Improve and document CSV reading options

[koperagen]

https://youtrack.jetbrains.com/issue/KT-69798
Unlike readCSV, readDelim exposes underlying apache common CSVFormat. There're a lot more configuration options. From the linked issue it seems arguments that readCSV provide are not enough to read every possible kind of file.
So at least readDelim needs documentation here
https://kotlin.github.io/dataframe/read.html
Or readCSV should have more options. But, i'd avoid exposing CSVFormat in readCSV

[zaleslaw]

We agreed to split the issue

• answer the use-case
• provide a default

[Jolanrensen]

For csv I suggest the following:

• Split readDelim and readCsv into different files, possibly a different module?
• Deprecate .read() overload, this should only exist in guess.kt
• KDocs for each overload
• Check which CSVFormat options we can put in the readDelim and readCsv functions to create overloads without it. (Like setting the quote mark)
• Keep one overload for readDelim and readCsv with CSVFormat as option so people can set advanced settings with it.

[unrec]

Thanks everyone for working on this issue.

I agree that exposing CSVFormat to readCSV fun is not good, but there should be a flexibility in CSV setup configuration.
I faced it with a .csv file with quote character in some row, so as I understand there are 2 possible options:

• be able to set specific quote character instead of "
• have an opportunity to disable quote char completely

This should be added as a optional param to readCsv fun instead of readDelim.

[Jolanrensen]

Will be solved by the new CSV implementation: "dataframe-csv". Setting a quote character is supported by default.

The new read functions will also expose a single overload where you can adjust additional CsvSpecs if we ever miss out on a parameter in the future.