A simple crate for reducing the boilerplate when writing parsers with [syn].
Structs
[syn(parenthesized)],
[syn(braced)],
[syn(bracketed)]:
Corresponds to the isonymous macros in syn.
Must be attached to Paren, Brace, and Bracket fields, respectively.
#[syn(in = Ident)]:
The field is read from inside the named delimiter pair.
#[parse(fn(ParseStream) -> syn::Result<T>)]:
A function used to parse the field,
often used with Punctuated::parse_terminated
or Attribute::parse_outer.
#[to_tokens(fn(&mut TokenStream, &T)]:
A function used to tokenize the field.
Often used with TokenStreamExt::append_all,
though for type resolution reasons this needs to be indirected through a closure expression.
Enums
#[parse(prefix = fn(ParseStream) -> syn::Result<_>)]:
A prefix used for all branches, before doing the peeking.
Useful when all branches support attributes, for example.
The return value is ignored, which gives somewhat suboptimal performance, since the prefix is parsed twice.
#[parse(peek = Token)]:
Checks whether the variant should be parsed.
Even if multiple peeks succeed, only the first successful variant is attempted.
#[parse(peek_func = fn(ParseStream) -> bool)]:
More powerful than peek (such as allowing peek2), but gives worse error messages on failure.
peek should be preferred when possible.
Alternatives
derive-syn-parsedoes not handleToTokens. It also seems to encourage throwing tokens away with itsprefixandpostfixattributes.parseluses its own types for parentheses, meaning the AST types have different API fromsyn's own.