This is a weird little crate for authors of proc macros (or build.rs scripts) with an uncommon need: it lets you take values you're working with in code-generation and emit code that produces those values. In other words, it transmogrifies a value in codegen-space into code-space.
Why would I want such a thing? Let's say your macro allows users to express a
value for some type in an alternate format--json, say. The macro might do
something as simple as using serde to deserialize the json into the target
type. You might additionally do some error checking or validation to produce
helpful messages for the user.
But now you have the value in codegen-space. To get it into code-space you
need to turn it into the tokens that produce the value. This is precisely the
Transmogrify trait does.
use transmogrify::Transmogrify;
#[derive(Transmogrify)]
#[transmogrify(prefix = my_crate_name)]
pub struct MyStruct {
// ...
}
let value = MyStruct {};
let tokens: proc_macro2::TokenStream = value.transmogrify();The Transmogrify derive macro requires the presense of an attribute of
the form #[transmogrify(prefix = <path>)] where the path is the path-prefix
to the type on which you're deriving Transmogrify. This should be the "good",
consumer-visible path to the type that users of your crate would use. (Note
that we don't use std::any::type_name because that may produce a path that
contains mods that are inaccessible to crate consumers.)