utf8helper Namespace Reference

Classes
struct	character
struct	Parser
	Parse and transform an UTF8 string. More...
class	UTF8String

Typedefs
using	byte = uint8_t
using	codepoint = uint32_t
using	Transform = codepoint(*)(const character *, int context)
	functor implements desired transformation of the character It has 2 arguments:

Functions
static Parser::Exit	_p0 (Parser *p, byte bb)
static Parser::Exit	_p1_u2 (Parser *p, byte bb)
	1 byte: RFC 3629:#4: Valid UTF-8 matches the following syntax 00-7F
static Parser::Exit	_p1_u3 (Parser *p, byte bb)
static Parser::Exit	_p1_u4 (Parser *p, byte bb)
static Parser::Exit	_p2_u3 (Parser *p, byte bb)
static Parser::Exit	_p2_u4 (Parser *p, byte bb)
static Parser::Exit	_p3_u4 (Parser *p, byte bb)
static int	_u_size (codepoint u)
static char *	_u_string (char *buf, codepoint u)
codepoint	TransformCapitalize (const character *, int)
codepoint	TransformLower (const character *, int)
codepoint	TransformNop (const character *, int)
codepoint	TransformNormalize (const character *, int)
codepoint	TransformTransliterate (const character *, int)
codepoint	TransformUpper (const character *, int)
std::string	UTF8Capitalize (const std::string &text)
std::string	UTF8Normalize (const std::string &text)
std::string	UTF8ToLower (const std::string &text)
std::string	UTF8ToUpper (const std::string &text)
std::string	UTF8Transliterate (const std::string &text)

Variables
const character	charmap_c2 []
const character	charmap_c3 []
const character	charmap_c4 []
const character	charmap_c5 []
const character	charmap_c6 []
const character	charmap_c7 []
const character	charmap_c8 []
const character	charmap_c9 []
const character	charmap_ca []
const character	charmap_cb []
const character	charmap_cc []
const character	charmap_cd []
const character	charmap_ce []
const character	charmap_cf []
const character	charmap_d0 []
const character	charmap_d1 []
const character	charmap_d2 []
const character	charmap_d3 []
const character	charmap_d4 []
const character	charmap_d5 []
const character	charmap_d6 []
const character	charmap_e1_82 []
const character	charmap_e1_83 []
const character	charmap_e1_b8 []
const character	charmap_e1_b9 []
const character	charmap_e1_ba []
const character	charmap_e1_bb []
const character	charmap_e1_bc []
const character	charmap_e1_bd []
const character	charmap_e1_be []
const character	charmap_e1_bf []
const character	charmap_e2_80 []
const character	charmap_e2_81 []
const character	charmap_e2_82 []
const character	charmap_e2_b4 []
const character	charmap_f0_90_92 []
const character	charmap_f0_90_93 []
const character	charmap_f0_9e_a4 []
const character	charmap_us7ascii []
constexpr int	IsBreaker = 0x02
constexpr int	IsControl = 0x04
constexpr int	IsDiacritic = 0x10
constexpr int	IsModifier = 0x08
constexpr int	IsPunctuation = 0x20
constexpr int	IsSpace = 0x01
constexpr int	None = 0x00
constexpr codepoint	NullCodepoint = 0
const character *	pagemap_16 [32]
const character *	pagemap_24_e1 []
const character *	pagemap_24_e2 []
const character *	pagemap_32_f0_90 []
const character *	pagemap_32_f0_9e []

Typedef Documentation

byte

using utf8helper::byte = typedef uint8_t

codepoint

using utf8helper::codepoint = typedef uint32_t

Transform

using utf8helper::Transform = typedef codepoint (*)(const character*, int context)

functor implements desired transformation of the character It has 2 arguments:

• The character struct matching the current code point
• The context, it is the category of the preceding sequence It must return the new code point to be dumped instead, or NullCodepoint to discard the sequence.

Function Documentation

_p0()

static Parser::Exit utf8helper::_p0 ( Parser *  p, byte  bb  )

static

References _p1_u2(), _p1_u3(), _p1_u4(), _u_size(), utf8helper::Parser::b, utf8helper::character::category, charmap_us7ascii, utf8helper::Parser::context, utf8helper::Parser::Continue, utf8helper::Parser::Done, utf8helper::Parser::Error, utf8helper::Parser::func, NullCodepoint, utf8helper::Parser::run, utf8helper::Parser::u, and utf8helper::Parser::u_size.

Referenced by _p1_u2(), _p1_u3(), _p1_u4(), _p2_u3(), _p2_u4(), _p3_u4(), and utf8helper::Parser::Reset().

_p1_u2()

static Parser::Exit utf8helper::_p1_u2 ( Parser *  p, byte  bb  )

static

1 byte: RFC 3629:#4: Valid UTF-8 matches the following syntax 00-7F

2 bytes: RFC 3629:#4: Valid UTF-8 matches the following syntax C2-DF 80-BF

3 bytes: RFC 3629:#4: Valid UTF-8 matches the following syntax E0 A0-BF 80-BF E1-EC 80-BF 80-BF ED 80-9F 80-BF EE-EF 80-BF 80-BF RFC 3629:#6: [EF,BB,BF] is BOM on start, else ZERO WIDTH NO-BREAK SPACE

4 bytes: RFC 3629:#4: Valid UTF-8 matches the following syntax F0 90-BF 80-BF 80-BF F1-F3 80-BF 80-BF 80-BF F4 80-8F 80-BF 80-BF

References _p0(), _u_size(), utf8helper::Parser::b, utf8helper::character::category, utf8helper::Parser::context, utf8helper::Parser::Continue, utf8helper::Parser::Done, utf8helper::Parser::func, None, NullCodepoint, pagemap_16, utf8helper::Parser::run, utf8helper::Parser::u, and utf8helper::Parser::u_size.

Referenced by _p0().

_p1_u3()

static Parser::Exit utf8helper::_p1_u3 ( Parser *  p, byte  bb  )

static

References _p0(), _p2_u3(), utf8helper::Parser::b, utf8helper::Parser::Continue, and utf8helper::Parser::run.

Referenced by _p0().

_p1_u4()

static Parser::Exit utf8helper::_p1_u4 ( Parser *  p, byte  bb  )

static

References _p0(), _p2_u4(), utf8helper::Parser::b, utf8helper::Parser::Continue, and utf8helper::Parser::run.

Referenced by _p0().

_p2_u3()

static Parser::Exit utf8helper::_p2_u3 ( Parser *  p, byte  bb  )

static

References _p0(), _u_size(), utf8helper::Parser::b, utf8helper::character::category, utf8helper::Parser::context, utf8helper::Parser::Continue, utf8helper::Parser::Done, utf8helper::Parser::func, None, NullCodepoint, pagemap_24_e1, pagemap_24_e2, utf8helper::Parser::run, utf8helper::Parser::u, and utf8helper::Parser::u_size.

Referenced by _p1_u3().

_p2_u4()

static Parser::Exit utf8helper::_p2_u4 ( Parser *  p, byte  bb  )

static

References _p0(), _p3_u4(), utf8helper::Parser::b, utf8helper::Parser::Continue, and utf8helper::Parser::run.

Referenced by _p1_u4().

_p3_u4()

static Parser::Exit utf8helper::_p3_u4 ( Parser *  p, byte  bb  )

static

References _p0(), _u_size(), utf8helper::Parser::b, utf8helper::character::category, utf8helper::Parser::context, utf8helper::Parser::Continue, utf8helper::Parser::Done, utf8helper::Parser::func, None, NullCodepoint, pagemap_32_f0_90, pagemap_32_f0_9e, utf8helper::Parser::run, utf8helper::Parser::u, and utf8helper::Parser::u_size.

Referenced by _p2_u4().

_u_size()

static int utf8helper::_u_size ( codepoint  u )

inlinestatic

Referenced by _p0(), _p1_u2(), _p2_u3(), _p3_u4(), and utf8helper::UTF8String::Remove().

_u_string()

static char * utf8helper::_u_string ( char *  buf, codepoint  u  )

inlinestatic

Referenced by utf8helper::UTF8String::Append(), utf8helper::UTF8String::Insert(), utf8helper::UTF8String::Substr(), utf8helper::UTF8String::ToStdString(), and utf8helper::UTF8String::Transform().

TransformCapitalize()

codepoint utf8helper::TransformCapitalize ( const character *  ch, int  context  )

extern

References IsBreaker, IsControl, IsSpace, utf8helper::character::lower, and utf8helper::character::upper.

Referenced by UTF8Capitalize().

TransformLower()

codepoint utf8helper::TransformLower ( const character *  ch, int  context  )

extern

References utf8helper::character::lower.

Referenced by UTF8ToLower().

TransformNop()

codepoint utf8helper::TransformNop ( const character *  ch, int  context  )

extern

References utf8helper::character::code.

TransformNormalize()

codepoint utf8helper::TransformNormalize ( const character *  ch, int  context  )

extern

References utf8helper::character::category, IsBreaker, IsControl, IsSpace, utf8helper::character::lower, and NullCodepoint.

Referenced by UTF8Normalize().

TransformTransliterate()

codepoint utf8helper::TransformTransliterate ( const character *  ch, int  context  )

extern

References utf8helper::character::category, IsSpace, NullCodepoint, and utf8helper::character::translate.

Referenced by UTF8Transliterate().

TransformUpper()

codepoint utf8helper::TransformUpper ( const character *  ch, int  context  )

extern

References utf8helper::character::upper.

Referenced by UTF8ToUpper().

UTF8Capitalize()

std::string utf8helper::UTF8Capitalize ( const std::string &  text )

extern

References utf8helper::UTF8String::ToStdString(), and TransformCapitalize().

UTF8Normalize()

std::string utf8helper::UTF8Normalize ( const std::string &  text )

extern

References utf8helper::UTF8String::ToStdString(), and TransformNormalize().

Referenced by osmscout::UTF8NormForLookup().

UTF8ToLower()

std::string utf8helper::UTF8ToLower ( const std::string &  text )

extern

References utf8helper::UTF8String::ToStdString(), and TransformLower().

Referenced by osmscout::UTF8StringToLower().

UTF8ToUpper()

std::string utf8helper::UTF8ToUpper ( const std::string &  text )

extern

References utf8helper::UTF8String::ToStdString(), and TransformUpper().

Referenced by osmscout::UTF8StringToUpper().

UTF8Transliterate()

std::string utf8helper::UTF8Transliterate ( const std::string &  text )

extern

References utf8helper::UTF8String::ToStdString(), and TransformTransliterate().

Referenced by osmscout::UTF8Transliterate().

Variable Documentation

charmap_c2

const character utf8helper::charmap_c2

charmap_c3

const character utf8helper::charmap_c3

charmap_c4

const character utf8helper::charmap_c4

charmap_c5

const character utf8helper::charmap_c5

charmap_c6

const character utf8helper::charmap_c6

charmap_c7

const character utf8helper::charmap_c7

charmap_c8

const character utf8helper::charmap_c8

charmap_c9

const character utf8helper::charmap_c9

charmap_ca

const character utf8helper::charmap_ca

charmap_cb

const character utf8helper::charmap_cb

charmap_cc

const character utf8helper::charmap_cc

charmap_cd

const character utf8helper::charmap_cd

charmap_ce

const character utf8helper::charmap_ce

charmap_cf

const character utf8helper::charmap_cf

charmap_d0

const character utf8helper::charmap_d0

charmap_d1

const character utf8helper::charmap_d1

charmap_d2

const character utf8helper::charmap_d2

charmap_d3

const character utf8helper::charmap_d3

charmap_d4

const character utf8helper::charmap_d4

charmap_d5

const character utf8helper::charmap_d5

charmap_d6

const character utf8helper::charmap_d6

charmap_e1_82

const character utf8helper::charmap_e1_82

charmap_e1_83

const character utf8helper::charmap_e1_83

charmap_e1_b8

const character utf8helper::charmap_e1_b8

charmap_e1_b9

const character utf8helper::charmap_e1_b9

charmap_e1_ba

const character utf8helper::charmap_e1_ba

charmap_e1_bb

const character utf8helper::charmap_e1_bb

charmap_e1_bc

const character utf8helper::charmap_e1_bc

charmap_e1_bd

const character utf8helper::charmap_e1_bd

charmap_e1_be

const character utf8helper::charmap_e1_be

charmap_e1_bf

const character utf8helper::charmap_e1_bf

charmap_e2_80

const character utf8helper::charmap_e2_80

charmap_e2_81

const character utf8helper::charmap_e2_81

charmap_e2_82

const character utf8helper::charmap_e2_82

charmap_e2_b4

const character utf8helper::charmap_e2_b4

charmap_f0_90_92

const character utf8helper::charmap_f0_90_92

charmap_f0_90_93

const character utf8helper::charmap_f0_90_93

charmap_f0_9e_a4

const character utf8helper::charmap_f0_9e_a4

charmap_us7ascii

const character utf8helper::charmap_us7ascii

Referenced by _p0().

IsBreaker

constexpr int utf8helper::IsBreaker = 0x02

constexpr

Referenced by utf8helper::Parser::Reset(), TransformCapitalize(), and TransformNormalize().

IsControl

constexpr int utf8helper::IsControl = 0x04

constexpr

Referenced by TransformCapitalize(), and TransformNormalize().

IsDiacritic

constexpr int utf8helper::IsDiacritic = 0x10

constexpr

IsModifier

constexpr int utf8helper::IsModifier = 0x08

constexpr

IsPunctuation

constexpr int utf8helper::IsPunctuation = 0x20

constexpr

IsSpace

constexpr int utf8helper::IsSpace = 0x01

constexpr

Referenced by utf8helper::Parser::Reset(), TransformCapitalize(), TransformNormalize(), and TransformTransliterate().

None

constexpr int utf8helper::None = 0x00

constexpr

Referenced by _p1_u2(), _p2_u3(), and _p3_u4().

NullCodepoint

constexpr codepoint utf8helper::NullCodepoint = 0

constexpr

Referenced by _p0(), _p1_u2(), _p2_u3(), _p3_u4(), utf8helper::Parser::Reset(), TransformNormalize(), and TransformTransliterate().

pagemap_16

const character * utf8helper::pagemap_16

Initial value:

= {
  
  nullptr, nullptr, charmap_c2, charmap_c3, charmap_c4, charmap_c5, charmap_c6, charmap_c7,
  charmap_c8, charmap_c9, charmap_ca, charmap_cb, charmap_cc, charmap_cd, charmap_ce, charmap_cf,
  charmap_d0, charmap_d1, charmap_d2, charmap_d3, charmap_d4, charmap_d5, charmap_d6, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
}

Referenced by _p1_u2().

pagemap_24_e1

const character * utf8helper::pagemap_24_e1

Initial value:

= {
  
  nullptr, nullptr, charmap_e1_82, charmap_e1_83, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  charmap_e1_b8, charmap_e1_b9, charmap_e1_ba, charmap_e1_bb, charmap_e1_bc, charmap_e1_bd, charmap_e1_be, charmap_e1_bf,
}

Referenced by _p2_u3().

pagemap_24_e2

const character * utf8helper::pagemap_24_e2

Initial value:

= {
  
  charmap_e2_80, charmap_e2_81, charmap_e2_82, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, charmap_e2_b4, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
}

Referenced by _p2_u3().

pagemap_32_f0_90

const character * utf8helper::pagemap_32_f0_90

Initial value:

= {
  
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, charmap_f0_90_92, charmap_f0_90_93, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
}

Referenced by _p3_u4().

pagemap_32_f0_9e

const character * utf8helper::pagemap_32_f0_9e

Initial value:

= {
  
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, charmap_f0_9e_a4, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
  nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr,
}

Referenced by _p3_u4().