A placeholder error message is printed when parsing files with ambiguous options

[adamroyjones]

I've noticed that the csv gem produces a placeholder error message when attempting to parse a TSV with ambiguous parsing options.

If one were to parse the string

foo\t\tbar

when directed to strip whitespace (i.e., to treat whitespace as insignificant) and use tabs as field separators (i.e., to treat whitespace as significant), then should one parse it as ["foo\t\tbar"], ["foo", "bar"], or ["foo", "", "bar"]? There's no way to choose a parsing strategy that wouldn't cause a reasonable surprise to someone.

I think an error should be produced that informs the user that this is a problem; at the moment, this doesn't happen. With Ruby 3.0.2 and csv 3.2.1, the file

require "csv"

sep = "\t"
tsv_file = "example.tsv"

File.open(tsv_file, "w") { |f| f.puts("foo#{sep}bar") }
CSV.read(tsv_file, col_sep: sep, strip: true)

produces the error

csv-3.2.1/lib/csv/parser.rb:935:in `parse_quotable_robust': TODO: Meaningful message in line 1. (CSV::MalformedCSVError)

I think this is misleading; the file is not really "malformed".

Do you agree with the above? If so, I'm happy to offer up a merge request around this (e.g. to check the options before beginning the parsing, to make sure that if strip is set to true then col_sep must not be whitespace).

[kou]

Good catch! Could you open a pull request?

[adamroyjones]

Thanks for getting back to me so quickly—I'm glad I'm not going mad and that you agree!

I'll work on it and raise a merge request soon.