source: mainline/common/printf/printf_core.c@ e77c3ed

/*
 * Copyright (c) 2001-2004 Jakub Jermar
 * Copyright (c) 2006 Josef Cejka
 * Copyright (c) 2009 Martin Decky
 * Copyright (c) 2025 Jiří Zárevúcky
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 *
 * - Redistributions of source code must retain the above copyright
 *   notice, this list of conditions and the following disclaimer.
 * - Redistributions in binary form must reproduce the above copyright
 *   notice, this list of conditions and the following disclaimer in the
 *   documentation and/or other materials provided with the distribution.
 * - The name of the author may not be used to endorse or promote products
 *   derived from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

/** @addtogroup libc
 * @{
 */
/**
 * @file
 * @brief Printing functions.
 */

#include <_bits/uchar.h>
#include <_bits/wint_t.h>
#include <assert.h>
#include <ctype.h>
#include <errno.h>
#include <limits.h>
#include <macros.h>
#include <printf_core.h>
#include <stddef.h>
#include <stdint.h>
#include <stdlib.h>
#include <str.h>

/* Disable float support in kernel, because we usually disable floating operations there. */
#if __STDC_HOSTED__
#define HAS_FLOAT
#endif

#ifdef HAS_FLOAT
#include <double_to_str.h>
#include <ieee_double.h>
#endif

/** show prefixes 0x or 0 */
#define __PRINTF_FLAG_PREFIX       0x00000001

/** show the decimal point even if no fractional digits present */
#define __PRINTF_FLAG_DECIMALPT    0x00000001

/** signed / unsigned number */
#define __PRINTF_FLAG_SIGNED       0x00000002

/** print leading zeroes */
#define __PRINTF_FLAG_ZEROPADDED   0x00000004

/** align to left */
#define __PRINTF_FLAG_LEFTALIGNED  0x00000010

/** always show + sign */
#define __PRINTF_FLAG_SHOWPLUS     0x00000020

/** print space instead of plus */
#define __PRINTF_FLAG_SPACESIGN    0x00000040

/** show big characters */
#define __PRINTF_FLAG_BIGCHARS     0x00000080

/** number has - sign */
#define __PRINTF_FLAG_NEGATIVE     0x00000100

/** don't print trailing zeros in the fractional part */
#define __PRINTF_FLAG_NOFRACZEROS  0x00000200

/**
 * Buffer big enough for 64-bit number printed in base 2.
 */
#define PRINT_NUMBER_BUFFER_SIZE  64

/** Get signed or unsigned integer argument */
#define PRINTF_GET_INT_ARGUMENT(type, ap, flags) \
        ({ \
                unsigned type res; \
                \
                if ((flags) & __PRINTF_FLAG_SIGNED) { \
                        signed type arg = va_arg((ap), signed type); \
                        \
                        if (arg < 0) { \
                                res = -arg; \
                                (flags) |= __PRINTF_FLAG_NEGATIVE; \
                        } else \
                                res = arg; \
                } else \
                        res = va_arg((ap), unsigned type); \
                \
                res; \
        })

/** Enumeration of possible arguments types.
 */
typedef enum {
        PrintfQualifierByte = 0,
        PrintfQualifierShort,
        PrintfQualifierInt,
        PrintfQualifierLong,
        PrintfQualifierLongLong,
        PrintfQualifierPointer,
} qualifier_t;

static const char _digits_small[] = "0123456789abcdef";
static const char _digits_big[] = "0123456789ABCDEF";

static const char _nullstr[] = "(NULL)";
static const char _replacement[] = u8"�";
static const char _spaces[] = "                                               ";
static const char _zeros[] = "000000000000000000000000000000000000000000000000";

static void _set_errno(errno_t rc)
{
        #ifdef errno
        errno = rc;
        #endif
}

static size_t _utf8_bytes(char32_t c)
{
        if (c < 0x80)
                return 1;

        if (c < 0x800)
                return 2;

        if (c < 0xD800)
                return 3;

        /* Surrogate code points, invalid in UTF-32. */
        if (c < 0xE000)
                return sizeof(_replacement) - 1;

        if (c < 0x10000)
                return 3;

        if (c < 0x110000)
                return 4;

        /* Invalid character. */
        return sizeof(_replacement) - 1;
}

/** Counts characters and utf8 bytes in a wide string up to a byte limit.
 * @param max_bytes    Byte length limit for string's utf8 conversion.
 * @param[out] len     The number of wide characters
 * @return  Number of utf8 bytes that the first *len characters in the string
 *          will convert to. Will always be less than max_bytes.
 */
static size_t _utf8_wstr_bytes_len(char32_t *s, size_t max_bytes, size_t *len)
{
        size_t bytes = 0;
        size_t i;

        for (i = 0; bytes < max_bytes && s[i]; i++) {
                size_t next = _utf8_bytes(s[i]);
                if (max_bytes - bytes < next)
                        break;

                bytes += next;
        }

        *len = i;
        return bytes;
}

#define TRY(expr) ({ errno_t rc = (expr); if (rc != EOK) return rc; })

static inline void _saturating_add(size_t *a, size_t b)
{
        size_t s = *a + b;
        /* Only works because size_t is unsigned. */
        *a = (s < b) ? SIZE_MAX : s;
}

static errno_t _write_bytes(const char *buf, size_t n, printf_spec_t *ps,
    size_t *written_bytes)
{
        errno_t rc = ps->write(buf, n, ps->data);
        if (rc != EOK)
                return rc;

        _saturating_add(written_bytes, n);
        return EOK;
}

/** Write one UTF-32 character. */
static errno_t _write_uchar(char32_t ch, printf_spec_t *ps,
    size_t *written_bytes)
{
        char utf8[4];
        size_t offset = 0;

        if (chr_encode(ch, utf8, &offset, sizeof(utf8)) == EOK)
                return _write_bytes(utf8, offset, ps, written_bytes);

        /* Invalid character. */
        return _write_bytes(_replacement, sizeof(_replacement) - 1, ps, written_bytes);
}

/** Write n UTF-32 characters. */
static errno_t _write_chars(const char32_t *buf, size_t n, printf_spec_t *ps,
    size_t *written_bytes)
{
        for (size_t i = 0; i < n; i++)
                TRY(_write_uchar(buf[i], ps, written_bytes));

        return EOK;
}

static errno_t _write_char(char c, printf_spec_t *ps, size_t *written_bytes)
{
        return _write_bytes(&c, 1, ps, written_bytes);
}

static errno_t _write_spaces(size_t n, printf_spec_t *ps, size_t *written_bytes)
{
        size_t max_spaces = sizeof(_spaces) - 1;

        while (n > max_spaces) {
                TRY(_write_bytes(_spaces, max_spaces, ps, written_bytes));
                n -= max_spaces;
        }

        return _write_bytes(_spaces, n, ps, written_bytes);
}

static errno_t _write_zeros(size_t n, printf_spec_t *ps, size_t *written_bytes)
{
        size_t max_zeros = sizeof(_zeros) - 1;

        while (n > max_zeros) {
                TRY(_write_bytes(_zeros, max_zeros, ps, written_bytes));
                n -= max_zeros;
        }

        return _write_bytes(_zeros, n, ps, written_bytes);
}

/** Print one formatted ASCII character.
 *
 * @param ch    Character to print.
 * @param width Width modifier.
 * @param flags Flags that change the way the character is printed.
 */
static errno_t _format_char(const char c, size_t width, uint32_t flags,
    printf_spec_t *ps, size_t *written_bytes)
{
        size_t bytes = 1;

        if (width <= bytes)
                return _write_char(c, ps, written_bytes);

        if (flags & __PRINTF_FLAG_LEFTALIGNED) {
                TRY(_write_char(c, ps, written_bytes));
                TRY(_write_spaces(width - bytes, ps, written_bytes));
        } else {
                TRY(_write_spaces(width - bytes, ps, written_bytes));
                TRY(_write_char(c, ps, written_bytes));
        }

        return EOK;
}

/** Print one formatted wide character.
 *
 * @param ch    Character to print.
 * @param width Width modifier.
 * @param flags Flags that change the way the character is printed.
 */
static errno_t _format_uchar(const char32_t ch, size_t width, uint32_t flags,
    printf_spec_t *ps, size_t *written_bytes)
{
        /*
         * All widths in printf() are specified in bytes. It might seem nonsensical
         * with unicode text, but that's the way the function is defined. The width
     * is barely useful if you want column alignment in terminal, but keep in
     * mind that counting code points is only marginally better for that.
     * Characters can span more than one unicode code point, even in languages
     * based on latin alphabet, and a single unicode code point can occupy two
     * spaces in east asian scripts.
     *
     * What the width can actually be useful for is padding, when you need the
     * output to fill an exact number of bytes in a file. That use would break
     * if we did our own thing here.
     */

    size_t bytes = _utf8_bytes(ch);

        if (width <= bytes)
                return _write_uchar(ch, ps, written_bytes);

        if (flags & __PRINTF_FLAG_LEFTALIGNED) {
                TRY(_write_uchar(ch, ps, written_bytes));
                TRY(_write_spaces(width - bytes, ps, written_bytes));
        } else {
                TRY(_write_spaces(width - bytes, ps, written_bytes));
                TRY(_write_uchar(ch, ps, written_bytes));
        }

        return EOK;
}

/** Print string.
 *
 * @param str       String to be printed.
 * @param width     Width modifier.
 * @param precision Precision modifier.
 * @param flags     Flags that modify the way the string is printed.
 */
static errno_t _format_cstr(const char *str, size_t width, int precision,
    uint32_t flags, printf_spec_t *ps, size_t *written_bytes)
{
        if (str == NULL)
                str = _nullstr;

        /* Negative precision == unspecified. */
        size_t max_bytes = (precision < 0) ? SIZE_MAX : (size_t) precision;
        size_t bytes = str_nsize(str, max_bytes);

        if (width <= bytes)
                return _write_bytes(str, bytes, ps, written_bytes);

        if (flags & __PRINTF_FLAG_LEFTALIGNED) {
                TRY(_write_bytes(str, bytes, ps, written_bytes));
                TRY(_write_spaces(width - bytes, ps, written_bytes));
        } else {
                TRY(_write_spaces(width - bytes, ps, written_bytes));
                TRY(_write_bytes(str, bytes, ps, written_bytes));
        }

        return EOK;
}

/** Print wide string.
 *
 * @param str       Wide string to be printed.
 * @param width     Width modifier.
 * @param precision Precision modifier.
 * @param flags     Flags that modify the way the string is printed.
 */
static errno_t _format_wstr(char32_t *str, size_t width, int precision,
    uint32_t flags, printf_spec_t *ps, size_t *written_bytes)
{
        if (!str)
                return _format_cstr(_nullstr, width, precision, flags, ps, written_bytes);

        /* Width and precision are always byte-based. See _format_uchar() */
        /* Negative precision == unspecified. */
        size_t max_bytes = (precision < 0) ? SIZE_MAX : (size_t) precision;

        size_t len;
        size_t bytes = _utf8_wstr_bytes_len(str, max_bytes, &len);

        if (width <= bytes)
                return _write_chars(str, len, ps, written_bytes);

        if (flags & __PRINTF_FLAG_LEFTALIGNED) {
                TRY(_write_chars(str, len, ps, written_bytes));
                TRY(_write_spaces(width - bytes, ps, written_bytes));
        } else {
                TRY(_write_spaces(width - bytes, ps, written_bytes));
                TRY(_write_chars(str, len, ps, written_bytes));
        }

        return EOK;
}

static char _sign(uint32_t flags)
{
        if (!(flags & __PRINTF_FLAG_SIGNED))
                return 0;

        if (flags & __PRINTF_FLAG_NEGATIVE)
                return '-';

        if (flags & __PRINTF_FLAG_SHOWPLUS)
                return '+';

        if (flags & __PRINTF_FLAG_SPACESIGN)
                return ' ';

        return 0;
}

/** Print a number in a given base.
 *
 * Print significant digits of a number in given base.
 *
 * @param num       Number to print.
 * @param width     Width modifier.
 * @param precision Precision modifier.
 * @param base      Base to print the number in (must be between 2 and 16).
 * @param flags     Flags that modify the way the number is printed.
 */
static errno_t _format_number(uint64_t num, size_t width, int precision, int base,
    uint32_t flags, printf_spec_t *ps, size_t *written_bytes)
{
        assert(base >= 2 && base <= 16);

        /* Default precision for numeric output is 1. */
        size_t min_digits = (precision < 0) ? 1 : precision;

        bool bigchars = flags & __PRINTF_FLAG_BIGCHARS;
        bool prefix = flags & __PRINTF_FLAG_PREFIX;
        bool left_aligned = flags & __PRINTF_FLAG_LEFTALIGNED;
        bool zero_padded = flags & __PRINTF_FLAG_ZEROPADDED;

        const char *digits = bigchars ? _digits_big : _digits_small;

        char buffer[PRINT_NUMBER_BUFFER_SIZE];
        char *end = &buffer[PRINT_NUMBER_BUFFER_SIZE];

        /* Write number to the buffer. */
        int offset = 0;
        while (num > 0) {
                end[--offset] = digits[num % base];
                num /= base;
        }

        char *number = &end[offset];
        size_t number_len = end - number;
        char sign = _sign(flags);

        if (left_aligned) {
                /* Space padded right-aligned. */
                size_t real_size = max(number_len, min_digits);

                if (sign) {
                        TRY(_write_char(sign, ps, written_bytes));
                        real_size++;
                }

                if (prefix && base == 2 && number_len > 0) {
                        TRY(_write_bytes(bigchars ? "0B" : "0b", 2, ps, written_bytes));
                        real_size += 2;
                }

                if (prefix && base == 16 && number_len > 0) {
                        TRY(_write_bytes(bigchars ? "0X" : "0x", 2, ps, written_bytes));
                        real_size += 2;
                }

                if (min_digits > number_len) {
                        TRY(_write_zeros(min_digits - number_len, ps, written_bytes));
                } else if (prefix && base == 8) {
                        TRY(_write_zeros(1, ps, written_bytes));
                        real_size++;
                }

                TRY(_write_bytes(number, number_len, ps, written_bytes));

                if (width > real_size)
                        TRY(_write_spaces(width - real_size, ps, written_bytes));

                return EOK;
        }

        /* Zero padded number (ignored when left aligned or if precision is specified). */
        if (precision < 0 && zero_padded) {
                size_t real_size = number_len;

                if (sign) {
                        TRY(_write_char(sign, ps, written_bytes));
                        real_size++;
                }

                if (prefix && base == 2 && number_len > 0) {
                        TRY(_write_bytes(bigchars ? "0B" : "0b", 2, ps, written_bytes));
                        real_size += 2;
                }

                if (prefix && base == 16 && number_len > 0) {
                        TRY(_write_bytes(bigchars ? "0X" : "0x", 2, ps, written_bytes));
                        real_size += 2;
                }

                if (width > real_size)
                        TRY(_write_zeros(width - real_size, ps, written_bytes));
                else if (number_len == 0 || (prefix && base == 8))
                        TRY(_write_char('0', ps, written_bytes));

                return _write_bytes(number, number_len, ps, written_bytes);
        }

        /* Space padded right-aligned. */
        size_t real_size = max(number_len, min_digits);
        if (sign)
                real_size++;

        if (prefix && (base == 2 || base == 16) && number_len > 0)
                real_size += 2;

        if (prefix && base == 8 && number_len >= min_digits)
                real_size += 1;

        if (width > real_size)
                TRY(_write_spaces(width - real_size, ps, written_bytes));

        if (sign)
                TRY(_write_char(sign, ps, written_bytes));

        if (prefix && base == 2 && number_len > 0)
                TRY(_write_bytes(bigchars ? "0B" : "0b", 2, ps, written_bytes));

        if (prefix && base == 16 && number_len > 0)
                TRY(_write_bytes(bigchars ? "0X" : "0x", 2, ps, written_bytes));

        if (min_digits > number_len)
                TRY(_write_zeros(min_digits - number_len, ps, written_bytes));
        else if (prefix && base == 8)
                TRY(_write_char('0', ps, written_bytes));

        return _write_bytes(number, number_len, ps, written_bytes);
}

#ifdef HAS_FLOAT

/** Unformatted double number string representation. */
typedef struct {
        /** Buffer with len digits, no sign or leading zeros. */
        char *str;
        /** Number of digits in str. */
        int len;
        /** Decimal exponent, ie number = str * 10^dec_exp */
        int dec_exp;
        /** True if negative. */
        bool neg;
} double_str_t;

/** Returns the sign character or 0 if no sign should be printed. */
static char _get_sign_char(bool negative, uint32_t flags)
{
        if (negative) {
                return '-';
        } else if (flags & __PRINTF_FLAG_SHOWPLUS) {
                return '+';
        } else if (flags & __PRINTF_FLAG_SPACESIGN) {
                return ' ';
        } else {
                return 0;
        }
}

/** Prints a special double (ie NaN, infinity) padded to width characters. */
static errno_t _format_special(ieee_double_t val, int width, uint32_t flags,
    printf_spec_t *ps, size_t *written_bytes)
{
        assert(val.is_special);

        char sign = _get_sign_char(val.is_negative, flags);

        const int str_len = 3;
        const char *str;

        if (flags & __PRINTF_FLAG_BIGCHARS) {
                str = val.is_infinity ? "INF" : "NAN";
        } else {
                str = val.is_infinity ? "inf" : "nan";
        }

        int padding_len = max(0, width - ((sign ? 1 : 0) + str_len));

        /* Leading padding. */
        if (!(flags & __PRINTF_FLAG_LEFTALIGNED))
                TRY(_write_spaces(padding_len, ps, written_bytes));

        if (sign)
                TRY(_write_char(sign, ps, written_bytes));

        TRY(_write_bytes(str, str_len, ps, written_bytes));

        /* Trailing padding. */
        if (flags & __PRINTF_FLAG_LEFTALIGNED)
                TRY(_write_spaces(padding_len, ps, written_bytes));

        return EOK;
}

/** Trims trailing zeros but leaves a single "0" intact. */
static void _fp_trim_trailing_zeros(char *buf, int *len, int *dec_exp)
{
        /* Cut the zero off by adjusting the exponent. */
        while (2 <= *len && '0' == buf[*len - 1]) {
                --*len;
                ++*dec_exp;
        }
}

/** Textually round up the last digit thereby eliminating it. */
static void _fp_round_up(char *buf, int *len, int *dec_exp)
{
        assert(1 <= *len);

        char *last_digit = &buf[*len - 1];

        int carry = ('5' <= *last_digit);

        /* Cut the digit off by adjusting the exponent. */
        --*len;
        ++*dec_exp;
        --last_digit;

        if (carry) {
                /* Skip all the digits to cut off/round to zero. */
                while (buf <= last_digit && '9' == *last_digit) {
                        --last_digit;
                }

                /* last_digit points to the last digit to round but not '9' */
                if (buf <= last_digit) {
                        *last_digit += 1;
                        int new_len = last_digit - buf + 1;
                        *dec_exp += *len - new_len;
                        *len = new_len;
                } else {
                        /* All len digits rounded to 0. */
                        buf[0] = '1';
                        *dec_exp += *len;
                        *len = 1;
                }
        } else {
                /* The only digit was rounded to 0. */
                if (last_digit < buf) {
                        buf[0] = '0';
                        *dec_exp = 0;
                        *len = 1;
                }
        }
}

/** Format and print the double string repressentation according
 *  to the %f specifier.
 */
static errno_t _format_double_str_fixed(double_str_t *val_str, int precision, int width,
    uint32_t flags, printf_spec_t *ps, size_t *written_bytes)
{
        int len = val_str->len;
        char *buf = val_str->str;
        int dec_exp = val_str->dec_exp;

        assert(0 < len);
        assert(0 <= precision);
        assert(0 <= dec_exp || -dec_exp <= precision);

        /* Number of integral digits to print (at least leading zero). */
        int int_len = max(1, len + dec_exp);

        char sign = _get_sign_char(val_str->neg, flags);

        /* Fractional portion lengths. */
        int last_frac_signif_pos = max(0, -dec_exp);
        int leading_frac_zeros = max(0, last_frac_signif_pos - len);
        int signif_frac_figs = min(last_frac_signif_pos, len);
        int trailing_frac_zeros = precision - last_frac_signif_pos;
        char *buf_frac = buf + len - signif_frac_figs;

        if (flags & __PRINTF_FLAG_NOFRACZEROS)
                trailing_frac_zeros = 0;

        int frac_len = leading_frac_zeros + signif_frac_figs + trailing_frac_zeros;

        bool has_decimal_pt = (0 < frac_len) || (flags & __PRINTF_FLAG_DECIMALPT);

        /* Number of non-padding chars to print. */
        int num_len = (sign ? 1 : 0) + int_len + (has_decimal_pt ? 1 : 0) + frac_len;

        int padding_len = max(0, width - num_len);

        /* Leading padding and sign. */

        if (!(flags & (__PRINTF_FLAG_LEFTALIGNED | __PRINTF_FLAG_ZEROPADDED)))
                TRY(_write_spaces(padding_len, ps, written_bytes));

        if (sign)
                TRY(_write_char(sign, ps, written_bytes));

        if (flags & __PRINTF_FLAG_ZEROPADDED)
                TRY(_write_zeros(padding_len, ps, written_bytes));

        /* Print the intergral part of the buffer. */

        int buf_int_len = min(len, len + dec_exp);

        if (0 < buf_int_len) {
                TRY(_write_bytes(buf, buf_int_len, ps, written_bytes));

                /* Print trailing zeros of the integral part of the number. */
                TRY(_write_zeros(int_len - buf_int_len, ps, written_bytes));
        } else {
                /* Single leading integer 0. */
                TRY(_write_char('0', ps, written_bytes));
        }

        /* Print the decimal point and the fractional part. */
        if (has_decimal_pt) {
                TRY(_write_char('.', ps, written_bytes));

                /* Print leading zeros of the fractional part of the number. */
                TRY(_write_zeros(leading_frac_zeros, ps, written_bytes));

                /* Print significant digits of the fractional part of the number. */
                if (0 < signif_frac_figs)
                        TRY(_write_bytes(buf_frac, signif_frac_figs, ps, written_bytes));

                /* Print trailing zeros of the fractional part of the number. */
                TRY(_write_zeros(trailing_frac_zeros, ps, written_bytes));
        }

        /* Trailing padding. */
        if (flags & __PRINTF_FLAG_LEFTALIGNED)
                TRY(_write_spaces(padding_len, ps, written_bytes));

        return EOK;
}

/** Convert, format and print a double according to the %f specifier.
 *
 * @param g     Double to print.
 * @param precision Number of fractional digits to print. If 0 no
 *              decimal point will be printed unless the flag
 *              __PRINTF_FLAG_DECIMALPT is specified.
 * @param width Minimum number of characters to display. Pads
 *              with '0' or ' ' depending on the set flags;
 * @param flags Printf flags.
 * @param ps    Printing functions.
 */
static errno_t _format_double_fixed(double g, int precision, int width,
        uint32_t flags, printf_spec_t *ps, size_t *written_bytes)
{
        if (flags & __PRINTF_FLAG_LEFTALIGNED) {
                flags &= ~__PRINTF_FLAG_ZEROPADDED;
        }

        if (flags & __PRINTF_FLAG_DECIMALPT) {
                flags &= ~__PRINTF_FLAG_NOFRACZEROS;
        }

        ieee_double_t val = extract_ieee_double(g);

        if (val.is_special) {
                return _format_special(val, width, flags, ps, written_bytes);
        }

        char buf[MAX_DOUBLE_STR_BUF_SIZE];
        const size_t buf_size = MAX_DOUBLE_STR_BUF_SIZE;
        double_str_t val_str;

        val_str.str = buf;
        val_str.neg = val.is_negative;

        if (0 <= precision) {
                /*
                 * Request one more digit so we can round the result. The last
                 * digit it returns may have an error of at most +/- 1.
                 */
                val_str.len = double_to_fixed_str(val, -1, precision + 1, buf, buf_size,
                    &val_str.dec_exp);

                /*
                 * Round using the last digit to produce precision fractional digits.
                 * If less than precision+1 fractional digits were output the last
                 * digit is definitely inaccurate so also round to get rid of it.
                 */
                _fp_round_up(buf, &val_str.len, &val_str.dec_exp);

                /* Rounding could have introduced trailing zeros. */
                if (flags & __PRINTF_FLAG_NOFRACZEROS) {
                        _fp_trim_trailing_zeros(buf, &val_str.len, &val_str.dec_exp);
                }
        } else {
                /* Let the implementation figure out the proper precision. */
                val_str.len = double_to_short_str(val, buf, buf_size, &val_str.dec_exp);

                /* Precision needed for the last significant digit. */
                precision = max(0, -val_str.dec_exp);
        }

        return _format_double_str_fixed(&val_str, precision, width, flags, ps, written_bytes);
}

/** Prints the decimal exponent part of a %e specifier formatted number. */
static errno_t _format_exponent(int exp_val, uint32_t flags, printf_spec_t *ps,
    size_t *written_bytes)
{
        char exp_ch = (flags & __PRINTF_FLAG_BIGCHARS) ? 'E' : 'e';
        TRY(_write_char(exp_ch, ps, written_bytes));

        char exp_sign = (exp_val < 0) ? '-' : '+';
        TRY(_write_char(exp_sign, ps, written_bytes));

        /* Print the exponent. */
        exp_val = abs(exp_val);

        char exp_str[4] = { 0 };

        exp_str[0] = '0' + exp_val / 100;
        exp_str[1] = '0' + (exp_val % 100) / 10;
        exp_str[2] = '0' + (exp_val % 10);

        int exp_len = (exp_str[0] == '0') ? 2 : 3;
        const char *exp_str_start = &exp_str[3] - exp_len;

        return _write_bytes(exp_str_start, exp_len, ps, written_bytes);
}

/** Format and print the double string repressentation according
 *  to the %e specifier.
 */
static errno_t _format_double_str_scient(double_str_t *val_str, int precision,
    int width, uint32_t flags, printf_spec_t *ps, size_t *written_bytes)
{
        int len = val_str->len;
        int dec_exp = val_str->dec_exp;
        char *buf  = val_str->str;

        assert(0 < len);

        char sign = _get_sign_char(val_str->neg, flags);
        bool has_decimal_pt = (0 < precision) || (flags & __PRINTF_FLAG_DECIMALPT);
        int dec_pt_len = has_decimal_pt ? 1 : 0;

        /* Fractional part lengths. */
        int signif_frac_figs = len - 1;
        int trailing_frac_zeros = precision - signif_frac_figs;

        if (flags & __PRINTF_FLAG_NOFRACZEROS) {
                trailing_frac_zeros = 0;
        }

        int frac_len = signif_frac_figs + trailing_frac_zeros;

        int exp_val = dec_exp + len - 1;
        /* Account for exponent sign and 'e'; minimum 2 digits. */
        int exp_len = 2 + (abs(exp_val) >= 100 ? 3 : 2);

        /* Number of non-padding chars to print. */
        int num_len = (sign ? 1 : 0) + 1 + dec_pt_len + frac_len + exp_len;

        int padding_len = max(0, width - num_len);

        if (!(flags & (__PRINTF_FLAG_LEFTALIGNED | __PRINTF_FLAG_ZEROPADDED)))
                TRY(_write_spaces(padding_len, ps, written_bytes));

        if (sign)
                TRY(_write_char(sign, ps, written_bytes));

        if (flags & __PRINTF_FLAG_ZEROPADDED)
                TRY(_write_zeros(padding_len, ps, written_bytes));

        /* Single leading integer. */
        TRY(_write_char(buf[0], ps, written_bytes));

        /* Print the decimal point and the fractional part. */
        if (has_decimal_pt) {
                TRY(_write_char('.', ps, written_bytes));

                /* Print significant digits of the fractional part of the number. */
                if (0 < signif_frac_figs)
                        TRY(_write_bytes(buf + 1, signif_frac_figs, ps, written_bytes));

                /* Print trailing zeros of the fractional part of the number. */
                TRY(_write_zeros(trailing_frac_zeros, ps, written_bytes));
        }

        /* Print the exponent. */
        TRY(_format_exponent(exp_val, flags, ps, written_bytes));

        if (flags & __PRINTF_FLAG_LEFTALIGNED)
                TRY(_write_spaces(padding_len, ps, written_bytes));

        return EOK;
}

/** Convert, format and print a double according to the %e specifier.
 *
 * Note that if g is large, the output may be huge (3e100 prints
 * with at least 100 digits).
 *
 * %e style: [-]d.dddde+dd
 *  left-justified:  [-]d.dddde+dd[space_pad]
 *  right-justified: [space_pad][-][zero_pad]d.dddde+dd
 *
 * @param g     Double to print.
 * @param precision Number of fractional digits to print, ie
 *              precision + 1 significant digits to display. If 0 no
 *              decimal point will be printed unless the flag
 *              __PRINTF_FLAG_DECIMALPT is specified. If negative
 *              the shortest accurate number will be printed.
 * @param width Minimum number of characters to display. Pads
 *              with '0' or ' ' depending on the set flags;
 * @param flags Printf flags.
 * @param ps    Printing functions.
 */
static errno_t _format_double_scientific(double g, int precision, int width,
    uint32_t flags, printf_spec_t *ps, size_t *written_bytes)
{
        if (flags & __PRINTF_FLAG_LEFTALIGNED)
                flags &= ~__PRINTF_FLAG_ZEROPADDED;

        ieee_double_t val = extract_ieee_double(g);

        if (val.is_special)
                return _format_special(val, width, flags, ps, written_bytes);

        char buf[MAX_DOUBLE_STR_BUF_SIZE];
        const size_t buf_size = MAX_DOUBLE_STR_BUF_SIZE;
        double_str_t val_str;

        val_str.str = buf;
        val_str.neg = val.is_negative;

        if (0 <= precision) {
                /*
                 * Request one more digit (in addition to the leading integer)
                 * so we can round the result. The last digit it returns may
                 * have an error of at most +/- 1.
                 */
                val_str.len = double_to_fixed_str(val, precision + 2, -1, buf, buf_size,
                    &val_str.dec_exp);

                /*
                 * Round the extra digit to produce precision+1 significant digits.
                 * If less than precision+2 significant digits were returned the last
                 * digit is definitely inaccurate so also round to get rid of it.
                 */
                _fp_round_up(buf, &val_str.len, &val_str.dec_exp);

                /* Rounding could have introduced trailing zeros. */
                if (flags & __PRINTF_FLAG_NOFRACZEROS) {
                        _fp_trim_trailing_zeros(buf, &val_str.len, &val_str.dec_exp);
                }
        } else {
                /* Let the implementation figure out the proper precision. */
                val_str.len = double_to_short_str(val, buf, buf_size, &val_str.dec_exp);

                /* Use all produced digits. */
                precision = val_str.len - 1;
        }

        return _format_double_str_scient(&val_str, precision, width, flags, ps, written_bytes);
}

/** Convert, format and print a double according to the %g specifier.
 *
 * %g style chooses between %f and %e.
 *
 * @param g     Double to print.
 * @param precision Number of significant digits to display; excluding
 *              any leading zeros from this count. If negative
 *              the shortest accurate number will be printed.
 * @param width Minimum number of characters to display. Pads
 *              with '0' or ' ' depending on the set flags;
 * @param flags Printf flags.
 * @param ps    Printing functions.
 *
 * @return The number of characters printed; negative on failure.
 */
static errno_t _format_double_generic(double g, int precision, int width,
    uint32_t flags, printf_spec_t *ps, size_t *written_bytes)
{
        ieee_double_t val = extract_ieee_double(g);

        if (val.is_special)
                return _format_special(val, width, flags, ps, written_bytes);

        char buf[MAX_DOUBLE_STR_BUF_SIZE];
        const size_t buf_size = MAX_DOUBLE_STR_BUF_SIZE;
        int dec_exp;
        int len;

        /* Honor the user requested number of significant digits. */
        if (0 <= precision) {
                /*
                 * Do a quick and dirty conversion of a single digit to determine
                 * the decimal exponent.
                 */
                len = double_to_fixed_str(val, 1, -1, buf, buf_size, &dec_exp);
                assert(0 < len);

                precision = max(1, precision);

                if (-4 <= dec_exp && dec_exp < precision) {
                        precision = precision - (dec_exp + 1);
                        return _format_double_fixed(g, precision, width,
                            flags | __PRINTF_FLAG_NOFRACZEROS, ps, written_bytes);
                } else {
                        --precision;
                        return _format_double_scientific(g, precision, width,
                            flags | __PRINTF_FLAG_NOFRACZEROS, ps, written_bytes);
                }
        } else {
                /* Convert to get the decimal exponent and digit count. */
                len = double_to_short_str(val, buf, buf_size, &dec_exp);
                assert(0 < len);

                if (flags & __PRINTF_FLAG_LEFTALIGNED) {
                        flags &= ~__PRINTF_FLAG_ZEROPADDED;
                }

                double_str_t val_str;
                val_str.str = buf;
                val_str.len = len;
                val_str.neg = val.is_negative;
                val_str.dec_exp = dec_exp;

                int first_digit_pos = len + dec_exp;
                int last_digit_pos = dec_exp;

                /* The whole number (15 digits max) fits between dec places 15 .. -6 */
                if (len <= 15 && -6 <= last_digit_pos && first_digit_pos <= 15) {
                        /* Precision needed for the last significant digit. */
                        precision = max(0, -val_str.dec_exp);
                        return _format_double_str_fixed(&val_str, precision, width, flags, ps, written_bytes);
                } else {
                        /* Use all produced digits. */
                        precision = val_str.len - 1;
                        return _format_double_str_scient(&val_str, precision, width, flags, ps, written_bytes);
                }
        }
}

/** Convert, format and print a double according to the specifier.
 *
 * Depending on the specifier it prints the double using the styles
 * %g, %f or %e by means of print_double_generic(), print_double_fixed(),
 * print_double_scientific().
 *
 * @param g     Double to print.
 * @param spec  Specifier of the style to print in; one of: 'g','G','f','F',
 *              'e','E'.
 * @param precision Number of fractional digits to display. If negative
 *              the shortest accurate number will be printed for style %g;
 *              negative precision defaults to 6 for styles %f, %e.
 * @param width Minimum number of characters to display. Pads
 *              with '0' or ' ' depending on the set flags;
 * @param flags Printf flags.
 * @param ps    Printing functions.
 */
static errno_t _format_double(double g, char spec, int precision, int width,
    uint32_t flags, printf_spec_t *ps, size_t *written_chars)
{
        switch (spec) {
        case 'F':
                flags |= __PRINTF_FLAG_BIGCHARS;
                /* Fallthrough */
        case 'f':
                precision = (precision < 0) ? 6 : precision;
                return _format_double_fixed(g, precision, width, flags, ps, written_chars);

        case 'E':
                flags |= __PRINTF_FLAG_BIGCHARS;
                /* Fallthrough */
        case 'e':
                precision = (precision < 0) ? 6 : precision;
                return _format_double_scientific(g, precision, width, flags, ps, written_chars);

        case 'G':
                flags |= __PRINTF_FLAG_BIGCHARS;
                /* Fallthrough */
        case 'g':
                return _format_double_generic(g, precision, width, flags, ps, written_chars);

        default:
                assert(false);
                return -1;
        }
}

#endif

static const char *_strchrnul(const char *s, int c)
{
        while (*s != c && *s != 0)
                s++;
        return s;
}

/** Read a sequence of digits from the format string as a number.
 * If the number has too many digits to fit in int, returns INT_MAX.
 */
static int _read_num(const char *fmt, size_t *i)
{
        const char *s;
        unsigned n = 0;

        for (s = &fmt[*i]; isdigit(*s); s++) {
                unsigned digit = (*s - '0');

                /* Check for overflow */
                if (n > INT_MAX / 10 || n * 10 > INT_MAX - digit) {
                        n = INT_MAX;
                        while (isdigit(*s))
                                s++;
                        break;
                }

                n = n * 10 + digit;
        }

        *i = s - fmt;
        return n;
}

static uint32_t _parse_flags(const char *fmt, size_t *i)
{
        uint32_t flags = 0;

        while (true) {
                switch (fmt[(*i)++]) {
                case '#':
                        flags |= __PRINTF_FLAG_PREFIX;
                        flags |= __PRINTF_FLAG_DECIMALPT;
                        continue;
                case '-':
                        flags |= __PRINTF_FLAG_LEFTALIGNED;
                        continue;
                case '+':
                        flags |= __PRINTF_FLAG_SHOWPLUS;
                        continue;
                case ' ':
                        flags |= __PRINTF_FLAG_SPACESIGN;
                        continue;
                case '0':
                        flags |= __PRINTF_FLAG_ZEROPADDED;
                        continue;
                }

                --*i;
                break;
        }

        return flags;
}

static bool _eat_char(const char *s, size_t *idx, int c)
{
        if (s[*idx] != c)
                return false;

        (*idx)++;
        return true;
}

static qualifier_t _read_qualifier(const char *s, size_t *idx)
{
        switch (s[(*idx)++]) {
        case 't': /* ptrdiff_t */
        case 'z': /* size_t */
                if (sizeof(ptrdiff_t) == sizeof(int))
                        return PrintfQualifierInt;
                else
                        return PrintfQualifierLong;

        case 'h':
                if (_eat_char(s, idx, 'h'))
                        return PrintfQualifierByte;
                else
                        return PrintfQualifierShort;

        case 'l':
                if (_eat_char(s, idx, 'l'))
                        return PrintfQualifierLongLong;
                else
                        return PrintfQualifierLong;

        case 'j':
                return PrintfQualifierLongLong;

        default:
                --*idx;

                /* Unspecified */
                return PrintfQualifierInt;
        }
}

/** Print formatted string.
 *
 * Print string formatted according to the fmt parameter and variadic arguments.
 * Each formatting directive must have the following form:
 *
 *  \% [ FLAGS ] [ WIDTH ] [ .PRECISION ] [ TYPE ] CONVERSION
 *
 * FLAGS:@n
 *  - "#" Force to print prefix. For \%o conversion, the prefix is 0, for
 *        \%x and \%X prefixes are 0x and 0X and for conversion \%b the
 *        prefix is 0b.
 *
 *  - "-" Align to left.
 *
 *  - "+" Print positive sign just as negative.
 *
 *  - " " If the printed number is positive and "+" flag is not set,
 *        print space in place of sign.
 *
 *  - "0" Print 0 as padding instead of spaces. Zeroes are placed between
 *        sign and the rest of the number. This flag is ignored if "-"
 *        flag is specified.
 *
 * WIDTH:@n
 *  - Specify the minimal width of a printed argument. If it is bigger,
 *    width is ignored. If width is specified with a "*" character instead of
 *    number, width is taken from parameter list. And integer parameter is
 *    expected before parameter for processed conversion specification. If
 *    this value is negative its absolute value is taken and the "-" flag is
 *    set.
 *
 * PRECISION:@n
 *  - Value precision. For numbers it specifies minimum valid numbers.
 *    Smaller numbers are printed with leading zeroes. Bigger numbers are not
 *    affected. Strings with more than precision characters are cut off. Just
 *    as with width, an "*" can be used used instead of a number. An integer
 *    value is then expected in parameters. When both width and precision are
 *    specified using "*", the first parameter is used for width and the
 *    second one for precision.
 *
 * TYPE:@n
 *  - "hh" Signed or unsigned char.@n
 *  - "h"  Signed or unsigned short.@n
 *  - ""   Signed or unsigned int (default value).@n
 *  - "l"  Signed or unsigned long int.@n
 *         If conversion is "c", the character is wint_t (wide character).@n
 *         If conversion is "s", the string is char32_t * (wide string).@n
 *  - "ll" Signed or unsigned long long int.@n
 *  - "z"  Signed or unsigned ssize_t or site_t.@n
 *
 * CONVERSION:@n
 *  - % Print percentile character itself.
 *
 *  - c Print single character. The character is expected to be plain
 *      ASCII (e.g. only values 0 .. 127 are valid).@n
 *      If type is "l", then the character is expected to be wide character
 *      (e.g. values 0 .. 0x10ffff are valid).
 *
 *  - s Print zero terminated string. If a NULL value is passed as
 *      value, "(NULL)" is printed instead.@n
 *      If type is "l", then the string is expected to be wide string.
 *
 *  - P, p Print value of a pointer. Void * value is expected and it is
 *         printed in hexadecimal notation with prefix (as with
 *         \%#0.8X / \%#0.8x for 32-bit or \%#0.16lX / \%#0.16lx for 64-bit
 *         long pointers).
 *
 *  - b Print value as unsigned binary number. Prefix is not printed by
 *      default. (Nonstandard extension.)
 *
 *  - o Print value as unsigned octal number. Prefix is not printed by
 *      default.
 *
 *  - d, i Print signed decimal number. There is no difference between d
 *         and i conversion.
 *
 *  - u Print unsigned decimal number.
 *
 *  - X, x Print hexadecimal number with upper- or lower-case. Prefix is
 *         not printed by default.
 *
 * All other characters from fmt except the formatting directives are printed
 * verbatim.
 *
 * @param fmt Format NULL-terminated string.
 *
 * @return Number of characters printed, negative value on failure.
 *
 */
int printf_core(const char *fmt, printf_spec_t *ps, va_list ap)
{
        errno_t rc = EOK;
        size_t nxt = 0;  /* Index of the next character from fmt */

        size_t counter = 0;   /* Number of characters printed */

        while (rc == EOK) {
                /* Find the next specifier and write all the bytes before it. */
                const char *s = _strchrnul(&fmt[nxt], '%');
                size_t bytes = s - &fmt[nxt];
                rc = _write_bytes(&fmt[nxt], bytes, ps, &counter);
                if (rc != EOK)
                        break;

                nxt += bytes;

                /* Check for end of string. */
                if (_eat_char(fmt, &nxt, 0))
                        break;

                /* We must be at the start of a specifier. */
                bool spec = _eat_char(fmt, &nxt, '%');
                assert(spec);

                /* Parse modifiers */
                uint32_t flags = _parse_flags(fmt, &nxt);

                /* Width & '*' operator */
                int width = -1;
                if (_eat_char(fmt, &nxt, '*')) {
                        /* Get width value from argument list */
                        width = va_arg(ap, int);

                        if (width < 0) {
                                /* Negative width sets '-' flag */
                                width = (width == INT_MIN) ? INT_MAX : -width;
                                flags |= __PRINTF_FLAG_LEFTALIGNED;
                        }
                } else {
                        width = _read_num(fmt, &nxt);
                }

                /* Precision and '*' operator */
                int precision = -1;
                if (_eat_char(fmt, &nxt, '.')) {
                        if (_eat_char(fmt, &nxt, '*')) {
                                /* Get precision value from the argument list */
                                precision = va_arg(ap, int);

                                /* Negative is treated as omitted. */
                                if (precision < 0)
                                        precision = -1;
                        } else {
                                precision = _read_num(fmt, &nxt);
                        }
                }

                qualifier_t qualifier = _read_qualifier(fmt, &nxt);
                unsigned int base = 10;
                char specifier = fmt[nxt++];

                switch (specifier) {
                /*
                 * String and character conversions.
                 */
                case 's':
                        if (qualifier == PrintfQualifierLong)
                                rc = _format_wstr(va_arg(ap, char32_t *), width, precision, flags, ps, &counter);
                        else
                                rc = _format_cstr(va_arg(ap, char *), width, precision, flags, ps, &counter);
                        continue;

                case 'c':
                        if (qualifier == PrintfQualifierLong)
                                rc = _format_uchar(va_arg(ap, wint_t), width, flags, ps, &counter);
                        else
                                rc = _format_char(va_arg(ap, int), width, flags, ps, &counter);
                        continue;

                /*
                 * Floating point values
                 */
                case 'G':
                case 'g':
                case 'F':
                case 'f':
                case 'E':
                case 'e':;
#ifdef HAS_FLOAT
                        rc = _format_double(va_arg(ap, double), specifier, precision,
                            width, flags, ps, &counter);
#else
                        rc = _format_cstr("<float unsupported>", width, -1, 0, ps, &counter);
#endif
                        continue;

                /*
                 * Integer values
                 */
                case 'P':
                        /* Pointer */
                        flags |= __PRINTF_FLAG_BIGCHARS;
                        /* Fallthrough */
                case 'p':
                        flags |= __PRINTF_FLAG_PREFIX;
                        flags |= __PRINTF_FLAG_ZEROPADDED;
                        base = 16;
                        qualifier = PrintfQualifierPointer;
                        break;
                case 'b':
                        base = 2;
                        break;
                case 'o':
                        base = 8;
                        break;
                case 'd':
                case 'i':
                        flags |= __PRINTF_FLAG_SIGNED;
                        break;
                case 'u':
                        break;
                case 'X':
                        flags |= __PRINTF_FLAG_BIGCHARS;
                        /* Fallthrough */
                case 'x':
                        base = 16;
                        break;

                case '%':
                        /* Percentile itself */
                        rc = _write_char('%', ps, &counter);
                        continue;

                /*
                 * Bad formatting.
                 */
                default:
                        rc = EINVAL;
                        continue;
                }

                /* Print integers */
                uint64_t number;

                switch (qualifier) {
                case PrintfQualifierByte:
                        number = PRINTF_GET_INT_ARGUMENT(int, ap, flags);
                        break;
                case PrintfQualifierShort:
                        number = PRINTF_GET_INT_ARGUMENT(int, ap, flags);
                        break;
                case PrintfQualifierInt:
                        number = PRINTF_GET_INT_ARGUMENT(int, ap, flags);
                        break;
                case PrintfQualifierLong:
                        number = PRINTF_GET_INT_ARGUMENT(long, ap, flags);
                        break;
                case PrintfQualifierLongLong:
                        number = PRINTF_GET_INT_ARGUMENT(long long, ap, flags);
                        break;
                case PrintfQualifierPointer:
                        precision = sizeof(void *) << 1;
                        number = (uint64_t) (uintptr_t) va_arg(ap, void *);
                        break;
                default:
                        /* Unknown qualifier */
                        rc = EINVAL;
                        continue;
                }

                rc = _format_number(number, width, precision, base, flags, ps, &counter);
        }

        if (rc != EOK) {
                _set_errno(rc);
                return -1;
        }

        if (counter > INT_MAX) {
                _set_errno(EOVERFLOW);
                return -1;
        }

        return (int) counter;
}

/** @}
 */