Support adjustment of more CJK punctuation (#2456)

4 files changed, 56 insertions, 11 deletions

diff --git a/crates/typst-library/src/layout/par.rs b/crates/typst-library/src/layout/par.rs
index 5d21c8e9..bb280069 100644
--- a/crates/typst-library/src/layout/par.rs
+++ b/crates/typst-library/src/layout/par.rs
@@ -17,7 +17,7 @@ use crate::math::EquationElem;
 use crate::prelude::*;
 use crate::text::{
     char_is_cjk_script, is_gb_style, shape, LinebreakElem, Quoter, Quotes, ShapedGlyph,
-    ShapedText, SmartquoteElem, SpaceElem, TextElem,
+    ShapedText, SmartquoteElem, SpaceElem, TextElem, BEGIN_PUNCT_PAT, END_PUNCT_PAT,
 };
 
 /// Arranges text, spacing and inline-level elements into a paragraph.
@@ -1287,11 +1287,6 @@ fn line<'a>(
     let end = range.end;
     let mut justify = p.justify && end < p.bidi.text.len() && !mandatory;
 
-    // The CJK punctuation that can appear at the beginning or end of a line.
-    const BEGIN_PUNCT_PAT: &[char] = &['“', '‘', '《', '（', '『', '「'];
-    const END_PUNCT_PAT: &[char] =
-        &['”', '’', '，', '。', '、', '：', '；', '》', '）', '』', '」'];
-
     if range.is_empty() {
         return Line {
             bidi: &p.bidi,
diff --git a/crates/typst-library/src/text/shaping.rs b/crates/typst-library/src/text/shaping.rs
index 39122837..0cfffce0 100644
--- a/crates/typst-library/src/text/shaping.rs
+++ b/crates/typst-library/src/text/shaping.rs
@@ -1001,6 +1001,14 @@ fn assert_glyph_ranges_in_order(glyphs: &[ShapedGlyph], dir: Dir) {
     }
 }
 
+// The CJK punctuation that can appear at the beginning or end of a line.
+pub(crate) const BEGIN_PUNCT_PAT: &[char] =
+    &['“', '‘', '《', '〈', '（', '『', '「', '【', '〖', '〔', '［', '｛'];
+pub(crate) const END_PUNCT_PAT: &[char] = &[
+    '”', '’', '，', '．', '。', '、', '：', '；', '》', '〉', '）', '』', '」', '】',
+    '〗', '〕', '］', '｝', '？', '！',
+];
+
 /// Whether the glyph is a space.
 #[inline]
 fn is_space(c: char) -> bool {
@@ -1035,11 +1043,15 @@ fn is_cjk_left_aligned_punctuation(
         return true;
     }
 
-    if gb_style && matches!(c, '，' | '。' | '、' | '：' | '；') {
+    if gb_style && matches!(c, '，' | '。' | '．' | '、' | '：' | '；' | '！' | '？')
+    {
+        // In GB style, exclamations and question marks are also left aligned and can be adjusted.
+        // Note that they are not adjustable in other styles.
         return true;
     }
 
-    matches!(c, '》' | '）' | '』' | '」')
+    // See appendix A.3 https://www.w3.org/TR/clreq/#tables_of_chinese_punctuation_marks
+    matches!(c, '》' | '）' | '』' | '」' | '】' | '〗' | '〕' | '〉' | '］' | '｝')
 }
 
 /// See <https://www.w3.org/TR/clreq/#punctuation_width_adjustment>
@@ -1054,14 +1066,14 @@ fn is_cjk_right_aligned_punctuation(
     if matches!(c, '“' | '‘') && x_advance + stretchability.0 == Em::one() {
         return true;
     }
-
-    matches!(c, '《' | '（' | '『' | '「')
+    // See appendix A.3 https://www.w3.org/TR/clreq/#tables_of_chinese_punctuation_marks
+    matches!(c, '《' | '（' | '『' | '「' | '【' | '〖' | '〔' | '〈' | '［' | '｛')
 }
 
 /// See <https://www.w3.org/TR/clreq/#punctuation_width_adjustment>
 #[inline]
 fn is_cjk_center_aligned_punctuation(c: char, gb_style: bool) -> bool {
-    if !gb_style && matches!(c, '，' | '。' | '、' | '：' | '；') {
+    if !gb_style && matches!(c, '，' | '。' | '．' | '、' | '：' | '；') {
         return true;
     }
 
diff --git a/tests/ref/layout/cjk-punctuation-adjustment.png b/tests/ref/layout/cjk-punctuation-adjustment.png
new file mode 100644
index 00000000..71179751
--- /dev/null
+++ b/tests/ref/layout/cjk-punctuation-adjustment.png
diff --git a/tests/typ/layout/cjk-punctuation-adjustment.typ b/tests/typ/layout/cjk-punctuation-adjustment.typ
new file mode 100644
index 00000000..0f1f2894
--- /dev/null
+++ b/tests/typ/layout/cjk-punctuation-adjustment.typ
@@ -0,0 +1,38 @@
+#set page(width: 15em)
+
+// In the following example, the space between 》！ and ？ should be squeezed.
+// because zh-CN follows GB style
+#set text(lang: "zh", region: "CN", font: "Noto Serif CJK SC")
+原来，你也玩《原神》！？
+
+// However, in the following example, the space between 》！ and ？ should not be squeezed.
+// because zh-TW does not follow GB style
+#set text(lang: "zh", region: "TW", font: "Noto Serif CJK TC")
+原來，你也玩《原神》！ ？
+---
+
+#set text(lang: "zh", region: "CN", font: "Noto Serif CJK SC")
+《书名〈章节〉》 // the space between 〉 and 》 should be squeezed
+
+〔茸毛〕：很细的毛 // the space between 〕 and ： should be squeezed
+
+---
+#set page(width: 21em)
+#set text(lang: "zh", region: "CN", font: "Noto Serif CJK SC")
+
+// These examples contain extensive use of Chinese punctuation marks,
+// from 《Which parentheses should be used when applying parentheses?》.
+// link: https://archive.md/2bb1N
+
+
+（〔中〕医、〔中〕药、技）系列评审
+
+（长三角［长江三角洲］）（GB/T 16159—2012《汉语拼音正词法基本规则》）
+
+【爱因斯坦（Albert Einstein）】物理学家
+
+〔（2009）民申字第1622号〕
+
+“江南海北长相忆，浅水深山独掩扉。”（［唐］刘长卿《会赦后酬主簿所问》）
+
+参看1378页〖象形文字〗。（《现代汉语词典》修订本）