Difference between revisions of "cpp/io/c/fscanf"

Latest revision as of 10:24, 10 September 2023

Reads data from a variety of sources, interprets it according to format and stores the results into given locations.

[edit] Parameters

The format string consists of

• non-whitespace multibyte characters except %: each such character in the format string consumes exactly one identical character from the input stream, or causes the function to fail if the next character on the stream does not compare equal.
• whitespace characters: any single whitespace character in the format string consumes all available consecutive whitespace characters from the input (determined as if by calling isspace in a loop). Note that there is no difference between "\n", " ", "\t\t", or other whitespace in the format string.
• conversion specifications. Each conversion specification has the following format:

• introductory % character.

• (optional) assignment-suppressing character *. If this option is present, the function does not assign the result of the conversion to any receiving argument.

• (optional) integer number (greater than zero) that specifies maximum field width, that is, the maximum number of characters that the function is allowed to consume when doing the conversion specified by the current conversion specification. Note that %s and %[ may lead to buffer overflow if the width is not provided.

• (optional) length modifier that specifies the size of the receiving argument, that is, the actual destination type. This affects the conversion accuracy and overflow rules. The default destination type is different for each conversion type (see table below).

• conversion format specifier.

The following format specifiers are available:

For every conversion specifier other than n, the longest sequence of input characters which does not exceed any specified field width and which either is exactly what the conversion specifier expects or is a prefix of a sequence it would expect, is what's consumed from the stream. The first character, if any, after this consumed sequence remains unread. If the consumed sequence has length zero or if the consumed sequence cannot be converted as specified above, the matching failure occurs unless end-of-file, an encoding error, or a read error prevented input from the stream, in which case it is an input failure.

All conversion specifiers other than [, c, and n consume and discard all leading whitespace characters (determined as if by calling isspace) before attempting to parse the input. These consumed characters do not count towards the specified maximum field width.

The conversion specifiers lc, ls, and l[ perform multibyte-to-wide character conversion as if by calling mbrtowc with an mbstate_t object initialized to zero before the first character is converted.

The conversion specifiers s and [ always store the null terminator in addition to the matched characters. The size of the destination array must be at least one greater than the specified field width. The use of %s or %[, without specifying the destination array size, is as unsafe as std::gets.

The correct conversion specifications for the fixed-width integer types (int8_t, etc) are defined in the header <cinttypes> (although SCNdMAX, SCNuMAX, etc is synonymous with %jd, %ju, etc).

There is a sequence point after the action of each conversion specifier; this permits storing multiple fields in the same "sink" variable.

When parsing an incomplete floating-point value that ends in the exponent with no digits, such as parsing "100er" with the conversion specifier %f, the sequence "100e" (the longest prefix of a possibly valid floating-point number) is consumed, resulting in a matching error (the consumed sequence cannot be converted to a floating-point number), with "r" remaining. Some existing implementations do not follow this rule and roll back to consume only "100", leaving "er", e.g. glibc bug 1765.

[edit] Return value

Number of receiving arguments successfully assigned (which may be zero in case a matching failure occurred before the first receiving argument was assigned), or EOF if input failure occurs before the first receiving argument was assigned.

[edit] Complexity

Not guaranteed. Notably, some implementations of std::sscanf are O(N), where N = std::strlen(buffer) [1]. For performant string parsing, see std::from_chars.

[edit] Notes

Because most conversion specifiers first consume all consecutive whitespace, code such as

std::scanf("%d", &a);
std::scanf("%d", &b);

will read two integers that are entered on different lines (second %d will consume the newline left over by the first) or on the same line, separated by spaces or tabs (second %d will consume the spaces or tabs).

std::scanf("%d", &a);
std::scanf(" %c", &c); // ignore the endline after %d, then read a char

Note that some implementations of std::sscanf involve a call to std::strlen, which makes their runtime linear on the length of the entire string. This means that if std::sscanf is called in a loop to repeatedly parse values from the front of a string, your code might run in quadratic time (example).

[edit] Example

#include <clocale>
#include <cstdio>
#include <iostream>
 
int main()
{
    int i, j;
    float x, y;
    char str1[10], str2[4];
    wchar_t warr[2];
    std::setlocale(LC_ALL, "en_US.utf8");
 
    char input[] = "25 54.32E-1 Thompson 56789 0123 56ß水";
    // parse as follows:
    // %d: an integer 
    // %f: a floating-point value
    // %9s: a string of at most 9 non-whitespace characters
    // %2d: two-digit integer (digits 5 and 6)
    // %f: a floating-point value (digits 7, 8, 9)
    // %*d an integer which isn't stored anywhere
    // ' ': all consecutive whitespace
    // %3[0-9]: a string of at most 3 digits (digits 5 and 6)
    // %2lc: two wide characters, using multibyte to wide conversion
    const int ret = std::sscanf(input, "%d%f%9s%2d%f%*d %3[0-9]%2lc",
                                &i, &x, str1, &j, &y, str2, warr);
 
    std::cout << "Converted " << ret << " fields:\n"
                 "i = " << i << "\n"
                 "x = " << x << "\n"
                 "str1 = " << str1 << "\n"
                 "j = " << j << "\n"
                 "y = " << y << "\n"
                 "str2 = " << str2 << std::hex << "\n"
                 "warr[0] = U+" << (int)warr[0] << "\n"
                 "warr[1] = U+" << (int)warr[1] << '\n';
}

Converted 7 fields:
i = 25
x = 5.432
str1 = Thompson
j = 56
y = 789
str2 = 56
warr[0] = U+df
warr[1] = U+6c34

[edit] See also