utf16.c File Reference

#include <stdbool.h>
#include <stdlib.h>
#include <string.h>
#include <parserutils/charset/utf16.h>

Include dependency graph for utf16.c:

Functions
parserutils_error	parserutils_charset_utf16_to_ucs4 (const uint8_t *s, size_t len, uint32_t *ucs4, size_t *clen)
parserutils_error	parserutils_charset_utf16_from_ucs4 (uint32_t ucs4, uint8_t *s, size_t *len)
parserutils_error	parserutils_charset_utf16_length (const uint8_t *s, size_t max, size_t *len)
parserutils_error	parserutils_charset_utf16_char_byte_length (const uint8_t *s, size_t *len)
parserutils_error	parserutils_charset_utf16_prev (const uint8_t *s, uint32_t off, uint32_t *prevoff)
parserutils_error	parserutils_charset_utf16_next (const uint8_t *s, uint32_t len, uint32_t off, uint32_t *nextoff)
parserutils_error	parserutils_charset_utf16_next_paranoid (const uint8_t *s, uint32_t len, uint32_t off, uint32_t *nextoff)

Detailed Description

UTF-16 manipulation functions (implementation).

Function Documentation

parserutils_charset_utf16_char_byte_length()

parserutils_error parserutils_charset_utf16_char_byte_length	(	const uint8_t *	s,
		size_t *	len
	)

Calculate the length (in bytes) of a UTF-16 character

Parameters

s	Pointer to start of character
len	Pointer to location to receive length

Returns

PARSERUTILS_OK on success, appropriate error otherwise

parserutils_charset_utf16_from_ucs4()

parserutils_error parserutils_charset_utf16_from_ucs4	(	uint32_t	ucs4,
		uint8_t *	s,
		size_t *	len
	)

Convert a single UCS-4 character into a UTF-16 sequence

Parameters

ucs4	The character to process (0 <= c <= 0x7FFFFFFF) (host endian)
s	Pointer to 4 byte long output buffer
len	Pointer to location to receive length of multibyte sequence

Returns

PARSERUTILS_OK on success, appropriate error otherwise

parserutils_charset_utf16_length()

parserutils_error parserutils_charset_utf16_length	(	const uint8_t *	s,
		size_t	max,
		size_t *	len
	)

Calculate the length (in characters) of a bounded UTF-16 string

Parameters

s	The string
max	Maximum length
len	Pointer to location to receive length of string

Returns

PARSERUTILS_OK on success, appropriate error otherwise

parserutils_charset_utf16_next()

parserutils_error parserutils_charset_utf16_next	(	const uint8_t *	s,
		uint32_t	len,
		uint32_t	off,
		uint32_t *	nextoff
	)

Find next legal UTF-16 char in string

Parameters

s	The string (assumed valid)
len	Maximum offset in string
off	Offset in the string to start at
nextoff	Pointer to location to receive offset of first byte of next legal character

Returns

PARSERUTILS_OK on success, appropriate error otherwise

parserutils_charset_utf16_next_paranoid()

parserutils_error parserutils_charset_utf16_next_paranoid	(	const uint8_t *	s,
		uint32_t	len,
		uint32_t	off,
		uint32_t *	nextoff
	)

Find next legal UTF-16 char in string

Parameters

s	The string (assumed to be of dubious validity)
len	Maximum offset in string
off	Offset in the string to start at
nextoff	Pointer to location to receive offset of first byte of next legal character

Returns

PARSERUTILS_OK on success, appropriate error otherwise

parserutils_charset_utf16_prev()

parserutils_error parserutils_charset_utf16_prev	(	const uint8_t *	s,
		uint32_t	off,
		uint32_t *	prevoff
	)

Find previous legal UTF-16 char in string

Parameters

s	The string
off	Offset in the string to start at
prevoff	Pointer to location to receive offset of first byte of previous legal character

Returns

PARSERUTILS_OK on success, appropriate error otherwise

parserutils_charset_utf16_to_ucs4()

parserutils_error parserutils_charset_utf16_to_ucs4	(	const uint8_t *	s,
		size_t	len,
		uint32_t *	ucs4,
		size_t *	clen
	)

Convert a UTF-16 sequence into a single UCS-4 character

Parameters

s	The sequence to process
len	Length of sequence in bytes
ucs4	Pointer to location to receive UCS-4 character (host endian)
clen	Pointer to location to receive byte length of UTF-16 sequence

Returns

PARSERUTILS_OK on success, appropriate error otherwise