Use Unicode punctuation in emphasis flanking rules

Cargo.toml

@@ -20,6 +20,7 @@ codegen-units = 1
 [dependencies]
 html-escape = ">=0.2"
 pyo3 = ">=0.28"
+unicode-properties = "0.1"
 unicode-width = "0.2"
 
 [features]

src/inline.rs

@@ -6,6 +6,7 @@ use crate::entity::decode_entities as decode_html_entities;
 use crate::tagfilter::tagfilter_html;
 use crate::{MathMode, Options};
 use std::collections::{HashMap, HashSet};
+use unicode_properties::{GeneralCategoryGroup, UnicodeGeneralCategory};
 
 const ESCAPED_AMP: char = '\u{E000}';
 
@@ -601,11 +602,16 @@ struct Delimiter {
     active: bool,
 }
 
+// Spec "punctuation character": ASCII punctuation or Unicode general category P.
+fn is_flanking_punct(ch: char) -> bool {
+    ch.is_ascii_punctuation() || ch.general_category_group() == GeneralCategoryGroup::Punctuation
+}
+
 fn delimiter_run_flags(ch: char, before: char, after: char) -> (bool, bool) {
     let before_ws = before == '\0' || before.is_whitespace();
     let after_ws = after == '\0' || after.is_whitespace();
-    let before_punct = before.is_ascii_punctuation();
-    let after_punct = after.is_ascii_punctuation();
+    let before_punct = is_flanking_punct(before);
+    let after_punct = is_flanking_punct(after);
     let left = !after_ws && (!after_punct || before_ws || before_punct);
     let right = !before_ws && (!before_punct || after_ws || after_punct);
     match ch {

tests/test_focused.py

@@ -12,6 +12,12 @@ def test_dialect_fixture():
     expected = (FIX / "dialect.xhtml").read_text()
     assert _norm(to_xhtml(md)) == _norm(expected)
 
+def test_flanking_treats_unicode_punctuation_as_punctuation():
+    html = to_xhtml("(“***{{company_common_name}}***”)")
+    assert "“<em><strong>{{company_common_name}}</strong></em>”" in html
+    html = to_xhtml("“__{{x}}__”")
+    assert "“<strong>{{x}}</strong>”" in html
+
 def test_default_math_mode_is_brackets():
     html = to_xhtml("\\(y\\)\n\n\\[\nx^2\n\\]\n\n$x$")
     assert '<span class="math inline">y</span>' in html