Crate mailrs_rfc2047 

mailrs-rfc2047

RFC 2047 MIME encoded-word decoder. Decodes =?charset?(B|Q)?text?= header values (Subject, From display name, …) into UTF-8.

Supports the full WHATWG Encoding charset set via encoding_rs: UTF-8, ISO-8859-, Windows-125, ISO-2022-JP, Shift_JIS, EUC-JP, EUC-KR, Big5, GB18030, etc. Unknown charsets fall through to a lossy UTF-8 pass.

Quickstart

use mailrs_rfc2047::decode;

// ASCII inputs are returned borrowed — no allocation.
assert_eq!(decode(b"plain text"), "plain text");

// Base64 encoded UTF-8.
assert_eq!(decode(b"=?UTF-8?B?VGVzdA==?="), "Test");

// Q (quoted-printable) encoded.
assert_eq!(decode(b"=?UTF-8?Q?Hello_World?="), "Hello World");

// ISO-2022-JP (Japanese subject from real-world mail).
assert_eq!(
    decode(b"=?ISO-2022-JP?B?GyRCJDMkcyRLJEEkTxsoQg==?="),
    "こんにちは",
);

// Adjacent encoded-words collapse whitespace per RFC 2047 §6.2.
assert_eq!(
    decode(b"=?UTF-8?B?aGVsbG8=?= =?UTF-8?B?d29ybGQ=?="),
    "helloworld",
);

Pairing with mailrs-rfc5322

This crate is the typical companion of mailrs-rfc5322. mailrs-rfc5322::Message::header() returns raw header bytes; pass those bytes to mailrs_rfc2047::decode() to get the decoded text:

// (this example uses the `mailrs-rfc5322` crate as well; both ship
// independently. add both to your Cargo.toml to compile.)
use mailrs_rfc5322::Message;
use mailrs_rfc2047::decode;

fn extract_subject(msg_bytes: &[u8]) -> Option<String> {
    let m = Message::new(msg_bytes);
    let subject = m.header("Subject").map(|b| decode(b).into_owned())?;
    Some(subject)
}

What this crate is not

• Not an RFC 5322 parser. Use mailrs-rfc5322 for that.
• Not a MIME body decoder (multipart, Content-Transfer-Encoding). This only decodes encoded-words in headers.
• Not a charset detector. The charset is taken verbatim from the encoded-word token; if a message claims =?UTF-8?Q?…?= and the bytes are actually Shift_JIS, you get garbage.

Performance

Measured numbers in BUDGETS.md. Reproduce via cargo bench -p mailrs-rfc2047 --bench decode.

Headline: plain-ASCII inputs return as borrowed Cow::Borrowed(&str) with zero allocations and constant time (just a forward scan for =?). Encoded inputs go through one String allocation sized to the input length.

License

Apache-2.0 OR MIT. Internal layout: decode is the entry point. It scans for =?charset?(B|Q)?text?= tokens and replaces them with their UTF-8 decoding; ASCII runs are copied unchanged. Charset → UTF-8 conversion goes through encoding_rs::Encoding::for_label.

Functions

decode

Decode an RFC 2047 encoded header value into UTF-8.

encode

Encode a UTF-8 string as an RFC 2047 encoded-word if and only if it contains non-ASCII bytes. Pure-ASCII inputs are returned as borrowed Cow::Borrowed unchanged — no allocation, no wrapping.