4.61 sprintf, fprintf, printf, snprintf, asprintf, asnprintf—format output

Synopsis

     #include <stdio.h>
     
     int printf(const char *restrict format, ...);
     int fprintf(FILE *restrict fd, const char *restrict format, ...);
     int sprintf(char *restrict str, const char *restrict format, ...);
     int snprintf(char *restrict str, size_t size, const char *restrict format,
         ...);
     int asprintf(char **restrict strp, const char *restrict format, ...);
     char *asnprintf(char *restrict str, size_t *restrict size, const char *restrict format,
         ...);
     
     int _printf_r(struct _reent *ptr, const char *restrict format, ...);
     int _fprintf_r(struct _reent *ptr, FILE *restrict fd,
         const char *restrict format, ...);
     int _sprintf_r(struct _reent *ptr, char *restrict str,
         const char *restrict format, ...);
     int _snprintf_r(struct _reent *ptr, char *restrict str, size_t size,
         const char *restrict format, ...);
     int _asprintf_r(struct _reent *ptr, char **restrict strp,
         const char *restrict format, ...);
     char *_asnprintf_r(struct _reent *ptr, char *restrict str,
         size_t *restrict size, const char *restrict format, ...);
     

Description
printf accepts a series of arguments, applies to each a format specifier from *format, and writes the formatted data to stdout, without a terminating NUL character. The behavior of printf is undefined if there are not enough arguments for the format. printf returns when it reaches the end of the format string. If there are more arguments than the format requires, excess arguments are ignored.

fprintf is like printf, except that output is directed to the stream fd rather than stdout.

sprintf is like printf, except that output is directed to the buffer str, and a terminating NUL is output. Behavior is undefined if more output is generated than the buffer can hold.

snprintf is like sprintf, except that output is limited to at most size bytes, including the terminating NUL. As a special case, if size is 0, str can be NULL, and snprintf merely calculates how many bytes would be printed.

asprintf is like sprintf, except that the output is stored in a dynamically allocated buffer, pstr, which should be freed later with free.

asnprintf is like sprintf, except that the return type is either the original str if it was large enough, or a dynamically allocated string if the output exceeds *size; the length of the result is returned in *size. When dynamic allocation occurs, the contents of the original str may have been modified.

For sprintf, snprintf, and asnprintf, the behavior is undefined if the output *str overlaps with one of the arguments. Behavior is also undefined if the argument for %n within *format overlaps another argument.

format is a pointer to a character string containing two types of objects: ordinary characters (other than %), which are copied unchanged to the output, and conversion specifications, each of which is introduced by %. (To include % in the output, use %% in the format string.) A conversion specification has the following form:

            %[pos][flags][width][.prec][size]type

The fields of the conversion specification have the following meanings:

• pos

  Conversions normally consume arguments in the order that they are presented. However, it is possible to consume arguments out of order, and reuse an argument for more than one conversion specification (although the behavior is undefined if the same argument is requested with different types), by specifying pos, which is a decimal integer followed by '$'. The integer must be between 1 and <NL_ARGMAX> from limits.h, and if argument %n$ is requested, all earlier arguments must be requested somewhere within format. If positional parameters are used, then all conversion specifications except for %% must specify a position. This positional parameters method is a POSIX extension to the C standard definition for the functions.

• flags

  flags is an optional sequence of characters which control output justification, numeric signs, decimal points, trailing zeros, and octal and hex prefixes. The flag characters are minus (-), plus (+), space ( ), zero (0), sharp (#), and quote ('). They can appear in any combination, although not all flags can be used for all conversion specification types.

  '

  A POSIX extension to the C standard. However, this implementation presently treats it as a no-op, which is the default behavior for the C locale, anyway. (If it did what it is supposed to, when type were i, d, u, f, F, g, or G, the integer portion of the conversion would be formatted with thousands' grouping wide characters.)
  

  -

  The result of the conversion is left justified, and the right is padded with blanks. If you do not use this flag, the result is right justified, and padded on the left.
  

  +

  The result of a signed conversion (as determined by type of d, i, a, A, e, E, f, F, g, or G) will always begin with a plus or minus sign. (If you do not use this flag, positive values do not begin with a plus sign.)
  

  " " (space)

  If the first character of a signed conversion specification is not a sign, or if a signed conversion results in no characters, the result will begin with a space. If the space ( ) flag and the plus (+) flag both appear, the space flag is ignored.
  

  0

  If the type character is d, i, o, u, x, X, a, A, e, E, f, F, g, or G: leading zeros are used to pad the field width (following any indication of sign or base); no spaces are used for padding. If the zero (0) and minus (-) flags both appear, the zero (0) flag will be ignored. For d, i, o, u, x, and X conversions, if a precision prec is specified, the zero (0) flag is ignored.

  Note that 0 is interpreted as a flag, not as the beginning of a field width.
  

  #

  The result is to be converted to an alternative form, according to the type character.

  The alternative form output with the # flag depends on the type character:

  o

  Increases precision to force the first digit of the result to be a zero.
  

  x

  A non-zero result will have a 0x prefix.
  

  X

  A non-zero result will have a 0X prefix.
  

  a, A, e, E, f, or F

  The result will always contain a decimal point even if no digits follow the point. (Normally, a decimal point appears only if a digit follows it.) Trailing zeros are removed.
  

  g or G

  The result will always contain a decimal point even if no digits follow the point. Trailing zeros are not removed.
  

  all others

  Undefined.

• width

  width is an optional minimum field width. You can either specify it directly as a decimal integer, or indirectly by using instead an asterisk (*), in which case an int argument is used as the field width. If positional arguments are used, then the width must also be specified positionally as *m$, with m as a decimal integer. Negative field widths are treated as specifying the minus (-) flag for left justfication, along with a positive field width. The resulting format may be wider than the specified width.

• prec

  prec is an optional field; if present, it is introduced with `.' (a period). You can specify the precision either directly as a decimal integer or indirectly by using an asterisk (*), in which case an int argument is used as the precision. If positional arguments are used, then the precision must also be specified positionally as *m$, with m as a decimal integer. Supplying a negative precision is equivalent to omitting the precision. If only a period is specified the precision is zero. The effect depends on the conversion type.

  d, i, o, u, x, or X

  Minimum number of digits to appear. If no precision is given, defaults to 1.
  

  a or A

  Number of digits to appear after the decimal point. If no precision is given, the precision defaults to the minimum needed for an exact representation.
  

  e, E, f or F

  Number of digits to appear after the decimal point. If no precision is given, the precision defaults to 6.
  

  g or G

  Maximum number of significant digits. A precision of 0 is treated the same as a precision of 1. If no precision is given, the precision defaults to 6.
  

  s or S

  Maximum number of characters to print from the string. If no precision is given, the entire string is printed.
  

  all others

  undefined.

• size

  size is an optional modifier that changes the data type that the corresponding argument has. Behavior is unspecified if a size is given that does not match the type.

  hh

  With d, i, o, u, x, or X, specifies that the argument should be converted to a signed char or unsigned char before printing.

  With n, specifies that the argument is a pointer to a signed char.
  

  h

  With d, i, o, u, x, or X, specifies that the argument should be converted to a short or unsigned short before printing.

  With n, specifies that the argument is a pointer to a short.
  

  l

  With d, i, o, u, x, or X, specifies that the argument is a long or unsigned long.

  With c, specifies that the argument has type wint_t.

  With s, specifies that the argument is a pointer to wchar_t.

  With n, specifies that the argument is a pointer to a long.

  With a, A, e, E, f, F, g, or G, has no effect (because of vararg promotion rules, there is no need to distinguish between float and double).
  

  ll

  With d, i, o, u, x, or X, specifies that the argument is a long long or unsigned long long.

  With n, specifies that the argument is a pointer to a long long.
  

  j

  With d, i, o, u, x, or X, specifies that the argument is an intmax_t or uintmax_t.

  With n, specifies that the argument is a pointer to an intmax_t.
  

  z

  With d, i, o, u, x, or X, specifies that the argument is a size_t.

  With n, specifies that the argument is a pointer to a size_t.
  

  t

  With d, i, o, u, x, or X, specifies that the argument is a ptrdiff_t.

  With n, specifies that the argument is a pointer to a ptrdiff_t.
  

  L

  With a, A, e, E, f, F, g, or G, specifies that the argument is a long double.

• type

  type specifies what kind of conversion printf performs. Here is a table of these:

  %

  Prints the percent character (%).
  

  c

  Prints arg as single character. If the l size specifier is in effect, a multibyte character is printed.
  

  C

  Short for %lc. A POSIX extension to the C standard.
  

  s

  Prints the elements of a pointer to char until the precision or a null character is reached. If the l size specifier is in effect, the pointer is to an array of wchar_t, and the string is converted to multibyte characters before printing.
  

  S

  Short for %ls. A POSIX extension to the C standard.
  

  d or i

  Prints a signed decimal integer; takes an int. Leading zeros are inserted as necessary to reach the precision. A value of 0 with a precision of 0 produces an empty string.
  

  D

  Newlib extension, short for %ld.
  

  o

  Prints an unsigned octal integer; takes an unsigned. Leading zeros are inserted as necessary to reach the precision. A value of 0 with a precision of 0 produces an empty string.
  

  O

  Newlib extension, short for %lo.
  

  u

  Prints an unsigned decimal integer; takes an unsigned. Leading zeros are inserted as necessary to reach the precision. A value of 0 with a precision of 0 produces an empty string.
  

  U

  Newlib extension, short for %lu.
  

  x

  Prints an unsigned hexadecimal integer (using abcdef as digits beyond 9); takes an unsigned. Leading zeros are inserted as necessary to reach the precision. A value of 0 with a precision of 0 produces an empty string.
  

  X

  Like x, but uses ABCDEF as digits beyond 9.
  

  f

  Prints a signed value of the form [-]9999.9999, with the precision determining how many digits follow the decimal point; takes a double (remember that float promotes to double as a vararg). The low order digit is rounded to even. If the precision results in at most DECIMAL_DIG digits, the result is rounded correctly; if more than DECIMAL_DIG digits are printed, the result is only guaranteed to round back to the original value.

  If the value is infinite, the result is inf, and no zero padding is performed. If the value is not a number, the result is nan, and no zero padding is performed.
  

  F

  Like f, but uses INF and NAN for non-finite numbers.
  

  e

  Prints a signed value of the form [-]9.9999e[+|-]999; takes a double. The digit before the decimal point is non-zero if the value is non-zero. The precision determines how many digits appear between . and e, and the exponent always contains at least two digits. The value zero has an exponent of zero. If the value is not finite, it is printed like f.
  

  E

  Like e, but using E to introduce the exponent, and like F for non-finite values.
  

  g

  Prints a signed value in either f or e form, based on the given value and precision—an exponent less than -4 or greater than the precision selects the e form. Trailing zeros and the decimal point are printed only if necessary; takes a double.
  

  G

  Like g, except use F or E form.
  

  a

  Prints a signed value of the form [-]0x1.ffffp[+|-]9; takes a double. The letters abcdef are used for digits beyond 9. The precision determines how many digits appear after the decimal point. The exponent contains at least one digit, and is a decimal value representing the power of 2; a value of 0 has an exponent of 0. Non-finite values are printed like f.
  

  A

  Like a, except uses X, P, and ABCDEF instead of lower case.
  

  n

  Takes a pointer to int, and stores a count of the number of bytes written so far. No output is created.
  

  p

  Takes a pointer to void, and prints it in an implementation-defined format. This implementation is similar to %#tx), except that 0x appears even for the NULL pointer.
  

  m

  Prints the output of strerror(errno); no argument is required. A GNU extension.

_printf_r, _fprintf_r, _asprintf_r, _sprintf_r, _snprintf_r, _asnprintf_r are simply reentrant versions of the functions above.

If an error occurs, the result of printf, fprintf, snprintf, and asprintf is a negative value, and the result of asnprintf is NULL. No error returns occur for sprintf. For printf and fprintf, errno may be set according to fputc. For asprintf and asnprintf, errno may be set to ENOMEM if allocation fails, and for snprintf, errno may be set to EOVERFLOW if size or the output length exceeds INT_MAX.

The ANSI C standard specifies that implementations must support at least formatted output of up to 509 characters. This implementation has no inherent limit.

Depending on how newlib was configured, not all format specifiers are supported.

Supporting OS subroutines required: close, fstat, isatty, lseek, read, sbrk, write.