34 uint8_t
mask8(uint32_t b);
35 uint16_t
mask16(uint32_t c);
46 UTF8Encoder(
const wchar_t* unicodeBegin,
const wchar_t* unicodeEnd);
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);
64 uint8_t* appendChar(uint8_t* utf8, uint32_t cp);
83 UTF8Decoder(
const uint8_t* utf8Begin,
const uint8_t* utf8End);
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);
101 int32_t sequenceLength(uint32_t cp);
102 bool getSequence(uint32_t& cp, int32_t length);
103 bool isValidNext(uint32_t& cp);
122 UTF16Decoder(
const uint16_t* utf16Begin,
const uint16_t* utf16End);
132 int32_t decode(
wchar_t* unicode, int32_t length);
134 int32_t utf16to16(
wchar_t* unicode, int32_t length);
135 int32_t utf16to32(
wchar_t* unicode, int32_t length);
static const uint16_t TRAIL_SURROGATE_MAX
Definition: UTF8Stream.h:23
static const wchar_t UNICODE_REPLACEMENT_CHAR
Definition: UTF8Stream.h:28
static const wchar_t UNICODE_TERMINATOR
Definition: UTF8Stream.h:29
boost::shared_ptr< Reader > ReaderPtr
Definition: LuceneTypes.h:547
const uint16_t * utf16Begin
Definition: UTF8Stream.h:125
bool isSurrogate(uint32_t cp)
const uint16_t * utf16End
Definition: UTF8Stream.h:129
bool isTrailSurrogate(uint32_t cp)
bool isValidCodePoint(uint32_t cp)
static const uint16_t LEAD_SURROGATE_MAX
Definition: UTF8Stream.h:21
Definition: UTF8Stream.h:106
const uint8_t * utf8Begin
Definition: UTF8Stream.h:86
virtual uint32_t readNext()=0
const wchar_t * unicodeEnd
Definition: UTF8Stream.h:53
static const uint32_t SURROGATE_OFFSET
Definition: UTF8Stream.h:25
uint16_t mask16(uint32_t c)
Definition: UTF8Stream.h:81
static const uint16_t LEAD_SURROGATE_MIN
Definition: UTF8Stream.h:17
Definition: UTF8Stream.h:120
const wchar_t * unicodeBegin
Definition: UTF8Stream.h:49
static const uint16_t TRAIL_SURROGATE_MIN
Definition: UTF8Stream.h:22
Definition: UTF8Stream.h:44
uint8_t mask8(uint32_t b)
Definition: UTF8Stream.h:14
Base class for all Lucene classes.
Definition: LuceneObject.h:31
#define LUCENE_CLASS(Name)
Definition: LuceneObject.h:24
Definition: AbstractAllTermDocs.h:12
bool isLeadSurrogate(uint32_t cp)
static const uint16_t LEAD_OFFSET
Definition: UTF8Stream.h:24
static const uint32_t CODE_POINT_MAX
Definition: UTF8Stream.h:26
Definition: UTF8Stream.h:67
const uint8_t * utf8End
Definition: UTF8Stream.h:90
bool isOverlongSequence(uint32_t cp, int32_t length)