snprintf.c
- yet another portable implementation of snprintf

snprintf is a routine to convert numeric and string arguments to formatted strings. It is similar to sprintf(3) provided in a system's C library, yet it requires an additional argument - the buffer size - and it guarantees never to store anything beyond the given buffer, regardless of the format or arguments to be formatted. Some newer operating systems do provide snprintf in their C library, but many do not or do provide an inadequate (slow or idiosyncratic) version, which calls for a portable implementation of this routine.

Author

Terms and conditions ...

... for copying, distribution and modification, NO WARRANTY:
GNU GENERAL PUBLIC LICENSE, Version 2 or later

Features

careful adherence to specs regarding flags, field width and precision;
good performance for large string handling (large format, large argument or large paddings). Performance is similar to system's sprintf and in several cases significantly better (make sure you compile with optimizations turned on, tell the compiler the code is strict ANSI if necessary to give it more freedom for optimizations);
written in standard ANSI C - requires an ANSI C compiler.

Supported formats and data types

This snprintf only supports format specifiers: s, c, d, o, u, x, X, p (and synonyms: i, D, U, O - see below) with flags: '-', '+', ' ', '0' and '#'. An asterisk is supported for field width as well as precision.

Data type modifiers 'h' (short int), 'l' (long int) and 'll' (long long int) are supported.

NOTE:

If macro SNPRINTF_LONGLONG_SUPPORT is not defined (default) the data type modifier 'll' is recognized but treated the same as 'l', which may cause argument value truncation! Defining SNPRINTF_LONGLONG_SUPPORT requires that your system's sprintf also handles data type modifier 'll'. long long int is a language extension which may not be portable.

Conversion of numeric data (formats d, o, u, x, X, p) with data type modifiers (none or h, l, ll) is left to the system routine sprintf, but all handling of flags, field width and precision as well as c and s formats is done very carefully by this portable routine. If a string precision (truncation) is specified (e.g. %.8s) it is guaranteed the string beyond the specified precision will not be referenced.

Data type modifiers h, l and ll are ignored for c and s formats (data types wint_t and wchar_t are not supported).

The following common synonyms for conversion characters are supported:

i is a synonym for d
D is a synonym for ld, explicit data type modifiers are ignored
U is a synonym for lu, explicit data type modifiers are ignored
O is a synonym for lo, explicit data type modifiers are ignored

The following is specifically not supported:

flag ' (thousands' grouping character) is recognized but ignored
numeric formats: f, e, E, g, G and synonym F
data type modifier 'L' (long double) and 'q' (quad - use 'll' instead)
wide character/string formats: C, lc, S, ls
writeback of converted string length: conversion character n
the n$ specification for direct reference to n-th argument
locales

Availability

http://www.ijs.si/software/snprintf/

distribution: snprintf_1.2.tar.gz
md5 sum: snprintf_1.2.tar.gz.md5

Other implementations of snprintf

I am aware of some other portable implementations of snprintf. I do not claim they are public - please refer to their respective copyright and licensing terms. If you know of other versions please let me know.

a very thorough implementation (src/util_snprintf.c) by the Apache Group distributed with the Apache web server - http://www.apache.org/ . Does its own floating point conversions using routines ecvt(3), fcvt(3) and gcvt(3) from the standard C library or from the GNU libc.
This is from the code:
This software [...] was originally based on public domain software written at the National Center for Supercomputing Applications, University of Illinois, Urbana-Champaign.
[...] This code is based on, and used with the permission of, the SIO stdio-replacement strx_* functions by Panos Tsirigotis <panos@alumni.cs.colorado.edu> for xinetd.
original implementation from Prof. Patrick Powell <papowell@sdsu.edu>, Dept. Electrical and Computer Engineering, San Diego State University, San Diego, CA 92182-1309, published in Bugtraq archives for 3rd quarter (Jul-Aug) 1995. No floating point conversions.
Brandon Long's <blong@fiction.net> modified version of Prof. Patrick Powell's snprintf with contributions from others. With minimal floating point support.
implementation (src/snprintf.c) as distributed with sendmail - http://www.sendmail.org/ is a cleaned up Prof. Patrick Powell's version to compile properly and to support .precision and %lx.
implementation used by newlog (a replacement for syslog(3)) made available by the SOS Corporation. Enabling floating point support is a compile-time option.
implementation from Tomi Salo <ttsalo@ssh.fi> distributed with SSH 2.0 Unix Server. Not in public domain. Floating point conversions done by system's sprintf.
and for completeness: my portable version described in this very document available at http://www.ijs.si/software/snprintf/ .

In retrospect, it appears that a lot of effort was wasted by many people for not being aware of what others are doing. Sigh.

snprintf.c - yet another portable implementation of snprintf