Skip to content

Latest commit

 

History

History
200 lines (144 loc) · 5.45 KB

README.md

File metadata and controls

200 lines (144 loc) · 5.45 KB

C++ API

The C++ API provides fixed-width formatters for these types:

  • fixfmt::Bool for booleans
  • fixfmt::String for strings
  • fixfmt::Number for integer and floating-point numbers
  • fixfmt::TickTime for timestamps represented as UTC epoch timestamps

An instance of one of these classes represents a formatter with fixed configuration, and will always produce a string of the same width when formatting any value. Use the overloaded operator() to format a value.

All formatters have a get_width() accessor, which returns the formatted width of any value formatted by that formatter.

Booleans

For example, the Bool formatter with default arguments formats true as True and false as False, padding the true value to five characters to match the length of the false value.

fixfmt::Bool fmt;
std::cout << "[" << fmt(true) << "]\n"
          << "[" << fmt(false) << "]\n";

produces,

[True ]
[False]

Likewise,

fixfmt::Bool fmt("ABSOLUTELY", "no");
std::cout << "[" << fmt(true) << "]\n"
          << "[" << fmt(false) << "]\n";

produces,

[ABSOLUTELY]
[no        ]

The constructor's signature is,

Bool(string const& true_str, string const& false_str, size_t size, bool pad_left)

The size parameter allows you to specify a fixed width; otherwise the longer of the length of true_str and false_str is used. If pad_left is true, the shorter string is padded on the left instead of on the right.

Strings

The string formatter formats a string to a fixed width either by padding to lengthen or eliding to shorten. Padding consists of adding characters either to the left or right of the string. Eliding consists of replacing a sequence of characters in the beginning, middle, or end of the string with a shorter ellipsis.

fixfmt::String fmt(10);
std::cout << "[" << fmt("testing") << "]\n"
          << "[" << fmt("Hello, world!") << "]\n";

produces,

[testing   ]
[Hello, wo…]

Observe that the first string has been padded on the right with three spaces to achieve a length of 10, while in the second string, the last four characters have been replaced with a single character, a Unicode ellipsis ("three periods").

The constructor's signature is,

String(int size, string ellipsis, string pad, double position, bool pad_left)

ellipsis is the ellipsis used for shortening, while pad is the pad used for lengthening. position is the position in the string at which the elision occurs: 0.0 for the left, 1.0 for the right, or an intermediate number for elsewhere in the string.

Numbers

The number formatter formats a number with a fixed number of digits for each of the integral and fractional parts, called the "size" and "precision" respectively.

double const x = 123.45678;
std::cout << "[" << fixfmt::Number(5, 1)(x) << "]\n"
          << "[" << fixfmt::Number(3, 3)(x) << "]\n";

produces

[   123.5]
[ 123.457]

If the precision is 0, the decimal point is included but without any fractional digits. If the precision is Number::PRECISION_NONE, the decimal point is omitted and the number is formatted as an integer. Numbers are rounded half-to-even ("bankers' rounding").

std::cout << "[" << fixfmt::Number(5)(x) << "]\n";

produces

[   123]

Since Number has a quite a number of features you can specify, its constructor takes Number::Args struct:

struct fixfmt::Number::Args
{
  int      size           = 8;
  int      precision      = PRECISION_NONE;
  char     sign           = SIGN_NEGATIVE;
  char     pad            = ' ';
  char     point          = '.';
  string   nan            = "NaN";
  string   inf            = "inf";
  char     bad            = '#';
};

  • pad is the character used to pad the integral part to the size, and may be either Number::PAD_SPACE or Number::PAD_ZERO.

  • sign determines how the sign is presented:

    • Number::SIGN_NEGATIVE (the default) reserves one position for the sign, and displays a space for positive numbers, a minus sign for negative numbers.
    • Number::SIGN_ALWAYS is similar except that a plus sign is shown for positive numbers.
    • Number::SIGN_NONE does not reserve a position for the sign; negative numbers cannot be formatted.

  • nan and inf are the renderings for NaN and infinite values, respectively.

  • point is the character used for the decimal point.

  • bad is a single character that is used to fill the entire width if the number cannot be formatted, for instance if its integral part exceeds the capacity of the format size.

Use the C99 struct field initialization syntax to specify the ones you need when creating a Number. For example,

fixfmt::Number fmt({6, .pad=fixfmt::Number::PAD_ZERO});

See the unit tests for additional examples.

Times

Note: This time functionality is a placeholder, and will be redesigned in the future.

The tick time formatter formats UTC times represented as long ticks since the UNIX epoch (1970-01-01T00:00:00Z). A tick is an integral fraction of a second, generally whole seconds, milliseconds, microseconds, or nanoseconds.

The time is formatted in ISO-8601 extended time, with a fixed number of decimal digits of precision for the seconds.

fixfmt::TickTime fmt(fixfmt::TickTime::SCALE_NSEC, 4)
std::cout << "[" << fmt(1447686456947558144l) << "]\n";

produces

[2015-11-16T15:07:36.9476Z]