rename _convert to maybe_encode

[johnnychen94]

This is a sub PR of #61

Changes:

• rename _convert to maybe_encode
• add tests for maybe_encode
• more patches to ignore CRLF/LF differences

---

I'm unclear what should maybe_encode(format"TXT", df::DataFrame) and maybe_encode(format"SHA256", df::DataFrame) behave. The current behavior is maybe_encode(fmt, string(df))

Converting it using string can be prune to show changes. Hence I suggest we do something like

function maybe_encode(::Type{DataFormat{:TXT}}, df::DataFrame)
    save("tmp.csv", df)
    _ignore_crlf(read(String, "tmp.csv"))
end

function maybe_encode(::Type{DataFormat{:SHA256}}, df::DataFrame)
    save("tmp.csv", df)
    bytes2hex(sha256(read("tmp.csv")))
end

@oxinabox if you like this idea, I can make a PR for this change.

[oxinabox]

Why?

[johnnychen94]

This function expects only string types splitext, I added this to make sure nothing strange is passed here as a safety net.

[oxinabox]

Should we just say we are ignoring \r ?

Suggested change

	_ignore_crlf(x::AbstractString) = replace(x, "\r"=>"")
	_ignore_linefeed(x::AbstractString) = replace(x, "\r"=>"")

[johnnychen94]

I assume you're suggesting Carriage Return (\r) instead of linefeed (\n)?

I renamed to _ignore_CR with additional docstring.

[johnnychen94]

I'm merging this to avoid possible conflict and save myself some efforts from rebasing.

If there're any further comments, I'll address them in future PRs.