UNDEF HANDLING¶

By default, formatting codes tend to treat "undef" like Perl does: coercing it to an empty string or zero. This was a bad initial decision and will probably change. A "on_undef" handler can be provided when importing "errf" to setup a callback for how undefs should be handled. These two possibilities seem useful:

  # Very lax; undefs always turn into the same string:
  use String::Errf errf => { on_undef => sub { '(undef)' } };
  # Strict; undefs are never valid:
  use String::Errf errf => { on_undef => sub {
    Carp::croak("undef passed to $_[1]{literal}") } };
  } };

FORMATTING CODES¶

"errf" formatting codes require a set of arguments between the "%" and the formatting code letter. These arguments are placed in curly braces and separated by semicolons. The first argument is the name of the data to look for in the format data. For example, this is a valid use of "errf":

  errf "The current time in %{tz}s is %{now;local}t.", {
    tz  => $ENV{TZ},
    now => time,
  };

The second argument, if present, may be a compact form for multiple named arguments. The rest of the arguments will be named values in the form "name=value". The examples below should help clarify how arguments are passed. When an argument appears in both a compact and named form, the named form trumps the compact form.

The specific codes and their arguments are:

s for string

The "s" format code is for any string, and takes no arguments. It just includes the named item from the input data.

  errf "%{name}s", { name => 'John Smith' }; # returns "John Smith"

Remember, "errf" does not have any of the left- or right-padding formatting that "sprintf" provides. It is not meant for building tables, only strings.

i for integer

The "i" format code is used for integers. It takes one optional argument, "prefix", which defaults to the empty string. "prefix" may be given as the compact argument, standing alone. "prefix" is used to prefix non-negative integers. It may only be a plus sign.

  errf "%{x}i",    { x => 10 }; # returns "10"
  errf "%{x;+}i",  { x => 10 }; # returns "+10"
  errf "%{x;prefix=+}i",  { x => 10 }; # returns "+10"

The rounding behavior for non-integer values is not currently specified.

f for float (or fractional)

The "f" format code is for numbers with sub-integer precision. It works just like "i", but adds a "precision" argument which specifies how many decimal places of precision to display. The compact argument may be just the prefix or the prefix followed by a period followed by the precision.

  errf "%{x}f",     { x => 10.1234 }; # returns "10";
  errf "%{x;+}f",   { x => 10.1234 }; # returns "+10";
  errf "%{x;.2}f",  { x => 10.1234 }; # returns  "10.12";
  errf "%{x;+.2}f", { x => 10.1234 }; # returns "+10.12";
  errf "%{x;precision=.2}f",          { x => 10.1234 }; # returns  "10.12";
  errf "%{x;prefix=+;precision=.2}f", { x => 10.1234 }; # returns "+10.12";

t for time

The "t" format code is used to format timestamps provided in epoch seconds. It can be given two arguments: "type" and "tz".

"type" can be either date, time, or datetime, and indicates what part of the timestamp should be displayed. The default is datetime. "tz" requests that the timestamp be displayed in either UTC or the local time zone. The default is local.

The compact form is just "type" alone.

  # Assuming our local time zone is America/New_York...
  errf "%{x}t",               { x => 1280530906 }; # "2010-07-30 19:01:46"
  errf "%{x;type=date}t",     { x => 1280530906 }; # "2010-07-30"
  errf "%{x;type=time}t",     { x => 1280530906 }; # "19:01:46"
  errf "%{x;type=datetime}t", { x => 1280530906 }; # "2010-07-30 19:01:46"
  errf "%{x;tz=UTC}t",               { x => 1280530906 }; # "2010-07-30 23:01:46 UTC"
  errf "%{x;tz=UTC;type=date}t",     { x => 1280530906 }; # "2010-07-30 UTC"
  errf "%{x;tz=UTC;type=time}t",     { x => 1280530906 }; # "23:01:46 UTC"
  errf "%{x;tz=UTC;type=datetime}t", { x => 1280530906 }; # "2010-07-30 23:01:46 UTC"

n and N for numbered

The "n" and "N" format codes are for picking words based on number. It takes two of its own arguments, "singular" and "plural", as well as "prefix" and "precision" which may be used for formatting the number itself.

If the value being formatted is 1, the singular word is used. Otherwise, the plural form is used.

  errf "%{x;singular=dog;plural=dogs}n", { x => 0 }; # 0 dogs
  errf "%{x;singular=dog;plural=dogs}n", { x => 1 }; # 1 dog
  errf "%{x;singular=dog;plural=dogs}n", { x => 2 }; # 2 dogs
  errf "%{x;singular=dog;plural=dogs}n", { x => 1.4 }; # 1.4 dogs
  errf "%{x;singular=dog;plural=dogs;precision=1}n", { x => 1.4 }; # 1.4 dogs
  errf "%{x;singular=dog;plural=dogs;precision=0}n", { x => 1.4 }; # 1 dog

If "N" is used instead of "n", the number will not be included, only the chosen word.

  errf "%{x;singular=is;plural=are}N", { x => 0 }; # are
  errf "%{x;singular=is;plural=are}N", { x => 1 }; # is
  errf "%{x;singular=is;plural=are}N", { x => 2 }; # are
  errf "%{x;singular=is;plural=are}N", { x => 1.4 }; # 1.4 are
  errf "%{x;singular=is;plural=are;precision=1}N", { x => 1.4 }; # 1.4 are
  errf "%{x;singular=is;plural=are;precision=0}N", { x => 1.4 }; # 1 is

The compact form may take any of the following forms:

  word          - equivalent to singular=word
  word+suffix   - equivalent to singular=word;plural=wordsuffix
  word1/word2   - equivalent to singular=word;plural=word2

If no singular form is given, an exception is thrown. If no plural form is given, one will be generated according to some basic rules of English noun orthography.