fix writeCsv for OptionalColumn

[mcoady]

Description

writeCsv was outputting optional (non-Text) values as text including Just/Nothing in the file. You can see this behaviour doing a read/write/read cycle (see below). On Windows btw.

Example

Input

dataframe> df <- D.readCsv "input.csv"
dataframe> df
--------------------------------------
id  | fld_a  |    fld_b    |   fld_c
----|--------|-------------|----------
Int | Double | Maybe Text  | Maybe Int
----|--------|-------------|----------
0   | 0.0    | Nothing     | Nothing
1   | 0.5    | Just "base" | Just 10

Behaviour before

dataframe> D.writeCsv "before.csv" df >> D.readCsv "before.csv" >>= print
-------------------------------------------
id  | fld_a  |    fld_b    |     fld_c
----|--------|-------------|---------------
Int | Double | Maybe Text  |   Maybe Text
----|--------|-------------|---------------
0   | 0.0    | Nothing     | Nothing
1   | 0.5    | Just "base" | Just "Just 10"

Behaviour after

dataframe> D.writeCsv "after.csv" df >> D.readCsv "after.csv" >>= print
--------------------------------------
id  | fld_a  |    fld_b    |   fld_c
----|--------|-------------|----------
Int | Double | Maybe Text  | Maybe Int
----|--------|-------------|----------
0   | 0.0    | Nothing     | Nothing
1   | 0.5    | Just "base" | Just 10

[mchav]

Great catch and fix. I'll include some round trip tests in a follow up.

[mchav]

@mcoady text show doesn't seem to work for older ghcs. Can we in the mean time add another case statement matching Maybe Text?

[mcoady]

Ah, my bad - I see show was added in text-2.1.2. Yes, I can see the need to handle Maybe Text separately as I can see quoted strings in the output. I'll take a look. Apologies for a basic question, how are you managing the running of tests for different compiler versions? Is there a standard or preferred tool?

As an aside - excellent project btw! I'm looking to replace some python/polars with this so happy to provide feedback if I stumble across anything else, but liking it so far!

Thanks!