56 int32_t
encode(uint8_t* utf8, int32_t length);
58 int32_t
utf16to8(uint8_t* utf8, int32_t length);
59 int32_t
utf32to8(uint8_t* utf8, int32_t length);
93 int32_t
decode(
wchar_t* unicode, int32_t length);
95 int32_t
utf8to16(
wchar_t* unicode, int32_t length);
96 int32_t
utf8to32(
wchar_t* unicode, int32_t length);
132 int32_t
decode(
wchar_t* unicode, int32_t length);
#define LUCENE_CLASS(Name)
Definition LuceneObject.h:24
Base class for all Lucene classes.
Definition LuceneObject.h:31
Definition UTF8Stream.h:120
int32_t utf16to32(wchar_t *unicode, int32_t length)
int32_t utf16to16(wchar_t *unicode, int32_t length)
virtual uint32_t readNext()
const uint16_t * utf16End
Definition UTF8Stream.h:129
int32_t decode(wchar_t *unicode, int32_t length)
const uint16_t * utf16Begin
Definition UTF8Stream.h:128
UTF16Decoder(const uint16_t *utf16Begin, const uint16_t *utf16End)
Definition UTF8Stream.h:14
bool isLeadSurrogate(uint32_t cp)
uint8_t mask8(uint32_t b)
static const wchar_t UNICODE_TERMINATOR
Definition UTF8Stream.h:29
static const uint16_t LEAD_OFFSET
Definition UTF8Stream.h:24
static const uint16_t TRAIL_SURROGATE_MAX
Definition UTF8Stream.h:23
virtual uint32_t readNext()=0
bool isSurrogate(uint32_t cp)
static const uint16_t LEAD_SURROGATE_MAX
Definition UTF8Stream.h:21
bool isTrailSurrogate(uint32_t cp)
static const uint32_t CODE_POINT_MAX
Definition UTF8Stream.h:26
static const uint16_t LEAD_SURROGATE_MIN
Definition UTF8Stream.h:20
bool isValidCodePoint(uint32_t cp)
bool isOverlongSequence(uint32_t cp, int32_t length)
static const uint32_t SURROGATE_OFFSET
Definition UTF8Stream.h:25
static const wchar_t UNICODE_REPLACEMENT_CHAR
Definition UTF8Stream.h:28
uint16_t mask16(uint32_t c)
static const uint16_t TRAIL_SURROGATE_MIN
Definition UTF8Stream.h:22
Definition UTF8Stream.h:106
ReaderPtr reader
Definition UTF8Stream.h:114
virtual uint32_t readNext()
UTF8DecoderStream(const ReaderPtr &reader)
virtual ~UTF8DecoderStream()
Definition UTF8Stream.h:81
bool isValidNext(uint32_t &cp)
int32_t utf8to32(wchar_t *unicode, int32_t length)
int32_t sequenceLength(uint32_t cp)
virtual uint32_t readNext()
int32_t utf8to16(wchar_t *unicode, int32_t length)
UTF8Decoder(const uint8_t *utf8Begin, const uint8_t *utf8End)
bool getSequence(uint32_t &cp, int32_t length)
const uint8_t * utf8Begin
Definition UTF8Stream.h:89
const uint8_t * utf8End
Definition UTF8Stream.h:90
int32_t decode(wchar_t *unicode, int32_t length)
Definition UTF8Stream.h:67
virtual uint32_t readNext()
ReaderPtr reader
Definition UTF8Stream.h:75
virtual ~UTF8EncoderStream()
UTF8EncoderStream(const ReaderPtr &reader)
Definition UTF8Stream.h:44
const wchar_t * unicodeEnd
Definition UTF8Stream.h:53
int32_t utf32to8(uint8_t *utf8, int32_t length)
UTF8Encoder(const wchar_t *unicodeBegin, const wchar_t *unicodeEnd)
uint8_t * appendChar(uint8_t *utf8, uint32_t cp)
int32_t encode(uint8_t *utf8, int32_t length)
const wchar_t * unicodeBegin
Definition UTF8Stream.h:52
int32_t utf16to8(uint8_t *utf8, int32_t length)
virtual uint32_t readNext()
Definition AbstractAllTermDocs.h:12
boost::shared_ptr< Reader > ReaderPtr
Definition LuceneTypes.h:547