For example, on https://mikedilger.github.io/mime-multipart/mime_multipart/enum.Node.html, you link to "http://hyper.rs/hyper/hyper/header/struct.Headers.html", which no longer exists. "http://hyper.rs/hyper/v0.9.12/hyper/header/struct.Headers.html" would work.

It appears that your Display implementation calls itself recursively forever.

reproduce with:

     #[test]
     fn display_error() {
         let custom_error = std::io::Error::new(std::io::ErrorKind::Other, "oh no!");
         println!("{}", Error::Io(custom_error));
     }
    

Will output bytes that are invalid within RFC5322 emails if parts contain such characters.

Currently we don't generate anything, we only parse (parsing was the harder job). As soon as someone needs us to generate, we can do that.

At that time FilePart will need to make tempdir public, and size Optional, as users will pass these to us.

This crate is built upon an older, pre-async version of hyper, and as a result the API is also blocking. The hyper ecosystem has moved to async.

I've analyzed a little bit what I think should be done:

• Depend on https://docs.rs/http instead of hyper
• Depend on https://docs.rs/hyperx for various header parsers
• Change read_multipart/read_multipart_body to return futures::stream::Stream of nodes instead of Vec

Will you accept a PR?

I seem to have published an 0.5.1 version quite a while ago, but the github source does not have this code. I will be looking around to find it and will push it when I do.

Right now, for downstream crates which have data from hyper, we take Headers and the stream as separate arguments to parse_multipart().

However, in the future, data may come from sources other than Hyper (e.g. an email system, a file, etc). In those cases we should take just the stream, and parse off the headers ourselves. This will be easy to implement since we already parse internal headers ourselves (using httparse).

Result of generate_boundary contains '=' and '/', which is not correctly handle by some multipart implementations.
I have a very simple fix, hopefully you can merge it soon.
zyddnys@fc18017