std::base64

std/src/base64.mo

Base64 (RFC 4648 §4) and base64url (§5) codecs, pure Monad.

B64 is the standard alphabet (+//, =-padded); B64Url is the URL alphabet (-/_) and emits no padding. Both decoders accept their own alphabet with or without padding and reject non-canonical trailing bits, so a re-encode of what they return is byte-identical.

Lives in std: the codecs are pure and portable, and they are the single home for base64. motes/http's request-cookie/basic-auth decoder and motes/tui's OSC 52 encoder delegate here, as does std::random's Random.token.

U8 has no shift or bitwise operations, so encoding works in U32 (U8.to_u32/U32.to_u8) and decoding uses multiplication/division, the shape std's SHA-256 and http's decoder already use.

typeB64Error#
pub type B64Error {
    /// A byte outside the alphabet (including `=` where it cannot appear).
    invalid_char,
    /// `=` in a position the group rule forbids.
    invalid_padding,
    /// A non-`=` character after the padding.
    data_after_padding,
    /// The final group's unused low bits are not zero, so a re-encode would
    /// not reproduce the input.
    non_canonical_bits,
    /// A final group of one character, which carries no whole byte.
    bad_length
}

Why a base64 string was rejected. One variant per rejection rule, so a caller can branch on the reason; to_string is the only render point.

defB64Error.to_string#
pub def B64Error.to_string (e : B64Error) : String
defB64.encode_std#
pub def B64.encode_std (bs : List U8) : String

Standard alphabet, =-padded (RFC 4648 §4).

defB64.decode_std#
pub def B64.decode_std (s : String) : Result B64Error (List U8)

Standard alphabet; accepts padded or unpadded input.

defB64Url.encode#
pub def B64Url.encode (bs : List U8) : String

URL alphabet, unpadded (RFC 4648 §5).

defB64Url.decode#
pub def B64Url.decode (s : String) : Result B64Error (List U8)

URL alphabet; accepts padded or unpadded input.