How can I right pad an integer with zeros?

[derekmahar]

How can I left-align an integer and pad it to the right with zeros? For example, left-aligning 123 and padding it with zeros to its right would produce "123000".

Miller function fmtnum can right-align an integer and pad it with zeros to its left:

$ echo number=123 | mlr put '$result = fmtnum($number, "%06d");'
number=123,result=000123

Can fmtnum left-align and right-pad a number with zeros? Can format or some other function do this?

[derekmahar]

Turns out that I required a function that left-aligns and right-pads a string, not an integer. I needed this specifically in order to left-align and right-pad fractional seconds of an ISO 8601 timestamp string. I couldn't use an integer to represent the decimal digits of the fractional seconds because fractions such as "085" or "001" are technically not integers and even if Miller could convert them to integers, it would strip the significant leading zeros. Instead, I had to treat these fractional seconds as strings that preserved all digits, including leading zeros.

To achieve this goal, I wrote the following function right_pad that appends repeated copies of string padding to input string unpadded_string to a maximum length of width.

func right_pad(str unpadded_string, int width, str padding) {
  str padded_string = unpadded_string;

  for (; strlen(padded_string) <= width; padded_string .= padding) {};

  return truncate(padded_string, width);
}

I think Miller could benefit from right_pad or similar as a built-in string function.

[johnkerl]

@derekmahar great idea!! :) This is tracked in #1166.

[aborruso]

hi @derekmahar , probably It's not elegant and it's not what you're looking for.

Starting from this input file

a
1
12
253
426
2

and using powers of 10, and cell length, you could run something like

mlr --csv  put '$b=$a*pow(10,6-strlen(string($a)))' input.csv

to have

a	b
1	100000
12	120000
253	253000
426	426000
2	200000

[derekmahar]

@aborruso, your solution certainly solves the problem that I stated in the title and I did think of this one, but I chose a more general solution that right-pads any input string with repetitions of another string. Your powers-of-ten solution is probably much faster than my solution, though, so I should use this one specifically for integers.

[derekmahar]

Oh, I overlooked one important point. Your solution can't right-pad an "integer" string that has leading zeroes such as "001" or "023" because Miller won't convert it to an integer unless you first strip the leading zeroes. However, if you strip those leading zeroes, you would have to save them and then prepend them to the integer that results from multiplying the "stripped" integer by the necessary powers-of-ten.

[aborruso]

Your solution can't right-pad an "integer" string that has leading zeroes such as "001" or "023"

True, but 001 is not an integer. But you are right, I understood you: your question is no longer that of the original subject, but it is a generic one, related to rigth-pad, and not strictly to "integer with zeros".

[amp42]

I needed to left-pad a string, here's a slight mod to @derekmahar's function:

func lpad(unpadded_string, int width, string padding) {
  if (strlen(string(unpadded_string)) >= width) {return string(unpadded_string);}

  str padded_string = string(unpadded_string);
  for (; strlen(padded_string) < width; padded_string = padding . padded_string) {};
  return truncate(padded_string, width);
}

Some of my unpadded strings could pass as numbers (42) and some are number-looking strings (042), thus type coercion via string() and untyped unpadded_string argument. I stuck the function in $HOME/functions.mlr and used it for left-padding to a width of 3 with zeros:

mlr --load ~/functions.mlr --csv --from myfile.csv put '$foo=lpad($foo,3,"0")'

[derekmahar]

Why leave unpadded_string as untyped when you could just require that the client coerce the argument assigned to this parameter as a string? As a string parameter, unpadded_string could handle both number and number-looking strings that have leading zeros or the client might decide to strip leading zeros from the argument before passing it to lpad.

Note that unlike right_pad, lpad neglects to truncate its return value when unpadded_string is longer than the maximum width.

[amp42]

Why leave unpadded_string as untyped when you could just require that the client coerce the argument assigned to this parameter as a string?

...because mlr doesn't like that with numbers. With unpadded_string as str and inlined zero for padding:

func lpadzero(str unpadded_string, int width) {
  if (strlen(unpadded_string) >= width) {return unpadded_string;}

  str padded_string = unpadded_string;
  for (; strlen(padded_string) < width; padded_string = 0 . padded_string) {};
  return truncate(padded_string, width);
}

test.csv:

foo
011
002
3
04
55
666

mlr --load ~/functions.mlr --csv --from test.csv put '$foo=lpadzero($foo,3)' yields

mlr: couldn't assign variable str unpadded_string from value int 3

[derekmahar]

Right, so that's why you should instead call it using mlr --load ~/functions.mlr --csv --from test.csv put '$foo=lpadzero(string($foo),3)'.

[johnkerl]

@derekmahar @aborruso #1205 is in progress