utf8::internal Namespace Reference

Enumerations
enum	utf_error {   UTF8_OK , NOT_ENOUGH_ROOM , INVALID_LEAD , INCOMPLETE_SEQUENCE ,   OVERLONG_SEQUENCE , INVALID_CODE_POINT }

Functions
template<typename octet_type>
utfchar8_t	mask8 (octet_type oc)
template<typename u16_type>
utfchar16_t	mask16 (u16_type oc)
template<typename octet_type>
bool	is_trail (octet_type oc)
bool	is_lead_surrogate (utfchar32_t cp)
bool	is_trail_surrogate (utfchar32_t cp)
bool	is_surrogate (utfchar32_t cp)
bool	is_code_point_valid (utfchar32_t cp)
bool	is_in_bmp (utfchar32_t cp)
template<typename octet_iterator>
int	sequence_length (octet_iterator lead_it)
bool	is_overlong_sequence (utfchar32_t cp, int length)
template<typename octet_iterator>
utf_error	increase_safely (octet_iterator &it, const octet_iterator end)
	Helper for get_sequence_x.
template<typename octet_iterator>
utf_error	get_sequence_1 (octet_iterator &it, octet_iterator end, utfchar32_t &code_point)
	get_sequence_x functions decode utf-8 sequences of the length x
template<typename octet_iterator>
utf_error	get_sequence_2 (octet_iterator &it, octet_iterator end, utfchar32_t &code_point)
template<typename octet_iterator>
utf_error	get_sequence_3 (octet_iterator &it, octet_iterator end, utfchar32_t &code_point)
template<typename octet_iterator>
utf_error	get_sequence_4 (octet_iterator &it, octet_iterator end, utfchar32_t &code_point)
template<typename octet_iterator>
utf_error	validate_next (octet_iterator &it, octet_iterator end, utfchar32_t &code_point)
template<typename octet_iterator>
utf_error	validate_next (octet_iterator &it, octet_iterator end)
template<typename word_iterator>
utf_error	validate_next16 (word_iterator &it, word_iterator end, utfchar32_t &code_point)
template<typename octet_iterator, typename octet_type>
octet_iterator	append (utfchar32_t cp, octet_iterator result)
char *	append (utfchar32_t cp, char *result)
template<typename container_type>
std::back_insert_iterator< container_type >	append (utfchar32_t cp, std::back_insert_iterator< container_type > result)
template<typename octet_iterator>
octet_iterator	append (utfchar32_t cp, octet_iterator result)
template<typename word_iterator, typename word_type>
word_iterator	append16 (utfchar32_t cp, word_iterator result)
template<typename container_type>
std::back_insert_iterator< container_type >	append16 (utfchar32_t cp, std::back_insert_iterator< container_type > result)
template<typename word_iterator>
word_iterator	append16 (utfchar32_t cp, word_iterator result)

Variables
const utfchar16_t	LEAD_SURROGATE_MIN = 0xd800u
const utfchar16_t	LEAD_SURROGATE_MAX = 0xdbffu
const utfchar16_t	TRAIL_SURROGATE_MIN = 0xdc00u
const utfchar16_t	TRAIL_SURROGATE_MAX = 0xdfffu
const utfchar16_t	LEAD_OFFSET = 0xd7c0u
const utfchar32_t	SURROGATE_OFFSET = 0xfca02400u
const utfchar32_t	CODE_POINT_MAX = 0x0010ffffu

Enumeration Type Documentation

utf_error

enum utf8::internal::utf_error

Enumerator
UTF8_OK
NOT_ENOUGH_ROOM
INVALID_LEAD
INCOMPLETE_SEQUENCE
OVERLONG_SEQUENCE
INVALID_CODE_POINT

Function Documentation

append() [1/4]

char * utf8::internal::append ( utfchar32_t cp, char * result )

inline

References append().

append() [2/4]

template<typename octet_iterator>

octet_iterator utf8::internal::append	(	utfchar32_t	cp,
		octet_iterator	result )

References append().

append() [3/4]

template<typename octet_iterator, typename octet_type>

octet_iterator utf8::internal::append	(	utfchar32_t	cp,
		octet_iterator	result )

Referenced by utf8::append(), append(), append(), append(), and utf8::unchecked::append().

append() [4/4]

template<typename container_type>

std::back_insert_iterator< container_type > utf8::internal::append	(	utfchar32_t	cp,
		std::back_insert_iterator< container_type >	result )

References append().

append16() [1/3]

template<typename container_type>

std::back_insert_iterator< container_type > utf8::internal::append16	(	utfchar32_t	cp,
		std::back_insert_iterator< container_type >	result )

References append16().

append16() [2/3]

template<typename word_iterator>

word_iterator utf8::internal::append16	(	utfchar32_t	cp,
		word_iterator	result )

References append16().

append16() [3/3]

template<typename word_iterator, typename word_type>

word_iterator utf8::internal::append16	(	utfchar32_t	cp,
		word_iterator	result )

References is_in_bmp(), LEAD_OFFSET, TRAIL_SURROGATE_MIN, and UTF_CPP_STATIC_ASSERT.

Referenced by utf8::append16(), append16(), append16(), and utf8::unchecked::append16().

get_sequence_1()

template<typename octet_iterator>

utf_error utf8::internal::get_sequence_1	(	octet_iterator &	it,
		octet_iterator	end,
		utfchar32_t &	code_point )

get_sequence_x functions decode utf-8 sequences of the length x

References mask8(), NOT_ENOUGH_ROOM, and UTF8_OK.

Referenced by validate_next().

get_sequence_2()

template<typename octet_iterator>

utf_error utf8::internal::get_sequence_2	(	octet_iterator &	it,
		octet_iterator	end,
		utfchar32_t &	code_point )

References mask8(), NOT_ENOUGH_ROOM, UTF8_CPP_INCREASE_AND_RETURN_ON_ERROR, and UTF8_OK.

Referenced by validate_next().

get_sequence_3()

template<typename octet_iterator>

utf_error utf8::internal::get_sequence_3	(	octet_iterator &	it,
		octet_iterator	end,
		utfchar32_t &	code_point )

References mask8(), NOT_ENOUGH_ROOM, UTF8_CPP_INCREASE_AND_RETURN_ON_ERROR, and UTF8_OK.

Referenced by validate_next().

get_sequence_4()

template<typename octet_iterator>

utf_error utf8::internal::get_sequence_4	(	octet_iterator &	it,
		octet_iterator	end,
		utfchar32_t &	code_point )

References mask8(), NOT_ENOUGH_ROOM, UTF8_CPP_INCREASE_AND_RETURN_ON_ERROR, and UTF8_OK.

Referenced by validate_next().

increase_safely()

template<typename octet_iterator>

utf_error utf8::internal::increase_safely	(	octet_iterator &	it,
		const octet_iterator	end )

Helper for get_sequence_x.

References INCOMPLETE_SEQUENCE, is_trail(), NOT_ENOUGH_ROOM, and UTF8_OK.

is_code_point_valid()

bool utf8::internal::is_code_point_valid ( utfchar32_t cp )

inline

References CODE_POINT_MAX, and is_surrogate().

Referenced by utf8::append(), utf8::append16(), and validate_next().

is_in_bmp()

bool utf8::internal::is_in_bmp ( utfchar32_t cp )

inline

Referenced by append16().

is_lead_surrogate()

bool utf8::internal::is_lead_surrogate ( utfchar32_t cp )

inline

References LEAD_SURROGATE_MAX, and LEAD_SURROGATE_MIN.

Referenced by utf8::unchecked::next16(), utf8::unchecked::utf16to8(), utf8::utf16to8(), and validate_next16().

is_overlong_sequence()

bool utf8::internal::is_overlong_sequence ( utfchar32_t cp, int length )

inline

Referenced by validate_next().

is_surrogate()

bool utf8::internal::is_surrogate ( utfchar32_t cp )

inline

References LEAD_SURROGATE_MIN, and TRAIL_SURROGATE_MAX.

Referenced by is_code_point_valid(), and validate_next16().

is_trail()

template<typename octet_type>

bool utf8::internal::is_trail ( octet_type oc )

inline

References mask8().

Referenced by increase_safely(), utf8::prior(), utf8::unchecked::prior(), utf8::replace_invalid(), and utf8::unchecked::replace_invalid().

is_trail_surrogate()

bool utf8::internal::is_trail_surrogate ( utfchar32_t cp )

inline

References TRAIL_SURROGATE_MAX, and TRAIL_SURROGATE_MIN.

Referenced by utf8::utf16to8(), and validate_next16().

mask16()

template<typename u16_type>

utfchar16_t utf8::internal::mask16 ( u16_type oc )

inline

Referenced by utf8::unchecked::next16(), utf8::replace_invalid(), utf8::unchecked::replace_invalid(), utf8::unchecked::utf16to8(), and utf8::utf16to8().

mask8()

template<typename octet_type>

utfchar8_t utf8::internal::mask8 ( octet_type oc )

inline

Referenced by get_sequence_1(), get_sequence_2(), get_sequence_3(), get_sequence_4(), is_trail(), utf8::unchecked::next(), sequence_length(), and utf8::starts_with_bom().

sequence_length()

template<typename octet_iterator>

int utf8::internal::sequence_length

(

octet_iterator

lead_it

)

References mask8().

Referenced by utf8::unchecked::next(), utf8::unchecked::iterator< octet_iterator >::operator++(), utf8::unchecked::iterator< octet_iterator >::operator++(), and validate_next().

validate_next() [1/2]

template<typename octet_iterator>

utf_error utf8::internal::validate_next ( octet_iterator & it, octet_iterator end )

inline

References validate_next().

validate_next() [2/2]

template<typename octet_iterator>

utf_error utf8::internal::validate_next	(	octet_iterator &	it,
		octet_iterator	end,
		utfchar32_t &	code_point )

References get_sequence_1(), get_sequence_2(), get_sequence_3(), get_sequence_4(), INVALID_CODE_POINT, INVALID_LEAD, is_code_point_valid(), is_overlong_sequence(), NOT_ENOUGH_ROOM, OVERLONG_SEQUENCE, sequence_length(), and UTF8_OK.

Referenced by utf8::find_invalid(), utf8::next(), utf8::replace_invalid(), utf8::unchecked::replace_invalid(), and validate_next().

validate_next16()

template<typename word_iterator>

utf_error utf8::internal::validate_next16	(	word_iterator &	it,
		word_iterator	end,
		utfchar32_t &	code_point )

References INCOMPLETE_SEQUENCE, INVALID_LEAD, is_lead_surrogate(), is_surrogate(), is_trail_surrogate(), NOT_ENOUGH_ROOM, SURROGATE_OFFSET, UTF8_OK, and UTF_CPP_STATIC_ASSERT.

Referenced by utf8::next16().

Variable Documentation

CODE_POINT_MAX

const utfchar32_t utf8::internal::CODE_POINT_MAX = 0x0010ffffu

Referenced by is_code_point_valid().

LEAD_OFFSET

const utfchar16_t utf8::internal::LEAD_OFFSET = 0xd7c0u

Referenced by append16(), utf8::unchecked::utf8to16(), and utf8::utf8to16().

LEAD_SURROGATE_MAX

const utfchar16_t utf8::internal::LEAD_SURROGATE_MAX = 0xdbffu

Referenced by is_lead_surrogate().

LEAD_SURROGATE_MIN

const utfchar16_t utf8::internal::LEAD_SURROGATE_MIN = 0xd800u

Referenced by is_lead_surrogate(), and is_surrogate().

SURROGATE_OFFSET

const utfchar32_t utf8::internal::SURROGATE_OFFSET = 0xfca02400u

Referenced by utf8::unchecked::next16(), utf8::unchecked::utf16to8(), utf8::utf16to8(), and validate_next16().

TRAIL_SURROGATE_MAX

const utfchar16_t utf8::internal::TRAIL_SURROGATE_MAX = 0xdfffu

Referenced by is_surrogate(), and is_trail_surrogate().

TRAIL_SURROGATE_MIN

const utfchar16_t utf8::internal::TRAIL_SURROGATE_MIN = 0xdc00u

Referenced by append16(), is_trail_surrogate(), utf8::unchecked::utf8to16(), and utf8::utf8to16().