(csv) Allow to use multiple-characters delimiter to separate column: withColumnSeparator

[cowtowncoder]

(from FasterXML/jackson-dataformat-csv#141 by @jefferyyuan)

It's common that we use multiple-characters delimiter to separate columns in CVS: such as: [|]
Sometimes, we have no choice as third-party defines the csv format like that.

Sample data:

FirstName[|]LastName[|]Address
John[|]Smith[|]123 Main St.

FasterXML/jackson-dataformat-csv#92 supports to use multiple characters as array element separator, It would be great if same can be done for withColumnSeparator.

I did some google search: seems there is no java library that supports use multiple-characters as the column separator.

But the C# CsvHelper does
https://github.com/JoshClose/CsvHelper

[nrapopor]

it's been two years -- any news?

[cowtowncoder]

@nrapopor It's an OSS project so if you want to help, there are good ways to help. Asking "what's the ETA" is on bottom 5 list of helpful things.

FAQ: if there is progress, there will be update to issue.

[skill7899]

ORZ

[djay-S]

Hi @cowtowncoder ,
Instead of using
public CsvSchema withColumnSeparator(char sep){}
can we change it to
public CsvSchema withColumnSeparator(char[] sep).
and also changes the data type for _columnSeparator and other needed places.

Let me know of your suggestions.

[cowtowncoder]

@djay-S Problem is not the method defining separator but implementation behind it -- splitting columns is very performance-sensitive operation, on critical path. So the real work is in decoder, to try to implement it in a way that has negligible performance penalty.
This is the real problem.