Feature: allow interpolation of any block that returns ToTokens

[Pistonight]

Currently there is support for interpolating iterators but it often requires constructing the iterator in one place and using it in another

For example, deriving something from the enum variants:

let mut idents = Vec::new();
let mut exprs = Vec::new();
for ident in &my_enum.variants {
  idents.push(ident.clone());
  exprs.push(quote! { /* something */ })
}

// ... later
quote! {
   // ... some other stuff
  match x {
    #(  #idents => #exprs,  )*
  }
  // ... some other stuff
}

For large macros, it creates a lot of temporaries to be used later. Also personally, I found it easier to build and work with TokenStreams directly rather than levels of iterators in quote

let mut match_arms = TokenStream::new();
for ident in &my_enum.variants {
  match_arms.extend(quote! { #ident =>/* something */, });
}

// ... later
quote! {
   // ... some other stuff
  match x {
    #match_arms
  }
  // ... some other stuff
}

it would be nice if we could build match_arms inline in the quote, something like

quote! {
   // ... some other stuff
  match x {
    # {
      my_enum.variants.iter().fold(TokenStream::new(), |mut t, v| {
        let ident = &v.ident;
        t.extend(quote! { #ident => /* something */ });
        t
      })
    }
  }
  // ... some other stuff
}

or have it support a block that returns an iterator

quote! {
   // ... some other stuff
  match x {
    # {
      my_enum.variants.iter().map(|v| {
        quote! { #{ &v.ident } => /* something */ })
      })
    }
  }
  // ... some other stuff
}

If you have alternative approach I am happy to discuss as well. I haven't looked into the implementation details, just thinking about this from the user's perspective. Not sure how easy/hard this would be to implement but I am happy to look into that as well if you think this is worth pursuing

[dtolnay]

Per https://github.com/dtolnay/quote/pull/276#pullrequestreview-2105173018, I would prefer not to support this. Thanks anyway!

[Pistonight]

Ah, I tried to search for discussions on this in issues and didn't find any. I wonder if it would be worth to include a note that this will not be supported in the doc comment?