feat: Allow supported boolean Unicode properties in Java

pomsky-lib/src/exprs/char_class.rs

@@ -437,6 +437,16 @@ fn named_class_to_regex_unicode(
                     _ => {}
                 }
                 set.add_prop(RegexProperty::Other(o).negative_item(negative));
+            } else if flavor == RF::Java {
+                if pomsky_syntax::props_supported_in_java().binary_search(&o.as_str()).is_ok() {
+                    set.add_prop(RegexProperty::Other(o).negative_item(negative));
+                } else {
+                    return Err(CompileErrorKind::Unsupported(
+                        Feature::SpecificUnicodeProp,
+                        flavor,
+                    )
+                    .at(span));
+                }
             } else {
                 return Err(CompileErrorKind::Unsupported(Feature::UnicodeProp, flavor).at(span));
             }

pomsky-lib/src/regex/mod.rs

@@ -469,7 +469,6 @@ impl RegexProperty {
             },
             RegexProperty::Other(o) => {
                 if flavor == RegexFlavor::Java {
-                    // Currently disabled since only some boolean properties are supported in Java
                     buf.push_str("Is");
                 }
                 buf.push_str(o.as_str());

pomsky-syntax/JavaSupportedProps.txt

@@ -0,0 +1,20 @@
+Alphabetic
+Assigned
+Control
+Digit
+Emoji
+Emoji_Component
+Emoji_Modifier
+Emoji_Modifier_Base
+Emoji_Presentation
+Extended_Pictographic
+Hex_Digit
+Ideographic
+Join_Control
+Letter
+Lowercase
+Noncharacter_Code_Point
+Punctuation
+Titlecase
+Uppercase
+White_Space

pomsky-syntax/build.rs

@@ -5,6 +5,7 @@ fn main() {
     println!("cargo:rerun-if-changed=PropertyValueAliases.txt");
     println!("cargo:rerun-if-changed=SupportedBooleanProps.txt");
     println!("cargo:rerun-if-changed=DotNetSupportedBlocks.txt");
+    println!("cargo:rerun-if-changed=JavaSupportedProps.txt");
     generate_unicode_data();
 }
 
@@ -19,6 +20,9 @@ fn generate_unicode_data() {
     let [categories, scripts, blocks, bools] = parse_aliases(&aliases, &blocks);
 
     let dotnet_blocks = std::fs::read_to_string("DotNetSupportedBlocks.txt").unwrap();
+    let java_props = std::fs::read_to_string("JavaSupportedProps.txt").unwrap();
+    assert!(dotnet_blocks.lines().is_sorted());
+    assert!(java_props.lines().is_sorted());
 
     let out_dir = std::env::var("OUT_DIR").unwrap();
     let path = std::path::Path::new(&out_dir).join("unicode_data.rs");
@@ -104,14 +108,19 @@ static PARSE_LUT: &[(&str, GroupName)] = &[
 
 static DOTNET_SUPPORTED: &[&str] = &[
 {dotnet_supported}];
+
+static JAVA_SUPPORTED: &[&str] = &[
+{java_supported}];
 ",
             category_enum = generate_enum("Category", &categories, 0, 1),
             script_enum = generate_enum("Script", &scripts, 1, 1),
             block_enum = generate_enum("CodeBlock", &blocks, 1, 1),
             other_enum = generate_enum("OtherProperties", &bools, 1, 1),
             lut = lut.join(",\n    "),
             dotnet_supported =
-                dotnet_blocks.lines().map(|line| format!("    {line:?},\n")).collect::<String>()
+                dotnet_blocks.lines().map(|line| format!("    {line:?},\n")).collect::<String>(),
+            java_supported =
+                java_props.lines().map(|line| format!("    {line:?},\n")).collect::<String>(),
         ),
     )
     .unwrap();

pomsky-syntax/src/exprs/char_class/mod.rs

@@ -85,7 +85,7 @@ mod ascii;
 pub(crate) mod char_group;
 pub(crate) mod unicode;
 
-pub use unicode::{blocks_supported_in_dotnet, list_shorthands};
+pub use unicode::{blocks_supported_in_dotnet, list_shorthands, props_supported_in_java};
 
 /// A _character class_. Refer to the [module-level documentation](self) for
 /// details.

pomsky-syntax/src/exprs/char_class/unicode.rs

@@ -76,6 +76,10 @@ pub fn blocks_supported_in_dotnet() -> &'static [&'static str] {
     DOTNET_SUPPORTED
 }
 
+pub fn props_supported_in_java() -> &'static [&'static str] {
+    JAVA_SUPPORTED
+}
+
 /// Returns the list of all accepted shorthands.
 pub fn list_shorthands() -> impl Iterator<Item = (&'static str, GroupName)> {
     PARSE_LUT.iter().copied()

pomsky-syntax/src/lib.rs

@@ -33,4 +33,4 @@ pub use util::find_suggestion;
 use pretty_print::PrettyPrinter;
 
 #[doc(hidden)]
-pub use exprs::char_class::{blocks_supported_in_dotnet, list_shorthands};
+pub use exprs::char_class::{blocks_supported_in_dotnet, list_shorthands, props_supported_in_java};