First column name is always skipped when calling GetColumnNames() function

[exonimbus]

Excellent library by the way, but none of the column named based addressing works as expected which is a bit of a let down. Mostly when calling GetCell() the overload that uses columnname as a string for addressing, ifcalled with the first column's name, it will always fail to find it and throw an exception. Because of this type of addressing problems GetCell() will not work either when called with the name of the first column in the csv file. Very annoying defect, please fix it if it is at all possible. I am a quick getting around this problem by always using numeric column index addressing when calling GetCell, but GetColumnNames would be very valuable if working well, particularly when one doesn't know exactly how many columns to expect in a csv and it would be most excellent if column name discovery would work as expected (i.e. don't skip first column and report the actual number of columns instead of number of columns - 1). Many thanks for creating this library, it is indeed very useful otherwise.

[d99kris]

Hi @exonimbus - rapidcsv allows four different parsing "modes" with respect to row/column headers. By default (default LabelParams) it treats first row as column headers and allows access to first column as data - I think this is what you're looking for - see example. Perhaps you can share how you use rapidcsv (especially how Document() constructor is called, and any LabelParams in particular)?

The other three label/header modes are listed in Reading a File with Column and Row Headers.

[exonimbus]

Hi Kristofer,

Thank you for answering this. You are right, my problem was with the way I called the constructor, it was not very clear in the examples what each parameter does. Now I have it figured out, after some trial and error, and it looks like that default constructor with no parameters is the way to go in my simple case (i.e. no fancy addressing modes based on first row value etc.). I admit that I copied some of your examples without understanding how initialization is supposed to work and my mistake was that I passed in rapidcsv::LabelParams(0, 0) which created that unexpected behavior I mentioned above. It looks like the default (no params) constructor does the right job in my case and all columns are then considered when addressing, as one would by default expect, thanks for pointing this out for me. This has helped me immensily, I mean being able to use this library, which indeed is the only free one out there that does proper escaping the way Excel would do when saving csv's. Everything else out ther that I tried so far didn't pass some of my fringe usage cases, for this I am very grateful.

[d99kris]

Hi @exonimbus - I'm glad to hear it worked out. I can admit that the LabelParams is not super intuitive. I've been meaning
to try document it more clearly, maybe with some image/illustration, hopefully I get around to it eventually.

Thanks for the positive feedback 👍