patch legacy katakana middle dot

closes #2

Unicode 4.1 through Unicode 15 omitted these two characters from ID_Continue
by accident. However, this accident was corrected in Unicode 15.1. Any JS VM
that supports ES6+ but that uses a version of Unicode earlier than 15.1 will
consider these to be a syntax error, so we deliberately omit these characters
from the set of identifiers that are valid in both ES5 and ES6+. For more info
see 2.2 in https://www.unicode.org/L2/L2023/23160-utc176-properties-recs.pdf

generate/src/parse.rs

@@ -50,6 +50,16 @@ pub fn parse_id_properties(ucd_dir: &Path) -> Properties {
         set.extend(lo..=hi);
     }
 
+    // <https://github.com/evanw/esbuild/pull/3424>
+    // Unicode 4.1 through Unicode 15 omitted these two characters from ID_Continue
+    // by accident. However, this accident was corrected in Unicode 15.1. Any JS VM
+    // that supports ES6+ but that uses a version of Unicode earlier than 15.1 will
+    // consider these to be a syntax error, so we deliberately omit these characters
+    // from the set of identifiers that are valid in both ES5 and ES6+. For more info
+    // see 2.2 in https://www.unicode.org/L2/L2023/23160-utc176-properties-recs.pdf
+    properties.id_continue.remove(&0x30FB);
+    properties.id_continue.remove(&0xFF65);
+
     properties
 }
 

tests/compare.rs

@@ -14,6 +14,10 @@ fn compare_all_implementations() {
     let id_continue_roaring = roaring::id_continue_bitmap();
 
     for ch in '\0'..=char::MAX {
+        // See test legacy_katakana_middle_dot_patch in tests/patch.rs
+        if matches!(ch, '・' | '･') {
+            continue;
+        }
         let thought_to_be_start = unicode_id_start::is_id_start(ch);
         let thought_to_be_continue = unicode_id_start::is_id_continue(ch);
 

tests/patch.rs

@@ -0,0 +1,11 @@
+use unicode_id_start::is_id_continue_unicode;
+
+#[test]
+fn legacy_katakana_middle_dot_patch() {
+    // U+30FB KATAKANA MIDDLE DOT
+    // https://util.unicode.org/UnicodeJsps/character.jsp?a=30FB
+    assert!(!is_id_continue_unicode('・'));
+    // U+FF65 HALFWIDTH KATAKANA MIDDLE DOT
+    // https://util.unicode.org/UnicodeJsps/character.jsp?a=FF65
+    assert!(!is_id_continue_unicode('･'));
+}