Function collect_a_seq

freebsd/contrib/openzfs/module/unicode/u8_textprep.c:1375–1710

* The collect_a_seq() function checks on the given string s, collect * a sequence of characters at u8s, and return the sequence. While it collects * a sequence, it also applies case conversion, canonical or compatibility * decomposition, canonical decomposition, or some or all of them and * in that order. * * The collected sequence cannot be bigger than 32 characters since if * it is having

 * null byte.
 */
static size_t
collect_a_seq(size_t uv, uchar_t *u8s, uchar_t **source, uchar_t *slast,
    boolean_t is_it_toupper, boolean_t is_it_tolower,
    boolean_t canonical_decomposition, boolean_t compatibility_decomposition,
    boolean_t canonical_composition,
    int *errnum, u8_normalization_states_t *state)
{
	uchar_t *s;
	int sz;
	int saved_sz;
	size_t i;
	size_t j;
	size_t k;
	size_t l;
	uchar_t comb_class[U8_MAX_CHARS_A_SEQ];
	uchar_t disp[U8_MAX_CHARS_A_SEQ];
	uchar_t start[U8_MAX_CHARS_A_SEQ];
	uchar_t u8t[U8_MB_CUR_MAX] = { '\0' };
	uchar_t uts[U8_STREAM_SAFE_TEXT_MAX + 1];
	uchar_t tc;
	size_t last;
	size_t saved_last;
	uint32_t u1;

	/*
	 * Save the source string pointer which we will return a changed
	 * pointer if we do processing.
	 */
	s = *source;

	/*
	 * The following is a fallback for just in case callers are not
	 * checking the string boundaries before the calling.
	 */
	if (s >= slast) {
		u8s[0] = '\0';

		return (0);
	}

	/*
	 * As the first thing, let's collect a character and do case
	 * conversion if necessary.
	 */

	sz = u8_number_of_bytes[*s];

	if (sz < 0) {
		*errnum = EILSEQ;

		u8s[0] = *s++;
		u8s[1] = '\0';

		*source = s;

		return (1);
	}