_infinite() {
-                ratio = ratio.signum() * MAX_COST;
+                ratio = delta / (em / 2.0);
             }
 
+            // At some point, it doesn't matter any more.
+            ratio = ratio.min(10.0);
+
             // Determine the cost of the line.
             let mut cost = if ratio < if justify { MIN_RATIO } else { 0.0 } {
                 // The line is overfull. This is the case if
@@ -863,9 +867,9 @@ fn commit(
             && line.range.end < line.bidi.text.len()
             && line.fr.is_zero())
     {
-        let spaces = line.spaces();
-        if spaces > 0 {
-            justification = remaining / spaces as f64;
+        let justifiables = line.justifiables();
+        if justifiables > 0 {
+            justification = remaining / justifiables as f64;
             remaining = Length::zero();
         }
     }
diff --git a/src/library/text/shaping.rs b/src/library/text/shaping.rs
index 66936792..d398e56d 100644
--- a/src/library/text/shaping.rs
+++ b/src/library/text/shaping.rs
@@ -50,10 +50,15 @@ pub struct ShapedGlyph {
 }
 
 impl ShapedGlyph {
-    /// Whether the glyph is a justifiable space.
+    /// Whether the glyph is a space.
     pub fn is_space(&self) -> bool {
         self.c == ' '
     }
+
+    /// Whether the glyph is justifiable.
+    pub fn is_justifiable(&self) -> bool {
+        matches!(self.c, ' ' | '，' | '　' | '。' | '、')
+    }
 }
 
 /// A side you can go toward.
@@ -68,7 +73,7 @@ impl<'a> ShapedText<'a> {
     /// Build the shaped text's frame.
     ///
     /// The `justification` defines how much extra advance width each
-    /// [space glyph](ShapedGlyph::is_space) will get.
+    /// [justifiable glyph](ShapedGlyph::is_justifiable) will get.
     pub fn build(&self, fonts: &FontStore, justification: Length) -> Frame {
         let mut offset = Length::zero();
         let mut frame = Frame::new(self.size);
@@ -84,7 +89,7 @@ impl<'a> ShapedText<'a> {
                 .map(|glyph| Glyph {
                     id: glyph.glyph_id,
                     x_advance: glyph.x_advance
-                        + if glyph.is_space() {
+                        + if glyph.is_justifiable() {
                             frame.size.x += justification;
                             Em::from_length(justification, size)
                         } else {
@@ -115,16 +120,16 @@ impl<'a> ShapedText<'a> {
         frame
     }
 
-    /// How many spaces the text contains.
-    pub fn spaces(&self) -> usize {
-        self.glyphs.iter().filter(|g| g.is_space()).count()
+    /// How many justifiable glyphs the text contains.
+    pub fn justifiables(&self) -> usize {
+        self.glyphs.iter().filter(|g| g.is_justifiable()).count()
     }
 
     /// The width of the spaces in the text.
     pub fn stretch(&self) -> Length {
         self.glyphs
             .iter()
-            .filter(|g| g.is_space())
+            .filter(|g| g.is_justifiable())
             .map(|g| g.x_advance)
             .sum::<Em>()
             .resolve(self.styles.get(TextNode::SIZE))
diff --git a/tests/ref/text/justify.png b/tests/ref/text/justify.png
index 90f84bb6..aa01016f 100644
--- a/tests/ref/text/justify.png
+++ b/tests/ref/text/justify.png
diff --git a/tests/typ/text/justify.typ b/tests/typ/text/justify.typ
index eb8feb61..d19249ea 100644
--- a/tests/typ/text/justify.typ
+++ b/tests/typ/text/justify.typ
@@ -19,3 +19,9 @@ First line indents and hyphenation play nicely with justified text.
 #set par(justify: true)
 A B C \
 D
+
+---
+// Test that justificating chinese text is at least a bit sensible.
+#set page(width: 200pt)
+#set par(justify: true)
+中文维基百科使用汉字书写，汉字是汉族或华人的共同文字，是中国大陆、新加坡、马来西亚、台湾、香港、澳门的唯一官方文字或官方文字之一。25.9%，而美国和荷兰则分別占13.7%及8.2%。近年來，中国大陆地区的维基百科编辑者正在迅速增加；