Encode a collection to a CSV as a text Builder. For example, we can take the following columnar encoding of a person:

>>> :{
let colPerson :: Colonnade Headed Person Text
    colPerson = mconcat
      [ C.headed "Name" name
      , C.headed "Age" (T.pack . show . age)
      , C.headed "Company" (fromMaybe "N/A" . company)
      ]
:}

And we have the following people whom we wish to encode in this way:

>>> :{
let people :: [Person]
    people =
      [ Person "Chao" 26 (Just "Tectonic, Inc.")
      , Person "Elsie" 41 (Just "Globex Corporation")
      , Person "Arabella" 19 Nothing
      ]
:}

We pair the encoding with the rows to get a CSV:

>>> LTIO.putStr (TB.toLazyText (encodeCsv colPerson people))
Name,Age,Company
Chao,26,"Tectonic, Inc."
Elsie,41,Globex Corporation
Arabella,19,N/A

Streaming variant of encodeCsv. This is particularly useful when you need to produce millions of rows without having them all loaded into memory at the same time.

Encode a collection to a CSV as a bytestring Builder.

Backwards-compatibility alias for decodeHeadedCsvUtf8.

Decode a CSV whose first row is contains headers identify each column.

Decode a CSV without a header.

Uses the second argument to parse a CSV column whose header content matches the first column exactly.

Uses the argument to parse a CSV column.

Uses the second argument to parse a CSV column that is positioned at the index given by the first argument.

This just actually a specialization of the free applicative. Check out Control.Applicative.Free in the free library to learn more about this. The meanings of the fields are documented slightly more in the source code. Unfortunately, haddock does not play nicely with GADTs.

Maps over a Decolonnade that expects headers, converting these expected headers into the indices of the columns that they correspond to.

This adds one to the index because text editors consider line number to be one-based, not zero-based.