Linux vps-61133.fhnet.fr 4.9.0-19-amd64 #1 SMP Debian 4.9.320-2 (2022-06-30) x86_64
Apache/2.4.25 (Debian)
Server IP : 93.113.207.21 & Your IP : 216.73.216.35
Domains :
Cant Read [ /etc/named.conf ]
User : www-data
Terminal
Auto Root
Create File
Create Folder
Localroot Suggester
Backdoor Destroyer
Readme
/
usr /
include /
unicode /
Delete
Unzip
Name
Size
Permission
Date
Action
alphaindex.h
26.13
KB
-rw-r--r--
2021-10-12 12:29
appendable.h
8.25
KB
-rw-r--r--
2021-10-12 12:29
basictz.h
8.73
KB
-rw-r--r--
2021-10-12 12:29
brkiter.h
27.31
KB
-rw-r--r--
2021-10-12 12:29
bytestream.h
9.16
KB
-rw-r--r--
2021-10-12 12:29
bytestrie.h
19.19
KB
-rw-r--r--
2021-10-12 12:29
bytestriebuilder.h
7.06
KB
-rw-r--r--
2021-10-12 12:29
calendar.h
105.26
KB
-rw-r--r--
2021-10-12 12:29
caniter.h
7.24
KB
-rw-r--r--
2021-10-12 12:29
chariter.h
23.71
KB
-rw-r--r--
2021-10-12 12:29
choicfmt.h
23.74
KB
-rw-r--r--
2021-10-12 12:29
coleitr.h
13.57
KB
-rw-r--r--
2021-10-12 12:29
coll.h
55.46
KB
-rw-r--r--
2021-10-12 12:29
compactdecimalformat.h
16.56
KB
-rw-r--r--
2021-10-12 12:29
curramt.h
3.57
KB
-rw-r--r--
2021-10-12 12:29
currpinf.h
6.97
KB
-rw-r--r--
2021-10-12 12:29
currunit.h
2.81
KB
-rw-r--r--
2021-10-12 12:29
datefmt.h
40.07
KB
-rw-r--r--
2021-10-12 12:29
dbbi.h
1020
B
-rw-r--r--
2021-10-12 12:29
dcfmtsym.h
16.46
KB
-rw-r--r--
2021-10-12 12:29
decimfmt.h
90.45
KB
-rw-r--r--
2021-10-12 12:29
docmain.h
6.28
KB
-rw-r--r--
2021-10-12 12:29
dtfmtsym.h
37.05
KB
-rw-r--r--
2021-10-12 12:29
dtintrv.h
3.65
KB
-rw-r--r--
2021-10-12 12:29
dtitvfmt.h
42.76
KB
-rw-r--r--
2021-10-12 12:29
dtitvinf.h
18.34
KB
-rw-r--r--
2021-10-12 12:29
dtptngen.h
23.18
KB
-rw-r--r--
2021-10-12 12:29
dtrule.h
8.51
KB
-rw-r--r--
2021-10-12 12:29
enumset.h
1.94
KB
-rw-r--r--
2021-10-12 12:29
errorcode.h
4.67
KB
-rw-r--r--
2021-10-12 12:29
fieldpos.h
8.51
KB
-rw-r--r--
2021-10-12 12:29
filteredbrk.h
4.22
KB
-rw-r--r--
2021-10-12 12:29
fmtable.h
24.29
KB
-rw-r--r--
2021-10-12 12:29
format.h
12.33
KB
-rw-r--r--
2021-10-12 12:29
fpositer.h
2.87
KB
-rw-r--r--
2021-10-12 12:29
gender.h
3.07
KB
-rw-r--r--
2021-10-12 12:29
gregocal.h
31.46
KB
-rw-r--r--
2021-10-12 12:29
icudataver.h
938
B
-rw-r--r--
2021-10-12 12:29
icuplug.h
11.58
KB
-rw-r--r--
2021-10-12 12:29
idna.h
12.55
KB
-rw-r--r--
2021-10-12 12:29
listformatter.h
4.73
KB
-rw-r--r--
2021-10-12 12:29
localpointer.h
18.17
KB
-rw-r--r--
2021-10-12 12:29
locdspnm.h
6.92
KB
-rw-r--r--
2021-10-12 12:29
locid.h
31.3
KB
-rw-r--r--
2021-10-12 12:29
measfmt.h
10.43
KB
-rw-r--r--
2021-10-12 12:29
measunit.h
37.64
KB
-rw-r--r--
2021-10-12 12:29
measure.h
4.15
KB
-rw-r--r--
2021-10-12 12:29
messagepattern.h
33.53
KB
-rw-r--r--
2021-10-12 12:29
msgfmt.h
43.05
KB
-rw-r--r--
2021-10-12 12:29
normalizer2.h
28.17
KB
-rw-r--r--
2021-10-12 12:29
normlzr.h
30.65
KB
-rw-r--r--
2021-10-12 12:29
numfmt.h
45.85
KB
-rw-r--r--
2021-10-12 12:29
numsys.h
6.62
KB
-rw-r--r--
2021-10-12 12:29
parseerr.h
2.97
KB
-rw-r--r--
2021-10-12 12:29
parsepos.h
5.34
KB
-rw-r--r--
2021-10-12 12:29
platform.h
26.96
KB
-rw-r--r--
2021-10-12 12:29
plurfmt.h
25.55
KB
-rw-r--r--
2021-10-12 12:29
plurrule.h
17.16
KB
-rw-r--r--
2021-10-12 12:29
ptypes.h
3.36
KB
-rw-r--r--
2021-10-12 12:29
putil.h
6.23
KB
-rw-r--r--
2021-10-12 12:29
rbbi.h
29.2
KB
-rw-r--r--
2021-10-12 12:29
rbnf.h
46.66
KB
-rw-r--r--
2021-10-12 12:29
rbtz.h
15.36
KB
-rw-r--r--
2021-10-12 12:29
regex.h
84.75
KB
-rw-r--r--
2021-10-12 12:29
region.h
9
KB
-rw-r--r--
2021-10-12 12:29
reldatefmt.h
13.73
KB
-rw-r--r--
2021-10-12 12:29
rep.h
9.38
KB
-rw-r--r--
2021-10-12 12:29
resbund.h
17.89
KB
-rw-r--r--
2021-10-12 12:29
schriter.h
6.15
KB
-rw-r--r--
2021-10-12 12:29
scientificnumberformatter.h
6.48
KB
-rw-r--r--
2021-10-12 12:29
search.h
22.1
KB
-rw-r--r--
2021-10-12 12:29
selfmt.h
14.13
KB
-rw-r--r--
2021-10-12 12:29
simpleformatter.h
11.14
KB
-rw-r--r--
2021-10-12 12:29
simpletz.h
45.13
KB
-rw-r--r--
2021-10-12 12:29
smpdtfmt.h
69.24
KB
-rw-r--r--
2021-10-12 12:29
sortkey.h
11.01
KB
-rw-r--r--
2021-10-12 12:29
std_string.h
905
B
-rw-r--r--
2021-10-12 12:29
strenum.h
9.92
KB
-rw-r--r--
2021-10-12 12:29
stringpiece.h
6.38
KB
-rw-r--r--
2021-10-12 12:29
stringtriebuilder.h
14.56
KB
-rw-r--r--
2021-10-12 12:29
stsearch.h
21.18
KB
-rw-r--r--
2021-10-12 12:29
symtable.h
4.1
KB
-rw-r--r--
2021-10-12 12:29
tblcoll.h
36.24
KB
-rw-r--r--
2021-10-12 12:29
timezone.h
40.71
KB
-rw-r--r--
2021-10-12 12:29
tmunit.h
2.94
KB
-rw-r--r--
2021-10-12 12:29
tmutamt.h
4.69
KB
-rw-r--r--
2021-10-12 12:29
tmutfmt.h
7.61
KB
-rw-r--r--
2021-10-12 12:29
translit.h
54.41
KB
-rw-r--r--
2021-10-12 12:29
tzfmt.h
42.69
KB
-rw-r--r--
2021-10-12 12:29
tznames.h
16.25
KB
-rw-r--r--
2021-10-12 12:29
tzrule.h
35.21
KB
-rw-r--r--
2021-10-12 12:29
tztrans.h
5.96
KB
-rw-r--r--
2021-10-12 12:29
ubidi.h
88.57
KB
-rw-r--r--
2021-10-12 12:29
ubrk.h
20.05
KB
-rw-r--r--
2021-10-12 12:29
ucal.h
54.18
KB
-rw-r--r--
2021-10-12 12:29
ucasemap.h
16.45
KB
-rw-r--r--
2021-10-12 12:29
ucat.h
5.25
KB
-rw-r--r--
2021-10-12 12:29
uchar.h
122.74
KB
-rw-r--r--
2021-10-12 12:29
ucharstrie.h
20.8
KB
-rw-r--r--
2021-10-12 12:29
ucharstriebuilder.h
7.01
KB
-rw-r--r--
2021-10-12 12:29
uchriter.h
12.82
KB
-rw-r--r--
2021-10-12 12:29
uclean.h
11.03
KB
-rw-r--r--
2021-10-12 12:29
ucnv.h
82.81
KB
-rw-r--r--
2021-10-12 12:29
ucnv_cb.h
6.48
KB
-rw-r--r--
2021-10-12 12:29
ucnv_err.h
20.87
KB
-rw-r--r--
2021-10-12 12:29
ucnvsel.h
6.03
KB
-rw-r--r--
2021-10-12 12:29
ucol.h
60.51
KB
-rw-r--r--
2021-10-12 12:29
ucoleitr.h
9.34
KB
-rw-r--r--
2021-10-12 12:29
uconfig.h
11.76
KB
-rw-r--r--
2021-10-12 12:29
ucsdet.h
14.35
KB
-rw-r--r--
2021-10-12 12:29
ucurr.h
14.92
KB
-rw-r--r--
2021-10-12 12:29
udat.h
59.91
KB
-rw-r--r--
2021-10-12 12:29
udata.h
15.28
KB
-rw-r--r--
2021-10-12 12:29
udateintervalformat.h
6.68
KB
-rw-r--r--
2021-10-12 12:29
udatpg.h
23.5
KB
-rw-r--r--
2021-10-12 12:29
udisplaycontext.h
4.87
KB
-rw-r--r--
2021-10-12 12:29
uenum.h
7.76
KB
-rw-r--r--
2021-10-12 12:29
ufieldpositer.h
4.24
KB
-rw-r--r--
2021-10-12 12:29
uformattable.h
10.65
KB
-rw-r--r--
2021-10-12 12:29
ugender.h
1.89
KB
-rw-r--r--
2021-10-12 12:29
uidna.h
33.26
KB
-rw-r--r--
2021-10-12 12:29
uiter.h
22.67
KB
-rw-r--r--
2021-10-12 12:29
uldnames.h
10.34
KB
-rw-r--r--
2021-10-12 12:29
ulistformatter.h
4.43
KB
-rw-r--r--
2021-10-12 12:29
uloc.h
49.78
KB
-rw-r--r--
2021-10-12 12:29
ulocdata.h
10.57
KB
-rw-r--r--
2021-10-12 12:29
umachine.h
10.94
KB
-rw-r--r--
2021-10-12 12:29
umisc.h
1.23
KB
-rw-r--r--
2021-10-12 12:29
umsg.h
24.12
KB
-rw-r--r--
2021-10-12 12:29
unifilt.h
3.52
KB
-rw-r--r--
2021-10-12 12:29
unifunct.h
3.87
KB
-rw-r--r--
2021-10-12 12:29
unimatch.h
5.93
KB
-rw-r--r--
2021-10-12 12:29
unirepl.h
3.21
KB
-rw-r--r--
2021-10-12 12:29
uniset.h
63.86
KB
-rw-r--r--
2021-10-12 12:29
unistr.h
170.03
KB
-rw-r--r--
2021-10-12 12:29
unorm.h
20.18
KB
-rw-r--r--
2021-10-12 12:29
unorm2.h
25.11
KB
-rw-r--r--
2021-10-12 12:29
unum.h
49.13
KB
-rw-r--r--
2021-10-12 12:29
unumsys.h
7.03
KB
-rw-r--r--
2021-10-12 12:29
uobject.h
10.61
KB
-rw-r--r--
2021-10-12 12:29
upluralrules.h
4.47
KB
-rw-r--r--
2021-10-12 12:29
uregex.h
71.95
KB
-rw-r--r--
2021-10-12 12:29
uregion.h
9.58
KB
-rw-r--r--
2021-10-12 12:29
ureldatefmt.h
11.85
KB
-rw-r--r--
2021-10-12 12:29
urename.h
123.32
KB
-rw-r--r--
2021-10-12 12:29
urep.h
5.27
KB
-rw-r--r--
2021-10-12 12:29
ures.h
36.25
KB
-rw-r--r--
2021-10-12 12:29
uscript.h
25.24
KB
-rw-r--r--
2021-10-12 12:29
usearch.h
37.71
KB
-rw-r--r--
2021-10-12 12:29
uset.h
39.67
KB
-rw-r--r--
2021-10-12 12:29
usetiter.h
9.38
KB
-rw-r--r--
2021-10-12 12:29
ushape.h
17.89
KB
-rw-r--r--
2021-10-12 12:29
uspoof.h
45.43
KB
-rw-r--r--
2021-10-12 12:29
usprep.h
8.02
KB
-rw-r--r--
2021-10-12 12:29
ustdio.h
38.43
KB
-rw-r--r--
2021-10-12 12:29
ustream.h
1.8
KB
-rw-r--r--
2021-10-12 12:29
ustring.h
72.69
KB
-rw-r--r--
2021-10-12 12:29
ustringtrie.h
3.04
KB
-rw-r--r--
2021-10-12 12:29
utext.h
57.97
KB
-rw-r--r--
2021-10-12 12:29
utf.h
7.64
KB
-rw-r--r--
2021-10-12 12:29
utf16.h
18.63
KB
-rw-r--r--
2021-10-12 12:29
utf32.h
653
B
-rw-r--r--
2021-10-12 12:29
utf8.h
27.62
KB
-rw-r--r--
2021-10-12 12:29
utf_old.h
42.5
KB
-rw-r--r--
2021-10-12 12:29
utmscale.h
13.37
KB
-rw-r--r--
2021-10-12 12:29
utrace.h
13.09
KB
-rw-r--r--
2021-10-12 12:29
utrans.h
25.42
KB
-rw-r--r--
2021-10-12 12:29
utypes.h
31.05
KB
-rw-r--r--
2021-10-12 12:29
uvernum.h
5.62
KB
-rw-r--r--
2021-10-12 12:29
uversion.h
6.29
KB
-rw-r--r--
2021-10-12 12:29
vtzone.h
20.06
KB
-rw-r--r--
2021-10-12 12:29
Save
Rename
/* ******************************************************************************* * Copyright (C) 2010-2012, International Business Machines * Corporation and others. All Rights Reserved. ******************************************************************************* * file name: bytestrie.h * encoding: US-ASCII * tab size: 8 (not used) * indentation:4 * * created on: 2010sep25 * created by: Markus W. Scherer */ #ifndef __BYTESTRIE_H__ #define __BYTESTRIE_H__ /** * \file * \brief C++ API: Trie for mapping byte sequences to integer values. */ #include "unicode/utypes.h" #include "unicode/stringpiece.h" #include "unicode/uobject.h" #include "unicode/ustringtrie.h" U_NAMESPACE_BEGIN class ByteSink; class BytesTrieBuilder; class CharString; class UVector32; /** * Light-weight, non-const reader class for a BytesTrie. * Traverses a byte-serialized data structure with minimal state, * for mapping byte sequences to non-negative integer values. * * This class owns the serialized trie data only if it was constructed by * the builder's build() method. * The public constructor and the copy constructor only alias the data (only copy the pointer). * There is no assignment operator. * * This class is not intended for public subclassing. * @stable ICU 4.8 */ class U_COMMON_API BytesTrie : public UMemory { public: /** * Constructs a BytesTrie reader instance. * * The trieBytes must contain a copy of a byte sequence from the BytesTrieBuilder, * starting with the first byte of that sequence. * The BytesTrie object will not read more bytes than * the BytesTrieBuilder generated in the corresponding build() call. * * The array is not copied/cloned and must not be modified while * the BytesTrie object is in use. * * @param trieBytes The byte array that contains the serialized trie. * @stable ICU 4.8 */ BytesTrie(const void *trieBytes) : ownedArray_(NULL), bytes_(static_cast<const uint8_t *>(trieBytes)), pos_(bytes_), remainingMatchLength_(-1) {} /** * Destructor. * @stable ICU 4.8 */ ~BytesTrie(); /** * Copy constructor, copies the other trie reader object and its state, * but not the byte array which will be shared. (Shallow copy.) * @param other Another BytesTrie object. * @stable ICU 4.8 */ BytesTrie(const BytesTrie &other) : ownedArray_(NULL), bytes_(other.bytes_), pos_(other.pos_), remainingMatchLength_(other.remainingMatchLength_) {} /** * Resets this trie to its initial state. * @return *this * @stable ICU 4.8 */ BytesTrie &reset() { pos_=bytes_; remainingMatchLength_=-1; return *this; } /** * BytesTrie state object, for saving a trie's current state * and resetting the trie back to this state later. * @stable ICU 4.8 */ class State : public UMemory { public: /** * Constructs an empty State. * @stable ICU 4.8 */ State() { bytes=NULL; } private: friend class BytesTrie; const uint8_t *bytes; const uint8_t *pos; int32_t remainingMatchLength; }; /** * Saves the state of this trie. * @param state The State object to hold the trie's state. * @return *this * @see resetToState * @stable ICU 4.8 */ const BytesTrie &saveState(State &state) const { state.bytes=bytes_; state.pos=pos_; state.remainingMatchLength=remainingMatchLength_; return *this; } /** * Resets this trie to the saved state. * If the state object contains no state, or the state of a different trie, * then this trie remains unchanged. * @param state The State object which holds a saved trie state. * @return *this * @see saveState * @see reset * @stable ICU 4.8 */ BytesTrie &resetToState(const State &state) { if(bytes_==state.bytes && bytes_!=NULL) { pos_=state.pos; remainingMatchLength_=state.remainingMatchLength; } return *this; } /** * Determines whether the byte sequence so far matches, whether it has a value, * and whether another input byte can continue a matching byte sequence. * @return The match/value Result. * @stable ICU 4.8 */ UStringTrieResult current() const; /** * Traverses the trie from the initial state for this input byte. * Equivalent to reset().next(inByte). * @param inByte Input byte value. Values -0x100..-1 are treated like 0..0xff. * Values below -0x100 and above 0xff will never match. * @return The match/value Result. * @stable ICU 4.8 */ inline UStringTrieResult first(int32_t inByte) { remainingMatchLength_=-1; if(inByte<0) { inByte+=0x100; } return nextImpl(bytes_, inByte); } /** * Traverses the trie from the current state for this input byte. * @param inByte Input byte value. Values -0x100..-1 are treated like 0..0xff. * Values below -0x100 and above 0xff will never match. * @return The match/value Result. * @stable ICU 4.8 */ UStringTrieResult next(int32_t inByte); /** * Traverses the trie from the current state for this byte sequence. * Equivalent to * \code * Result result=current(); * for(each c in s) * if(!USTRINGTRIE_HAS_NEXT(result)) return USTRINGTRIE_NO_MATCH; * result=next(c); * return result; * \endcode * @param s A string or byte sequence. Can be NULL if length is 0. * @param length The length of the byte sequence. Can be -1 if NUL-terminated. * @return The match/value Result. * @stable ICU 4.8 */ UStringTrieResult next(const char *s, int32_t length); /** * Returns a matching byte sequence's value if called immediately after * current()/first()/next() returned USTRINGTRIE_INTERMEDIATE_VALUE or USTRINGTRIE_FINAL_VALUE. * getValue() can be called multiple times. * * Do not call getValue() after USTRINGTRIE_NO_MATCH or USTRINGTRIE_NO_VALUE! * @return The value for the byte sequence so far. * @stable ICU 4.8 */ inline int32_t getValue() const { const uint8_t *pos=pos_; int32_t leadByte=*pos++; // U_ASSERT(leadByte>=kMinValueLead); return readValue(pos, leadByte>>1); } /** * Determines whether all byte sequences reachable from the current state * map to the same value. * @param uniqueValue Receives the unique value, if this function returns TRUE. * (output-only) * @return TRUE if all byte sequences reachable from the current state * map to the same value. * @stable ICU 4.8 */ inline UBool hasUniqueValue(int32_t &uniqueValue) const { const uint8_t *pos=pos_; // Skip the rest of a pending linear-match node. return pos!=NULL && findUniqueValue(pos+remainingMatchLength_+1, FALSE, uniqueValue); } /** * Finds each byte which continues the byte sequence from the current state. * That is, each byte b for which it would be next(b)!=USTRINGTRIE_NO_MATCH now. * @param out Each next byte is appended to this object. * (Only uses the out.Append(s, length) method.) * @return the number of bytes which continue the byte sequence from here * @stable ICU 4.8 */ int32_t getNextBytes(ByteSink &out) const; /** * Iterator for all of the (byte sequence, value) pairs in a BytesTrie. * @stable ICU 4.8 */ class U_COMMON_API Iterator : public UMemory { public: /** * Iterates from the root of a byte-serialized BytesTrie. * @param trieBytes The trie bytes. * @param maxStringLength If 0, the iterator returns full strings/byte sequences. * Otherwise, the iterator returns strings with this maximum length. * @param errorCode Standard ICU error code. Its input value must * pass the U_SUCCESS() test, or else the function returns * immediately. Check for U_FAILURE() on output or use with * function chaining. (See User Guide for details.) * @stable ICU 4.8 */ Iterator(const void *trieBytes, int32_t maxStringLength, UErrorCode &errorCode); /** * Iterates from the current state of the specified BytesTrie. * @param trie The trie whose state will be copied for iteration. * @param maxStringLength If 0, the iterator returns full strings/byte sequences. * Otherwise, the iterator returns strings with this maximum length. * @param errorCode Standard ICU error code. Its input value must * pass the U_SUCCESS() test, or else the function returns * immediately. Check for U_FAILURE() on output or use with * function chaining. (See User Guide for details.) * @stable ICU 4.8 */ Iterator(const BytesTrie &trie, int32_t maxStringLength, UErrorCode &errorCode); /** * Destructor. * @stable ICU 4.8 */ ~Iterator(); /** * Resets this iterator to its initial state. * @return *this * @stable ICU 4.8 */ Iterator &reset(); /** * @return TRUE if there are more elements. * @stable ICU 4.8 */ UBool hasNext() const; /** * Finds the next (byte sequence, value) pair if there is one. * * If the byte sequence is truncated to the maximum length and does not * have a real value, then the value is set to -1. * In this case, this "not a real value" is indistinguishable from * a real value of -1. * @param errorCode Standard ICU error code. Its input value must * pass the U_SUCCESS() test, or else the function returns * immediately. Check for U_FAILURE() on output or use with * function chaining. (See User Guide for details.) * @return TRUE if there is another element. * @stable ICU 4.8 */ UBool next(UErrorCode &errorCode); /** * @return The NUL-terminated byte sequence for the last successful next(). * @stable ICU 4.8 */ const StringPiece &getString() const { return sp_; } /** * @return The value for the last successful next(). * @stable ICU 4.8 */ int32_t getValue() const { return value_; } private: UBool truncateAndStop(); const uint8_t *branchNext(const uint8_t *pos, int32_t length, UErrorCode &errorCode); const uint8_t *bytes_; const uint8_t *pos_; const uint8_t *initialPos_; int32_t remainingMatchLength_; int32_t initialRemainingMatchLength_; CharString *str_; StringPiece sp_; int32_t maxLength_; int32_t value_; // The stack stores pairs of integers for backtracking to another // outbound edge of a branch node. // The first integer is an offset from bytes_. // The second integer has the str_->length() from before the node in bits 15..0, // and the remaining branch length in bits 24..16. (Bits 31..25 are unused.) // (We could store the remaining branch length minus 1 in bits 23..16 and not use bits 31..24, // but the code looks more confusing that way.) UVector32 *stack_; }; private: friend class BytesTrieBuilder; /** * Constructs a BytesTrie reader instance. * Unlike the public constructor which just aliases an array, * this constructor adopts the builder's array. * This constructor is only called by the builder. */ BytesTrie(void *adoptBytes, const void *trieBytes) : ownedArray_(static_cast<uint8_t *>(adoptBytes)), bytes_(static_cast<const uint8_t *>(trieBytes)), pos_(bytes_), remainingMatchLength_(-1) {} // No assignment operator. BytesTrie &operator=(const BytesTrie &other); inline void stop() { pos_=NULL; } // Reads a compact 32-bit integer. // pos is already after the leadByte, and the lead byte is already shifted right by 1. static int32_t readValue(const uint8_t *pos, int32_t leadByte); static inline const uint8_t *skipValue(const uint8_t *pos, int32_t leadByte) { // U_ASSERT(leadByte>=kMinValueLead); if(leadByte>=(kMinTwoByteValueLead<<1)) { if(leadByte<(kMinThreeByteValueLead<<1)) { ++pos; } else if(leadByte<(kFourByteValueLead<<1)) { pos+=2; } else { pos+=3+((leadByte>>1)&1); } } return pos; } static inline const uint8_t *skipValue(const uint8_t *pos) { int32_t leadByte=*pos++; return skipValue(pos, leadByte); } // Reads a jump delta and jumps. static const uint8_t *jumpByDelta(const uint8_t *pos); static inline const uint8_t *skipDelta(const uint8_t *pos) { int32_t delta=*pos++; if(delta>=kMinTwoByteDeltaLead) { if(delta<kMinThreeByteDeltaLead) { ++pos; } else if(delta<kFourByteDeltaLead) { pos+=2; } else { pos+=3+(delta&1); } } return pos; } static inline UStringTrieResult valueResult(int32_t node) { return (UStringTrieResult)(USTRINGTRIE_INTERMEDIATE_VALUE-(node&kValueIsFinal)); } // Handles a branch node for both next(byte) and next(string). UStringTrieResult branchNext(const uint8_t *pos, int32_t length, int32_t inByte); // Requires remainingLength_<0. UStringTrieResult nextImpl(const uint8_t *pos, int32_t inByte); // Helper functions for hasUniqueValue(). // Recursively finds a unique value (or whether there is not a unique one) // from a branch. static const uint8_t *findUniqueValueFromBranch(const uint8_t *pos, int32_t length, UBool haveUniqueValue, int32_t &uniqueValue); // Recursively finds a unique value (or whether there is not a unique one) // starting from a position on a node lead byte. static UBool findUniqueValue(const uint8_t *pos, UBool haveUniqueValue, int32_t &uniqueValue); // Helper functions for getNextBytes(). // getNextBytes() when pos is on a branch node. static void getNextBranchBytes(const uint8_t *pos, int32_t length, ByteSink &out); static void append(ByteSink &out, int c); // BytesTrie data structure // // The trie consists of a series of byte-serialized nodes for incremental // string/byte sequence matching. The root node is at the beginning of the trie data. // // Types of nodes are distinguished by their node lead byte ranges. // After each node, except a final-value node, another node follows to // encode match values or continue matching further bytes. // // Node types: // - Value node: Stores a 32-bit integer in a compact, variable-length format. // The value is for the string/byte sequence so far. // One node bit indicates whether the value is final or whether // matching continues with the next node. // - Linear-match node: Matches a number of bytes. // - Branch node: Branches to other nodes according to the current input byte. // The node byte is the length of the branch (number of bytes to select from) // minus 1. It is followed by a sub-node: // - If the length is at most kMaxBranchLinearSubNodeLength, then // there are length-1 (key, value) pairs and then one more comparison byte. // If one of the key bytes matches, then the value is either a final value for // the string/byte sequence so far, or a "jump" delta to the next node. // If the last byte matches, then matching continues with the next node. // (Values have the same encoding as value nodes.) // - If the length is greater than kMaxBranchLinearSubNodeLength, then // there is one byte and one "jump" delta. // If the input byte is less than the sub-node byte, then "jump" by delta to // the next sub-node which will have a length of length/2. // (The delta has its own compact encoding.) // Otherwise, skip the "jump" delta to the next sub-node // which will have a length of length-length/2. // Node lead byte values. // 00..0f: Branch node. If node!=0 then the length is node+1, otherwise // the length is one more than the next byte. // For a branch sub-node with at most this many entries, we drop down // to a linear search. static const int32_t kMaxBranchLinearSubNodeLength=5; // 10..1f: Linear-match node, match 1..16 bytes and continue reading the next node. static const int32_t kMinLinearMatch=0x10; static const int32_t kMaxLinearMatchLength=0x10; // 20..ff: Variable-length value node. // If odd, the value is final. (Otherwise, intermediate value or jump delta.) // Then shift-right by 1 bit. // The remaining lead byte value indicates the number of following bytes (0..4) // and contains the value's top bits. static const int32_t kMinValueLead=kMinLinearMatch+kMaxLinearMatchLength; // 0x20 // It is a final value if bit 0 is set. static const int32_t kValueIsFinal=1; // Compact value: After testing bit 0, shift right by 1 and then use the following thresholds. static const int32_t kMinOneByteValueLead=kMinValueLead/2; // 0x10 static const int32_t kMaxOneByteValue=0x40; // At least 6 bits in the first byte. static const int32_t kMinTwoByteValueLead=kMinOneByteValueLead+kMaxOneByteValue+1; // 0x51 static const int32_t kMaxTwoByteValue=0x1aff; static const int32_t kMinThreeByteValueLead=kMinTwoByteValueLead+(kMaxTwoByteValue>>8)+1; // 0x6c static const int32_t kFourByteValueLead=0x7e; // A little more than Unicode code points. (0x11ffff) static const int32_t kMaxThreeByteValue=((kFourByteValueLead-kMinThreeByteValueLead)<<16)-1; static const int32_t kFiveByteValueLead=0x7f; // Compact delta integers. static const int32_t kMaxOneByteDelta=0xbf; static const int32_t kMinTwoByteDeltaLead=kMaxOneByteDelta+1; // 0xc0 static const int32_t kMinThreeByteDeltaLead=0xf0; static const int32_t kFourByteDeltaLead=0xfe; static const int32_t kFiveByteDeltaLead=0xff; static const int32_t kMaxTwoByteDelta=((kMinThreeByteDeltaLead-kMinTwoByteDeltaLead)<<8)-1; // 0x2fff static const int32_t kMaxThreeByteDelta=((kFourByteDeltaLead-kMinThreeByteDeltaLead)<<16)-1; // 0xdffff uint8_t *ownedArray_; // Fixed value referencing the BytesTrie bytes. const uint8_t *bytes_; // Iterator variables. // Pointer to next trie byte to read. NULL if no more matches. const uint8_t *pos_; // Remaining length of a linear-match node, minus 1. Negative if not in such a node. int32_t remainingMatchLength_; }; U_NAMESPACE_END #endif // __BYTESTRIE_H__