Question

I have been doing some testing with CSV.table. I have two small and almost identical CSV files, however one is missing the header row.

When I run CSV.table against the CSV file with the header row, everything works as expected.

When I run it against the CSV file without the header row I get:

NoMethodError: undefined method `encode' for nil:NilClass

I tried this with different types of data, with different types of headers, and get the same results.

I am curious to the magic of CSV.table. If I use CSV.parse with headers set to true, then it always makes the first row be headers no matter what. So, I have been using CSV.table to check if a CSV file being imported has a header row but I am not too comfortable with this because I don’t understand if or when it will or will not work the way I’m using it.

begin
  CSV.table(csv_file_path)
rescue
  # Add error to log or something.
end

Does anyone know?

P.S. I’ve already read through this and the source code it provides on each method – http://www.ruby-doc.org/stdlib-1.9.3/libdoc/csv/rdoc/CSV.html

Answer 1

There isn’t any magic involved, and it won’t work for you in general.

As you can see from the source, table literally just calls read with headers: true. But it also converts the header to symbols (header_converters: :symbol) and this is the key to why it appears to work.

You get an error without headers because you have a blank column in your first row of data (something like a,b,,d,e). The blank gets read in as nil, and since nil can’t be converted to a symbol, it blows up.

Try it with some data that doesn’t have a blank in the first row – you’ll see that table will treat that row of data as headers just like any of the other methods.