base64 text pipe

[ChristopherDavenport]

I was interested recently in a large file contents and translating it from its base64 to the string representation.

I think it would be fairly straightforward to implement but wanted to see if there was an interest.

[mpilquist]

I'd love to be able to reuse the methods on ByteVector for this -- perhaps with some small changes to scodec-bits to handle carry?

[mpilquist]

@ChristopherDavenport How are these signatures for the fs2.text object?

object text {

  def base64Decode[F[_]]: Pipe[F, String, Byte] =
    _.through(base64DecodeC).flatMap(Stream.chunk)

  def base64DecodeC[F[_]]: Pipe[F, String, Chunk[Byte]] = ??? 

  def base64Encode[F[_]]: Pipe[F, Byte, String] =
    _.chunks.through(base64EncodeC)

  def base64EncodeC[F[_]]: Pipe[F, Chunk[Byte], String] = ???
}

I think we can skip the chunk variants assuming they stay identities.

[djspiewak]

Is there a reason to have the separate chunk variants at all rather than just ensuring that the primary variants operate on chunks rather than individual bytes?

[mpilquist]

@djspiewak I think we can skip the chunk variants. I looked up why we have the utf8 chunk variants and found this: #609

[djspiewak]

Ah interesting. It's worth noting that it would probably be more efficient, even in the utf8 codecs, to exploit the chunkiness more explicitly. But, given that it's Strings, I'm guessing it wouldn't make a massive difference. It would definitely be much more noticeable with Chunk[Byte].

Either way, we definitely don't need the Cs