Add CJK unicode into built-in rules.

Make this change to add `CJK`, `HAN`, `HANGUL`, `KATAKANA`, `HIRAGANA` to built-in rules.

https://unicode.org/faq/han_cjk.html

- Chinese - `HAN`
- Japanese - `KATAKANA`, `HIRAGANA`
- Korean - `HANGUL`

So we can easy to to match the CJK chars.

meta/src/lib.rs

@@ -162,4 +162,10 @@ pub static UNICODE_PROPERTY_NAMES: &[&str] = &[
     "TITLECASE_LETTER",
     "UNASSIGNED",
     "UPPERCASE_LETTER",
+    /* Script */
+    "CJK",
+    "HAN",
+    "HANGUL",
+    "HIRAGANA",
+    "KATAKANA",
 ];

pest/src/unicode/binary.rs

@@ -4,7 +4,7 @@
 //
 // Unicode version: 15.0.0.
 //
-// ucd-generate 0.2.13 is available on crates.io.
+// ucd-generate 0.2.15 is available on crates.io.
 
 pub const BY_NAME: &'static [(&'static str, &'static ::ucd_trie::TrieSet)] = &[
   ("ASCII_Hex_Digit", ASCII_HEX_DIGIT), ("Alphabetic", ALPHABETIC),

pest/src/unicode/category.rs

@@ -4,7 +4,7 @@
 //
 // Unicode version: 15.0.0.
 //
-// ucd-generate 0.2.13 is available on crates.io.
+// ucd-generate 0.2.15 is available on crates.io.
 
 pub const BY_NAME: &'static [(&'static str, &'static ::ucd_trie::TrieSet)] = &[
   ("Cased_Letter", CASED_LETTER), ("Close_Punctuation", CLOSE_PUNCTUATION),

pest/src/unicode/mod.rs

@@ -49,6 +49,11 @@ char_property_functions! {
         PRIVATE_USE, PUNCTUATION, SEPARATOR, SPACE_SEPARATOR, SPACING_MARK, SURROGATE, SYMBOL,
         TITLECASE_LETTER, UNASSIGNED, UPPERCASE_LETTER,
     ];
+
+    mod script;
+    [
+        HAN, KATAKANA, HIRAGANA, HANGUL,
+    ];
 }
 
 pub fn by_name(name: &str) -> Option<Box<dyn Fn(char) -> bool>> {
@@ -64,5 +69,20 @@ pub fn by_name(name: &str) -> Option<Box<dyn Fn(char) -> bool>> {
         }
     }
 
+    for property in script::BY_NAME {
+        if name == property.0.to_uppercase() {
+            return Some(Box::new(move |c| property.1.contains_char(c)));
+        }
+    }
+
+    if name == "CJK" {
+        return Some(Box::new(|c| {
+            script::HAN.contains_char(c)
+                || script::HANGUL.contains_char(c)
+                || script::KATAKANA.contains_char(c)
+                || script::HIRAGANA.contains_char(c)
+        }));
+    }
+
     None
 }