Ruby CSV::Table sort in place - ruby

Question

I'm sorting a CSV::Table object. I have a table with headers ("date", "amount", "source") and O(50) entries.
Input:
data = CSV.table('filename.csv', headers:true) # note headers are :date, :source, :amount
amounts = []
data[:amount].each {|i| amounts << i.to_f}
data.sort_by! {|row| row[:amount]}
# error - not a defined function
data = data.sort_by {|row| row[:amount]}
# sorted but data is now an array not CSV::Table. would like to retain access to headers
I want a bang function to sort the table in place by the "amount" column without loosing the CSV::Table structure. Specifically, I want the result to be a CSV::Table, so that I still have access to the headers. Right now, I'm getting an Array, which is not what I want.
I'm sure there is an easier way to do this, especially with the CSV::Table class. Any help?

Answer 1

You can use:
CSV::Table.new(data) to convert Array to CSV::Table object if that is what you want.
sort_by is a method from Enumerable module which will always return an array when block is given as an argument

Answer 2

Suppose you define the following string:
txt=<<-CSV_TXT
Item, Type, Amount, Date
gasoline, expense, 200.00, 2022-01-01
Food, expense, 25.66, 2021-12-24
Plates, expense, 333.03, 2021-04-24
Presents, expense, 1500.01, 2021-12-15
Pay check, income, 2000, 2021-12-07
Consulting, income, 300, 2021-12-16
CSV_TXT
# for giggles, using a multi character separator of ', '
Now create a CSV Table from that (this in the IRB...):
> require 'csv'
=> true
> options={:col_sep=>", ", :headers=>true, :return_headers=>true}
=> {:col_sep=>", ", :headers=>true, :return_headers=>true}
> data=CSV.parse(txt, **options)
=> #<CSV::Table mode:col_or_row row_count:7>
We now have a CSV::Table with a header defined. If you use CSV::Table the header is not optional.
There are two ways (that I know of) you can now sort this table by the Date field and end up with a CSV::Table with the header unchanged. Neither is fully an 'in-place' solution.
The first, create a new CSV::Table after a round trip an array of CSV::Rows. The call to .sort_by creates that array of CSV::Rows for you and you can index a CSV::Row for sorting purposes. You use the first row of the existing table as the header argument:
> data=CSV::Table.new([data[0]]+data[1..].sort_by{ |r| r[3] })
=> #<CSV::Table mode:col_or_row row_count:7>
The second, is similar but allows the header to more easily be split off by using .to_a to create an array. This also allows the individual rows to be filtered or otherwise processed further:
> data=CSV.parse(txt, **options).to_a
=>
[["Item", "Type", "Amount", "Date"],
...
> header=data.shift.to_csv(**options)
=> "Item, Type, Amount, Date\n"
Now you have data with the header split off. With that array, you can sort or filter or process at will; then put back into an array of CSV strings. This is all in place:
> data.sort_by!{|r| r[3]}.map!{|r| r.to_csv(**options)}
=>
["Plates, expense, 333.03, 2021-04-24\n",
"\"Pay check\", income, 2000, 2021-12-07\n",
"Presents, expense, 1500.01, 2021-12-15\n",
"Consulting, income, 300, 2021-12-16\n",
"Food, expense, 25.66, 2021-12-24\n",
"gasoline, expense, 200.00, 2022-01-01\n"]
(Note the field with "Pay check" is quoted. If any character from a multi-character :col_sep is in a field, Ruby will quote it...)
To print the first, just use puts data since Ruby knows how to print a CSV::Table; to print the second, you can do puts header,data.join("")
For the second, to rejoin the header and data into a new table, use parse with options again:
> data_new=CSV.parse(header+data.join(""), **options)
=> #<CSV::Table mode:col_or_row row_count:7>