CSV (Line-by-Line) Records with Auto-Converted Numerics (Float Numbers) Encoding Rules - A Modern (Simple) Tabular Data Format incl. Numbers, Comments and More
Let's read:
1,2,3
"4","5","6"
What do you expect? In the "classic" vanilla format the comma-separated values once read in / parsed always are a list of string values. Period.
[["1", "2", "3"],
["4", "5", "6"]]Guess, what? There's a popular comma-separated values (CSV) convention / variant / dialect that now has an official specification:
Rule 1: Use "un-quoted" values for float numbers e.g. 1,2,3 or 1.0, 2.0, 3.0 etc.
Rule 2: Use quoted values for "non-numeric" strings e.g. "4", "5", "6" or "Hello, World!" etc.
Example:
1,2,3
"4","5","6"
returns / maps to
[[1.0, 2.0, 3.0],
["4", "5", "6"]]and now has an official name. Let's call it:
CSV <3 Numerics
A CSV <3 Numerics reader / parser lets you configure a list of values
that get auto-converted to NaN, that is, Not A Number.
Example:
1,2,NAN,#NAN
with the nan option ['NAN', '#NAN']
returns / maps to:
[[1.0, 2.0, NaN, NaN]]Note: The Not a Number (NaN) values are "un-quoted" values (like numbers) in the comma-separated values (CSV) format.
Yes, in CSV <3 Numberics you can use comments (starting with #) and blank lines
and leading and trailing (white)spaces - they all will get skipped and trimmed automatically.
Example:
# CSV <3 Numerics
1 , 2 , 3
"4" , "5" , "6"
returns / maps to
[[1.0, 2.0, 3.0],
["4", "5", "6"]]Note: (White)spaces inside quotes will NOT get trimmed:
" 4 " , " 5 " ," 6 "
returns / maps to
[[" 4 ", " 5 ", " 6 "]]See alternative CSV formats that include specified encoding rules for more types:
- CSV <3 JSON Format, see https://github.com/csvspecs/csv-json
- includes Nulls, Booleans, Arrays, Structs, and more
- does NOT include Dates
- CSV <3 YAML Format, see https://github.com/csvspecs/csv-yaml
- includes Nulls, Booleans, Dates, Arrays, Structs, and more
The recommended file format for CSV <3 Numerics files is .csv :-) or use .num.csv (to highlight
the fact of the numerics encoding rules).
The CSV <3 Numerics format is dedicated to the public domain.
Please post your comments to the wwwmake forum. Thanks!