A parser for WebIDL in Rust.
use webidl::*;
let lexer = Lexer::new("/* Example taken from emscripten site */\n\
enum EnumClass_EnumWithinClass {\n\
\"EnumClass::e_val\"\n\
};");
assert_eq!(lexer.collect::<Vec<_>>(),
vec![Ok((41, Token::Enum, 45)),
Ok((46, Token::Identifier("EnumClass_EnumWithinClass".to_string()), 71)),
Ok((72, Token::LeftBrace, 73)),
Ok((74, Token::StringLiteral("EnumClass::e_val".to_string()), 92)),
Ok((93, Token::RightBrace, 94)),
Ok((94, Token::Semicolon, 95))]);
use webidl::*;
use webidl::ast::*;
let result = parse_string("[Attribute] interface Node { };");
assert_eq!(result,
Ok(vec![Definition::Interface(Interface::NonPartial(NonPartialInterface {
extended_attributes: vec![
Box::new(ExtendedAttribute::NoArguments(
Other::Identifier("Attribute".to_string())))],
inherits: None,
members: vec![],
name: "Node".to_string()
}))]));
An example of a visitor implementation can be found here. Below is an example of how it is used:
use webidl::ast::*;
use webidl::visitor::*;
let ast = vec![Definition::Interface(Interface::NonPartial(NonPartialInterface {
extended_attributes: vec![
Box::new(ExtendedAttribute::NoArguments(
Other::Identifier("Attribute".to_string())))],
inherits: None,
members: vec![InterfaceMember::Attribute(Attribute::Regular(RegularAttribute {
extended_attributes: vec![],
inherits: false,
name: "attr".to_string(),
read_only: true,
type_: Box::new(Type {
extended_attributes: vec![],
kind: TypeKind::SignedLong,
nullable: true
})
}))],
name: "Node".to_string()
}))];
let mut visitor = PrettyPrintVisitor::new();
visitor.visit(&ast);
assert_eq!(visitor.get_output(),
"[Attribute]\ninterface Node {\n readonly attribute long? attr;\n};\n\n");
The parser is conformant with regards to the WebIDL grammar except for three points:
- Extended attributes, as described by the grammar, are not supported due to their lack of semantic meaning when parsed. Instead, limited forms are supported (as shown in the table). This parser allows a bit more flexibility when parsing extended attributes of the form
A=B
. The specification states thatA
andB
must be identifiers, but this parser acceptsB
as any token. If you would like for any extended attributes to be parsed (essentially any sequences of tokens), please consider looking at #8 to help resolve the problem with doing so. - This parser supports the old
implements
keyword that is no longer a part of the official specification. This is for backwards compatibility. - This parser supports the old
legacycaller
keyword that is no longer a part of the official specification. This is for backwards compatibility.