csvjson -key option processing

[wiluite]

To work successfully with the key option, all values ​​in the column must be unique. However, csvjson considers the same numeric values ​​represented by different strings to be different. Is it really correct? )

a,b,c
1/1/2020,1s,3.
01/01/2020,1 sec,3.0

csvjson  dummy.csv -k a
ValueError: Value 2020-01-01 is not unique in the key column.

csvjson  dummy.csv -k b
ValueError: Value 0:00:01 is not unique in the key column.

csvjson  dummy.csv -k c

{"3": {"a": "2020-01-01", "b": "0:00:01", "c": 3.0}, "3.0": {"a": "2020-01-01", "b": "0:00:01", "c": 3.0}}

[jpmckinney]

Hmm, yes, and different number of decimals has the same behavior:

$ printf 'a,b,c\n1/1/2020,1s,3.00\n01/01/2020,1 sec,3.0' | csvjson -k c
{"3.00": {"a": "2020-01-01", "b": "0:00:01", "c": 3.0}, "3.0": {"a": "2020-01-01", "b": "0:00:01", "c": 3.0}}

[jpmckinney]

Aha, it is because Decimal('3.00') == Decimal('3.0') but str(Decimal('3.00')) != str(Decimal('3.0')). Fixed in agate.