OXIESEC PANEL
- Current Dir:
/
/
usr
/
include
/
unicode
Server IP: 139.59.38.164
Upload:
Create Dir:
Name
Size
Modified
Perms
📁
..
-
10/28/2024 06:50:42 AM
rwxr-xr-x
📄
alphaindex.h
26.27 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
appendable.h
8.43 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
basictz.h
8.84 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
brkiter.h
27.54 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
bytestream.h
9.59 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
bytestrie.h
19.26 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
bytestriebuilder.h
7.1 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
calendar.h
105.35 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
caniter.h
7.35 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
casemap.h
25.31 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
char16ptr.h
7.29 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
chariter.h
24 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
choicfmt.h
23.85 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
coleitr.h
13.7 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
coll.h
55.61 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
compactdecimalformat.h
16.58 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
curramt.h
3.69 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
currpinf.h
7.08 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
currunit.h
3.49 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
datefmt.h
40.18 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
dbbi.h
1.11 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
dcfmtsym.h
17.66 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
decimfmt.h
89.73 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
docmain.h
6.56 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
dtfmtsym.h
37.64 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
dtintrv.h
3.76 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
dtitvfmt.h
42.87 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
dtitvinf.h
18.46 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
dtptngen.h
23.72 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
dtrule.h
8.62 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
edits.h
15.54 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
enumset.h
2.05 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
errorcode.h
4.78 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
fieldpos.h
8.63 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
filteredbrk.h
5.46 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
fmtable.h
24.39 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
format.h
12.44 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
fpositer.h
3.14 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
gender.h
3.18 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
gregocal.h
31.57 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
icudataver.h
1.03 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
icuplug.h
11.86 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
idna.h
12.63 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
listformatter.h
4.98 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
localpointer.h
18.19 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
locdspnm.h
7.05 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
locid.h
31.4 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
measfmt.h
11.25 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
measunit.h
37.74 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
measure.h
4.26 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
messagepattern.h
33.64 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
msgfmt.h
43.16 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
normalizer2.h
33.97 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
normlzr.h
30.74 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
nounit.h
2.6 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
numberformatter.h
64.98 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
numfmt.h
47.76 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
numsys.h
6.73 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
parseerr.h
3.08 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
parsepos.h
5.45 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
platform.h
27.71 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
plurfmt.h
25.66 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
plurrule.h
18.34 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ptypes.h
3.47 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
putil.h
6.34 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
rbbi.h
27.09 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
rbnf.h
49.76 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
rbtz.h
15.47 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
regex.h
84.91 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
region.h
9.12 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
reldatefmt.h
14.3 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
rep.h
9.5 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
resbund.h
18.01 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
schriter.h
6.26 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
scientificnumberformatter.h
6.59 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
search.h
22.21 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
selfmt.h
14.24 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
simpleformatter.h
11.43 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
simpletz.h
45.24 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
smpdtfmt.h
70.09 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
sortkey.h
11.12 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
std_string.h
1015 bytes
10/19/2021 07:44:28 PM
rw-r--r--
📄
strenum.h
10.04 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
stringoptions.h
5.89 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
stringpiece.h
6.46 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
stringtriebuilder.h
15.2 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
stsearch.h
21.29 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
symtable.h
4.21 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
tblcoll.h
36.4 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
timezone.h
40.83 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
tmunit.h
3.29 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
tmutamt.h
4.8 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
tmutfmt.h
7.79 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
translit.h
54.53 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
tzfmt.h
42.81 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
tznames.h
16.79 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
tzrule.h
35.32 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
tztrans.h
6.07 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ubidi.h
89.28 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ubiditransform.h
12.71 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ubrk.h
24.08 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucal.h
54.56 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucasemap.h
15.18 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucat.h
5.36 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uchar.h
131.61 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucharstrie.h
21.06 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucharstriebuilder.h
7.14 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uchriter.h
13.14 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uclean.h
11.24 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucnv.h
83.06 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucnv_cb.h
6.59 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucnv_err.h
20.98 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucnvsel.h
6.14 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucol.h
61.36 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucoleitr.h
9.46 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uconfig.h
11.91 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucsdet.h
14.67 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ucurr.h
15.15 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
udat.h
60.13 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
udata.h
15.53 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
udateintervalformat.h
6.79 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
udatpg.h
24.11 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
udisplaycontext.h
5.89 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uenum.h
7.9 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ufieldpositer.h
4.36 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uformattable.h
10.94 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ugender.h
2 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uidna.h
33.37 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uiter.h
22.77 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uldnames.h
10.45 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ulistformatter.h
4.54 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uloc.h
50.75 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ulocdata.h
11.26 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
umachine.h
13.1 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
umisc.h
1.33 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
umsg.h
24.23 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
unifilt.h
3.63 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
unifunct.h
3.98 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
unimatch.h
6.04 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
unirepl.h
3.32 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uniset.h
63.98 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
unistr.h
175.65 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
unorm.h
20.45 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
unorm2.h
24.66 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
unum.h
52.49 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
unumsys.h
7.14 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uobject.h
10.72 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
upluralrules.h
6.77 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uregex.h
72.05 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uregion.h
9.84 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ureldatefmt.h
12.22 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
urename.h
124.49 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
urep.h
5.38 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ures.h
36.52 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uscript.h
25.91 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
usearch.h
38.14 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uset.h
39.91 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
usetiter.h
9.49 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ushape.h
18 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uspoof.h
64.9 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
usprep.h
8.13 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ustdio.h
38.54 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ustream.h
1.8 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ustring.h
72.52 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
ustringtrie.h
3.15 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
utext.h
58.11 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
utf.h
7.86 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
utf16.h
22.49 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
utf32.h
763 bytes
10/19/2021 07:44:28 PM
rw-r--r--
📄
utf8.h
28.16 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
utf_old.h
43.78 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
utmscale.h
13.78 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
utrace.h
13.89 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
utrans.h
25.53 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
utypes.h
29.92 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uvernum.h
5.69 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
uversion.h
6.4 KB
10/19/2021 07:44:28 PM
rw-r--r--
📄
vtzone.h
20.17 KB
10/19/2021 07:44:28 PM
rw-r--r--
Editing: ucharstrie.h
Close
// © 2016 and later: Unicode, Inc. and others. // License & terms of use: http://www.unicode.org/copyright.html /* ******************************************************************************* * Copyright (C) 2010-2012, International Business Machines * Corporation and others. All Rights Reserved. ******************************************************************************* * file name: ucharstrie.h * encoding: UTF-8 * tab size: 8 (not used) * indentation:4 * * created on: 2010nov14 * created by: Markus W. Scherer */ #ifndef __UCHARSTRIE_H__ #define __UCHARSTRIE_H__ /** * \file * \brief C++ API: Trie for mapping Unicode strings (or 16-bit-unit sequences) * to integer values. */ #include "unicode/utypes.h" #include "unicode/unistr.h" #include "unicode/uobject.h" #include "unicode/ustringtrie.h" U_NAMESPACE_BEGIN class Appendable; class UCharsTrieBuilder; class UVector32; /** * Light-weight, non-const reader class for a UCharsTrie. * Traverses a char16_t-serialized data structure with minimal state, * for mapping strings (16-bit-unit 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 UCharsTrie : public UMemory { public: /** * Constructs a UCharsTrie reader instance. * * The trieUChars must contain a copy of a char16_t sequence from the UCharsTrieBuilder, * starting with the first char16_t of that sequence. * The UCharsTrie object will not read more char16_ts than * the UCharsTrieBuilder generated in the corresponding build() call. * * The array is not copied/cloned and must not be modified while * the UCharsTrie object is in use. * * @param trieUChars The char16_t array that contains the serialized trie. * @stable ICU 4.8 */ UCharsTrie(ConstChar16Ptr trieUChars) : ownedArray_(NULL), uchars_(trieUChars), pos_(uchars_), remainingMatchLength_(-1) {} /** * Destructor. * @stable ICU 4.8 */ ~UCharsTrie(); /** * Copy constructor, copies the other trie reader object and its state, * but not the char16_t array which will be shared. (Shallow copy.) * @param other Another UCharsTrie object. * @stable ICU 4.8 */ UCharsTrie(const UCharsTrie &other) : ownedArray_(NULL), uchars_(other.uchars_), pos_(other.pos_), remainingMatchLength_(other.remainingMatchLength_) {} /** * Resets this trie to its initial state. * @return *this * @stable ICU 4.8 */ UCharsTrie &reset() { pos_=uchars_; remainingMatchLength_=-1; return *this; } /** * UCharsTrie 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() { uchars=NULL; } private: friend class UCharsTrie; const char16_t *uchars; const char16_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 UCharsTrie &saveState(State &state) const { state.uchars=uchars_; 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 */ UCharsTrie &resetToState(const State &state) { if(uchars_==state.uchars && uchars_!=NULL) { pos_=state.pos; remainingMatchLength_=state.remainingMatchLength; } return *this; } /** * Determines whether the string so far matches, whether it has a value, * and whether another input char16_t can continue a matching string. * @return The match/value Result. * @stable ICU 4.8 */ UStringTrieResult current() const; /** * Traverses the trie from the initial state for this input char16_t. * Equivalent to reset().next(uchar). * @param uchar Input char value. Values below 0 and above 0xffff will never match. * @return The match/value Result. * @stable ICU 4.8 */ inline UStringTrieResult first(int32_t uchar) { remainingMatchLength_=-1; return nextImpl(uchars_, uchar); } /** * Traverses the trie from the initial state for the * one or two UTF-16 code units for this input code point. * Equivalent to reset().nextForCodePoint(cp). * @param cp A Unicode code point 0..0x10ffff. * @return The match/value Result. * @stable ICU 4.8 */ UStringTrieResult firstForCodePoint(UChar32 cp); /** * Traverses the trie from the current state for this input char16_t. * @param uchar Input char value. Values below 0 and above 0xffff will never match. * @return The match/value Result. * @stable ICU 4.8 */ UStringTrieResult next(int32_t uchar); /** * Traverses the trie from the current state for the * one or two UTF-16 code units for this input code point. * @param cp A Unicode code point 0..0x10ffff. * @return The match/value Result. * @stable ICU 4.8 */ UStringTrieResult nextForCodePoint(UChar32 cp); /** * Traverses the trie from the current state for this string. * 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. Can be NULL if length is 0. * @param length The length of the string. Can be -1 if NUL-terminated. * @return The match/value Result. * @stable ICU 4.8 */ UStringTrieResult next(ConstChar16Ptr s, int32_t length); /** * Returns a matching string'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 string so far. * @stable ICU 4.8 */ inline int32_t getValue() const { const char16_t *pos=pos_; int32_t leadUnit=*pos++; // U_ASSERT(leadUnit>=kMinValueLead); return leadUnit&kValueIsFinal ? readValue(pos, leadUnit&0x7fff) : readNodeValue(pos, leadUnit); } /** * Determines whether all strings 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 strings reachable from the current state * map to the same value. * @stable ICU 4.8 */ inline UBool hasUniqueValue(int32_t &uniqueValue) const { const char16_t *pos=pos_; // Skip the rest of a pending linear-match node. return pos!=NULL && findUniqueValue(pos+remainingMatchLength_+1, FALSE, uniqueValue); } /** * Finds each char16_t which continues the string from the current state. * That is, each char16_t c for which it would be next(c)!=USTRINGTRIE_NO_MATCH now. * @param out Each next char16_t is appended to this object. * @return the number of char16_ts which continue the string from here * @stable ICU 4.8 */ int32_t getNextUChars(Appendable &out) const; /** * Iterator for all of the (string, value) pairs in a UCharsTrie. * @stable ICU 4.8 */ class U_COMMON_API Iterator : public UMemory { public: /** * Iterates from the root of a char16_t-serialized UCharsTrie. * @param trieUChars The trie char16_ts. * @param maxStringLength If 0, the iterator returns full strings. * 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(ConstChar16Ptr trieUChars, int32_t maxStringLength, UErrorCode &errorCode); /** * Iterates from the current state of the specified UCharsTrie. * @param trie The trie whose state will be copied for iteration. * @param maxStringLength If 0, the iterator returns full strings. * 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 UCharsTrie &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 (string, value) pair if there is one. * * If the string 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 string for the last successful next(). * @stable ICU 4.8 */ const UnicodeString &getString() const { return str_; } /** * @return The value for the last successful next(). * @stable ICU 4.8 */ int32_t getValue() const { return value_; } private: UBool truncateAndStop() { pos_=NULL; value_=-1; // no real value for str return TRUE; } const char16_t *branchNext(const char16_t *pos, int32_t length, UErrorCode &errorCode); const char16_t *uchars_; const char16_t *pos_; const char16_t *initialPos_; int32_t remainingMatchLength_; int32_t initialRemainingMatchLength_; UBool skipValue_; // Skip intermediate value which was already delivered. UnicodeString str_; 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 uchars_. // The second integer has the str_.length() from before the node in bits 15..0, // and the remaining branch length in bits 31..16. // (We could store the remaining branch length minus 1 in bits 30..16 and not use the sign bit, // but the code looks more confusing that way.) UVector32 *stack_; }; private: friend class UCharsTrieBuilder; /** * Constructs a UCharsTrie 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. */ UCharsTrie(char16_t *adoptUChars, const char16_t *trieUChars) : ownedArray_(adoptUChars), uchars_(trieUChars), pos_(uchars_), remainingMatchLength_(-1) {} // No assignment operator. UCharsTrie &operator=(const UCharsTrie &other); inline void stop() { pos_=NULL; } // Reads a compact 32-bit integer. // pos is already after the leadUnit, and the lead unit has bit 15 reset. static inline int32_t readValue(const char16_t *pos, int32_t leadUnit) { int32_t value; if(leadUnit<kMinTwoUnitValueLead) { value=leadUnit; } else if(leadUnit<kThreeUnitValueLead) { value=((leadUnit-kMinTwoUnitValueLead)<<16)|*pos; } else { value=(pos[0]<<16)|pos[1]; } return value; } static inline const char16_t *skipValue(const char16_t *pos, int32_t leadUnit) { if(leadUnit>=kMinTwoUnitValueLead) { if(leadUnit<kThreeUnitValueLead) { ++pos; } else { pos+=2; } } return pos; } static inline const char16_t *skipValue(const char16_t *pos) { int32_t leadUnit=*pos++; return skipValue(pos, leadUnit&0x7fff); } static inline int32_t readNodeValue(const char16_t *pos, int32_t leadUnit) { // U_ASSERT(kMinValueLead<=leadUnit && leadUnit<kValueIsFinal); int32_t value; if(leadUnit<kMinTwoUnitNodeValueLead) { value=(leadUnit>>6)-1; } else if(leadUnit<kThreeUnitNodeValueLead) { value=(((leadUnit&0x7fc0)-kMinTwoUnitNodeValueLead)<<10)|*pos; } else { value=(pos[0]<<16)|pos[1]; } return value; } static inline const char16_t *skipNodeValue(const char16_t *pos, int32_t leadUnit) { // U_ASSERT(kMinValueLead<=leadUnit && leadUnit<kValueIsFinal); if(leadUnit>=kMinTwoUnitNodeValueLead) { if(leadUnit<kThreeUnitNodeValueLead) { ++pos; } else { pos+=2; } } return pos; } static inline const char16_t *jumpByDelta(const char16_t *pos) { int32_t delta=*pos++; if(delta>=kMinTwoUnitDeltaLead) { if(delta==kThreeUnitDeltaLead) { delta=(pos[0]<<16)|pos[1]; pos+=2; } else { delta=((delta-kMinTwoUnitDeltaLead)<<16)|*pos++; } } return pos+delta; } static const char16_t *skipDelta(const char16_t *pos) { int32_t delta=*pos++; if(delta>=kMinTwoUnitDeltaLead) { if(delta==kThreeUnitDeltaLead) { pos+=2; } else { ++pos; } } return pos; } static inline UStringTrieResult valueResult(int32_t node) { return (UStringTrieResult)(USTRINGTRIE_INTERMEDIATE_VALUE-(node>>15)); } // Handles a branch node for both next(uchar) and next(string). UStringTrieResult branchNext(const char16_t *pos, int32_t length, int32_t uchar); // Requires remainingLength_<0. UStringTrieResult nextImpl(const char16_t *pos, int32_t uchar); // Helper functions for hasUniqueValue(). // Recursively finds a unique value (or whether there is not a unique one) // from a branch. static const char16_t *findUniqueValueFromBranch(const char16_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 unit. static UBool findUniqueValue(const char16_t *pos, UBool haveUniqueValue, int32_t &uniqueValue); // Helper functions for getNextUChars(). // getNextUChars() when pos is on a branch node. static void getNextBranchUChars(const char16_t *pos, int32_t length, Appendable &out); // UCharsTrie data structure // // The trie consists of a series of char16_t-serialized nodes for incremental // Unicode string/char16_t sequence matching. (char16_t=16-bit unsigned integer) // The root node is at the beginning of the trie data. // // Types of nodes are distinguished by their node lead unit ranges. // After each node, except a final-value node, another node follows to // encode match values or continue matching further units. // // Node types: // - Final-value node: Stores a 32-bit integer in a compact, variable-length format. // The value is for the string/char16_t sequence so far. // - Match node, optionally with an intermediate value in a different compact format. // The value, if present, is for the string/char16_t sequence so far. // // Aside from the value, which uses the node lead unit's high bits: // // - Linear-match node: Matches a number of units. // - Branch node: Branches to other nodes according to the current input unit. // The node unit is the length of the branch (number of units 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 unit. // If one of the key units matches, then the value is either a final value for // the string so far, or a "jump" delta to the next node. // If the last unit matches, then matching continues with the next node. // (Values have the same encoding as final-value nodes.) // - If the length is greater than kMaxBranchLinearSubNodeLength, then // there is one unit and one "jump" delta. // If the input unit is less than the sub-node unit, 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. // Match-node lead unit values, after masking off intermediate-value bits: // 0000..002f: Branch node. If node!=0 then the length is node+1, otherwise // the length is one more than the next unit. // For a branch sub-node with at most this many entries, we drop down // to a linear search. static const int32_t kMaxBranchLinearSubNodeLength=5; // 0030..003f: Linear-match node, match 1..16 units and continue reading the next node. static const int32_t kMinLinearMatch=0x30; static const int32_t kMaxLinearMatchLength=0x10; // Match-node lead unit bits 14..6 for the optional intermediate value. // If these bits are 0, then there is no intermediate value. // Otherwise, see the *NodeValue* constants below. static const int32_t kMinValueLead=kMinLinearMatch+kMaxLinearMatchLength; // 0x0040 static const int32_t kNodeTypeMask=kMinValueLead-1; // 0x003f // A final-value node has bit 15 set. static const int32_t kValueIsFinal=0x8000; // Compact value: After testing and masking off bit 15, use the following thresholds. static const int32_t kMaxOneUnitValue=0x3fff; static const int32_t kMinTwoUnitValueLead=kMaxOneUnitValue+1; // 0x4000 static const int32_t kThreeUnitValueLead=0x7fff; static const int32_t kMaxTwoUnitValue=((kThreeUnitValueLead-kMinTwoUnitValueLead)<<16)-1; // 0x3ffeffff // Compact intermediate-value integer, lead unit shared with a branch or linear-match node. static const int32_t kMaxOneUnitNodeValue=0xff; static const int32_t kMinTwoUnitNodeValueLead=kMinValueLead+((kMaxOneUnitNodeValue+1)<<6); // 0x4040 static const int32_t kThreeUnitNodeValueLead=0x7fc0; static const int32_t kMaxTwoUnitNodeValue= ((kThreeUnitNodeValueLead-kMinTwoUnitNodeValueLead)<<10)-1; // 0xfdffff // Compact delta integers. static const int32_t kMaxOneUnitDelta=0xfbff; static const int32_t kMinTwoUnitDeltaLead=kMaxOneUnitDelta+1; // 0xfc00 static const int32_t kThreeUnitDeltaLead=0xffff; static const int32_t kMaxTwoUnitDelta=((kThreeUnitDeltaLead-kMinTwoUnitDeltaLead)<<16)-1; // 0x03feffff char16_t *ownedArray_; // Fixed value referencing the UCharsTrie words. const char16_t *uchars_; // Iterator variables. // Pointer to next trie unit to read. NULL if no more matches. const char16_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 // __UCHARSTRIE_H__