// © 2016 and later: Unicode, Inc. and others. // License & terms of use: http://www.unicode.org/copyright.html /* ********************************************************************** * Copyright (c) 2002-2014, International Business Machines * Corporation and others. All Rights Reserved. ********************************************************************** */ #ifndef USETITER_H #define USETITER_H #include "unicode/utypes.h" #if U_SHOW_CPLUSPLUS_API #include "unicode/uobject.h" #include "unicode/unistr.h" /** * \file * \brief C++ API: UnicodeSetIterator iterates over the contents of a UnicodeSet. */ U_NAMESPACE_BEGIN class UnicodeSet; class UnicodeString; /** * * UnicodeSetIterator iterates over the contents of a UnicodeSet. It * iterates over either code points or code point ranges. After all * code points or ranges have been returned, it returns the * multicharacter strings of the UnicodeSet, if any. * * This class is not intended for public subclassing. * * <p>To iterate over code points and strings, use a loop like this: * <pre> * UnicodeSetIterator it(set); * while (it.next()) { * processItem(it.getString()); * } * </pre> * <p>Each item in the set is accessed as a string. Set elements * consisting of single code points are returned as strings containing * just the one code point. * * <p>To iterate over code point ranges, instead of individual code points, * use a loop like this: * <pre> * UnicodeSetIterator it(set); * while (it.nextRange()) { * if (it.isString()) { * processString(it.getString()); * } else { * processCodepointRange(it.getCodepoint(), it.getCodepointEnd()); * } * } * </pre> * * To iterate over only the strings, start with <code>skipToStrings()</code>. * * @author M. Davis * @stable ICU 2.4 */ class U_COMMON_API UnicodeSetIterator U_FINAL : public UObject { /** * Value of <tt>codepoint</tt> if the iterator points to a string. * If <tt>codepoint == IS_STRING</tt>, then examine * <tt>string</tt> for the current iteration result. */ enum { IS_STRING = -1 }; /** * Current code point, or the special value <tt>IS_STRING</tt>, if * the iterator points to a string. */ UChar32 codepoint; /** * When iterating over ranges using <tt>nextRange()</tt>, * <tt>codepointEnd</tt> contains the inclusive end of the * iteration range, if <tt>codepoint != IS_STRING</tt>. If * iterating over code points using <tt>next()</tt>, or if * <tt>codepoint == IS_STRING</tt>, then the value of * <tt>codepointEnd</tt> is undefined. */ UChar32 codepointEnd; /** * If <tt>codepoint == IS_STRING</tt>, then <tt>string</tt> points * to the current string. If <tt>codepoint != IS_STRING</tt>, the * value of <tt>string</tt> is undefined. */ const UnicodeString* string; public: /** * Create an iterator over the given set. The iterator is valid * only so long as <tt>set</tt> is valid. * @param set set to iterate over * @stable ICU 2.4 */ UnicodeSetIterator(const UnicodeSet& set); /** * Create an iterator over nothing. <tt>next()</tt> and * <tt>nextRange()</tt> return false. This is a convenience * constructor allowing the target to be set later. * @stable ICU 2.4 */ UnicodeSetIterator(); /** * Destructor. * @stable ICU 2.4 */ virtual ~UnicodeSetIterator(); /** * Returns true if the current element is a string. If so, the * caller can retrieve it with <tt>getString()</tt>. If this * method returns false, the current element is a code point or * code point range, depending on whether <tt>next()</tt> or * <tt>nextRange()</tt> was called. * Elements of types string and codepoint can both be retrieved * with the function <tt>getString()</tt>. * Elements of type codepoint can also be retrieved with * <tt>getCodepoint()</tt>. * For ranges, <tt>getCodepoint()</tt> returns the starting codepoint * of the range, and <tt>getCodepointEnd()</tt> returns the end * of the range. * @stable ICU 2.4 */ inline UBool isString() const; /** * Returns the current code point, if <tt>isString()</tt> returned * false. Otherwise returns an undefined result. * @stable ICU 2.4 */ inline UChar32 getCodepoint() const; /** * Returns the end of the current code point range, if * <tt>isString()</tt> returned false and <tt>nextRange()</tt> was * called. Otherwise returns an undefined result. * @stable ICU 2.4 */ inline UChar32 getCodepointEnd() const; /** * Returns the current string, if <tt>isString()</tt> returned * true. If the current iteration item is a code point, a UnicodeString * containing that single code point is returned. * * Ownership of the returned string remains with the iterator. * The string is guaranteed to remain valid only until the iterator is * advanced to the next item, or until the iterator is deleted. * * @stable ICU 2.4 */ const UnicodeString& getString(); /** * Skips over the remaining code points/ranges, if any. * A following call to next() or nextRange() will yield a string, if there is one. * No-op if next() would return false, or if it would yield a string anyway. * * @return *this * @stable ICU 70 * @see UnicodeSet#strings() */ inline UnicodeSetIterator &skipToStrings() { // Finish code point/range iteration. range = endRange; endElement = -1; nextElement = 0; return *this; } /** * Advances the iteration position to the next element in the set, * which can be either a single code point or a string. * If there are no more elements in the set, return false. * * <p> * If <tt>isString() == true</tt>, the value is a * string, otherwise the value is a * single code point. Elements of either type can be retrieved * with the function <tt>getString()</tt>, while elements of * consisting of a single code point can be retrieved with * <tt>getCodepoint()</tt> * * <p>The order of iteration is all code points in sorted order, * followed by all strings sorted order. Do not mix * calls to <tt>next()</tt> and <tt>nextRange()</tt> without * calling <tt>reset()</tt> between them. The results of doing so * are undefined. * * @return true if there was another element in the set. * @stable ICU 2.4 */ UBool next(); /** * Returns the next element in the set, either a code point range * or a string. If there are no more elements in the set, return * false. If <tt>isString() == true</tt>, the value is a * string and can be accessed with <tt>getString()</tt>. Otherwise the value is a * range of one or more code points from <tt>getCodepoint()</tt> to * <tt>getCodepointeEnd()</tt> inclusive. * * <p>The order of iteration is all code points ranges in sorted * order, followed by all strings sorted order. Ranges are * disjoint and non-contiguous. The value returned from <tt>getString()</tt> * is undefined unless <tt>isString() == true</tt>. Do not mix calls to * <tt>next()</tt> and <tt>nextRange()</tt> without calling * <tt>reset()</tt> between them. The results of doing so are * undefined. * * @return true if there was another element in the set. * @stable ICU 2.4 */ UBool nextRange(); /** * Sets this iterator to visit the elements of the given set and * resets it to the start of that set. The iterator is valid only * so long as <tt>set</tt> is valid. * @param set the set to iterate over. * @stable ICU 2.4 */ void reset(const UnicodeSet& set); /** * Resets this iterator to the start of the set. * @stable ICU 2.4 */ void reset(); /** * ICU "poor man's RTTI", returns a UClassID for this class. * * @stable ICU 2.4 */ static UClassID U_EXPORT2 getStaticClassID(); /** * ICU "poor man's RTTI", returns a UClassID for the actual class. * * @stable ICU 2.4 */ virtual UClassID getDynamicClassID() const override; // ======================= PRIVATES =========================== private: // endElement and nextElements are really UChar32's, but we keep // them as signed int32_t's so we can do comparisons with // endElement set to -1. Leave them as int32_t's. /** The set */ const UnicodeSet* set; /** End range */ int32_t endRange; /** Range */ int32_t range; /** End element */ int32_t endElement; /** Next element */ int32_t nextElement; /** Next string */ int32_t nextString; /** String count */ int32_t stringCount; /** * Points to the string to use when the caller asks for a * string and the current iteration item is a code point, not a string. */ UnicodeString *cpString; /** Copy constructor. Disallowed. */ UnicodeSetIterator(const UnicodeSetIterator&) = delete; /** Assignment operator. Disallowed. */ UnicodeSetIterator& operator=(const UnicodeSetIterator&) = delete; /** Load range */ void loadRange(int32_t range); }; inline UBool UnicodeSetIterator::isString() const { return codepoint < 0; } inline UChar32 UnicodeSetIterator::getCodepoint() const { return codepoint; } inline UChar32 UnicodeSetIterator::getCodepointEnd() const { return codepointEnd; } U_NAMESPACE_END #endif /* U_SHOW_CPLUSPLUS_API */ #endif
Name | Type | Size | Permission | Actions |
---|---|---|---|---|
alphaindex.h | File | 26.53 KB | 0644 |
|
appendable.h | File | 8.53 KB | 0644 |
|
basictz.h | File | 9.98 KB | 0644 |
|
brkiter.h | File | 27.81 KB | 0644 |
|
bytestream.h | File | 10.75 KB | 0644 |
|
bytestrie.h | File | 20.78 KB | 0644 |
|
bytestriebuilder.h | File | 7.48 KB | 0644 |
|
calendar.h | File | 102.25 KB | 0644 |
|
caniter.h | File | 7.47 KB | 0644 |
|
casemap.h | File | 25.33 KB | 0644 |
|
char16ptr.h | File | 7.22 KB | 0644 |
|
chariter.h | File | 24.06 KB | 0644 |
|
choicfmt.h | File | 23.98 KB | 0644 |
|
coleitr.h | File | 13.78 KB | 0644 |
|
coll.h | File | 56.27 KB | 0644 |
|
compactdecimalformat.h | File | 6.88 KB | 0644 |
|
curramt.h | File | 3.78 KB | 0644 |
|
currpinf.h | File | 7.3 KB | 0644 |
|
currunit.h | File | 4.02 KB | 0644 |
|
datefmt.h | File | 40.71 KB | 0644 |
|
dbbi.h | File | 1.19 KB | 0644 |
|
dcfmtsym.h | File | 20.61 KB | 0644 |
|
decimfmt.h | File | 87.57 KB | 0644 |
|
displayoptions.h | File | 7.11 KB | 0644 |
|
docmain.h | File | 7.21 KB | 0644 |
|
dtfmtsym.h | File | 38.22 KB | 0644 |
|
dtintrv.h | File | 3.85 KB | 0644 |
|
dtitvfmt.h | File | 49.27 KB | 0644 |
|
dtitvinf.h | File | 18.63 KB | 0644 |
|
dtptngen.h | File | 28.69 KB | 0644 |
|
dtrule.h | File | 8.69 KB | 0644 |
|
edits.h | File | 20.74 KB | 0644 |
|
enumset.h | File | 2.08 KB | 0644 |
|
errorcode.h | File | 4.84 KB | 0644 |
|
fieldpos.h | File | 8.7 KB | 0644 |
|
filteredbrk.h | File | 5.37 KB | 0644 |
|
fmtable.h | File | 24.43 KB | 0644 |
|
format.h | File | 12.5 KB | 0644 |
|
formattedvalue.h | File | 9.75 KB | 0644 |
|
fpositer.h | File | 3.03 KB | 0644 |
|
gender.h | File | 3.35 KB | 0644 |
|
gregocal.h | File | 31.04 KB | 0644 |
|
icudataver.h | File | 1.02 KB | 0644 |
|
icuplug.h | File | 11.83 KB | 0644 |
|
idna.h | File | 12.71 KB | 0644 |
|
listformatter.h | File | 8.61 KB | 0644 |
|
localebuilder.h | File | 11.08 KB | 0644 |
|
localematcher.h | File | 26.84 KB | 0644 |
|
localpointer.h | File | 19.36 KB | 0644 |
|
locdspnm.h | File | 7.12 KB | 0644 |
|
locid.h | File | 47.66 KB | 0644 |
|
measfmt.h | File | 11.42 KB | 0644 |
|
measunit.h | File | 106.33 KB | 0644 |
|
measure.h | File | 4.33 KB | 0644 |
|
messagepattern.h | File | 33.71 KB | 0644 |
|
msgfmt.h | File | 44.17 KB | 0644 |
|
normalizer2.h | File | 33.66 KB | 0644 |
|
normlzr.h | File | 30.96 KB | 0644 |
|
nounit.h | File | 2.25 KB | 0644 |
|
numberformatter.h | File | 94.82 KB | 0644 |
|
numberrangeformatter.h | File | 25.02 KB | 0644 |
|
numfmt.h | File | 49.84 KB | 0644 |
|
numsys.h | File | 7.22 KB | 0644 |
|
parseerr.h | File | 3.08 KB | 0644 |
|
parsepos.h | File | 5.57 KB | 0644 |
|
platform.h | File | 28.55 KB | 0644 |
|
plurfmt.h | File | 25.25 KB | 0644 |
|
plurrule.h | File | 20.62 KB | 0644 |
|
ptypes.h | File | 3.49 KB | 0644 |
|
putil.h | File | 6.32 KB | 0644 |
|
rbbi.h | File | 28.46 KB | 0644 |
|
rbnf.h | File | 49.82 KB | 0644 |
|
rbtz.h | File | 15.77 KB | 0644 |
|
regex.h | File | 84.41 KB | 0644 |
|
region.h | File | 9.18 KB | 0644 |
|
reldatefmt.h | File | 22.22 KB | 0644 |
|
rep.h | File | 9.37 KB | 0644 |
|
resbund.h | File | 18.09 KB | 0644 |
|
schriter.h | File | 6.1 KB | 0644 |
|
scientificnumberformatter.h | File | 6.44 KB | 0644 |
|
search.h | File | 22.22 KB | 0644 |
|
selfmt.h | File | 14.35 KB | 0644 |
|
simpleformatter.h | File | 12.59 KB | 0644 |
|
simpletz.h | File | 45.65 KB | 0644 |
|
smpdtfmt.h | File | 71.93 KB | 0644 |
|
sortkey.h | File | 11.18 KB | 0644 |
|
std_string.h | File | 1.05 KB | 0644 |
|
strenum.h | File | 9.92 KB | 0644 |
|
stringoptions.h | File | 5.79 KB | 0644 |
|
stringpiece.h | File | 10.05 KB | 0644 |
|
stringtriebuilder.h | File | 15.47 KB | 0644 |
|
stsearch.h | File | 21.42 KB | 0644 |
|
symtable.h | File | 4.27 KB | 0644 |
|
tblcoll.h | File | 36.92 KB | 0644 |
|
timezone.h | File | 43.81 KB | 0644 |
|
tmunit.h | File | 3.4 KB | 0644 |
|
tmutamt.h | File | 4.91 KB | 0644 |
|
tmutfmt.h | File | 7.42 KB | 0644 |
|
translit.h | File | 65.82 KB | 0644 |
|
tzfmt.h | File | 42.93 KB | 0644 |
|
tznames.h | File | 16.85 KB | 0644 |
|
tzrule.h | File | 34.86 KB | 0644 |
|
tztrans.h | File | 6.13 KB | 0644 |
|
ubidi.h | File | 89.61 KB | 0644 |
|
ubiditransform.h | File | 12.71 KB | 0644 |
|
ubrk.h | File | 24.43 KB | 0644 |
|
ucal.h | File | 60.68 KB | 0644 |
|
ucasemap.h | File | 15.21 KB | 0644 |
|
ucat.h | File | 5.35 KB | 0644 |
|
uchar.h | File | 145.09 KB | 0644 |
|
ucharstrie.h | File | 22.53 KB | 0644 |
|
ucharstriebuilder.h | File | 7.48 KB | 0644 |
|
uchriter.h | File | 13.42 KB | 0644 |
|
uclean.h | File | 11.2 KB | 0644 |
|
ucnv.h | File | 83.46 KB | 0644 |
|
ucnv_cb.h | File | 6.58 KB | 0644 |
|
ucnv_err.h | File | 20.98 KB | 0644 |
|
ucnvsel.h | File | 6.24 KB | 0644 |
|
ucol.h | File | 61.95 KB | 0644 |
|
ucoleitr.h | File | 9.82 KB | 0644 |
|
uconfig.h | File | 12.07 KB | 0644 |
|
ucpmap.h | File | 5.54 KB | 0644 |
|
ucptrie.h | File | 22.51 KB | 0644 |
|
ucsdet.h | File | 14.69 KB | 0644 |
|
ucurr.h | File | 16.72 KB | 0644 |
|
udat.h | File | 62.36 KB | 0644 |
|
udata.h | File | 15.63 KB | 0644 |
|
udateintervalformat.h | File | 11.93 KB | 0644 |
|
udatpg.h | File | 30.18 KB | 0644 |
|
udisplaycontext.h | File | 5.94 KB | 0644 |
|
udisplayoptions.h | File | 8.8 KB | 0644 |
|
uenum.h | File | 7.79 KB | 0644 |
|
ufieldpositer.h | File | 4.41 KB | 0644 |
|
uformattable.h | File | 10.97 KB | 0644 |
|
uformattedvalue.h | File | 12.25 KB | 0644 |
|
ugender.h | File | 2.06 KB | 0644 |
|
uidna.h | File | 33.43 KB | 0644 |
|
uiter.h | File | 22.75 KB | 0644 |
|
uldnames.h | File | 10.48 KB | 0644 |
|
ulistformatter.h | File | 10.78 KB | 0644 |
|
uloc.h | File | 54.66 KB | 0644 |
|
ulocdata.h | File | 11.3 KB | 0644 |
|
umachine.h | File | 15.73 KB | 0644 |
|
umisc.h | File | 1.34 KB | 0644 |
|
umsg.h | File | 24.25 KB | 0644 |
|
umutablecptrie.h | File | 8.3 KB | 0644 |
|
unifilt.h | File | 4 KB | 0644 |
|
unifunct.h | File | 4.05 KB | 0644 |
|
unimatch.h | File | 6.1 KB | 0644 |
|
unirepl.h | File | 3.38 KB | 0644 |
|
uniset.h | File | 66.12 KB | 0644 |
|
unistr.h | File | 170.53 KB | 0644 |
|
unorm.h | File | 20.55 KB | 0644 |
|
unorm2.h | File | 24.68 KB | 0644 |
|
unum.h | File | 56.6 KB | 0644 |
|
unumberformatter.h | File | 30.26 KB | 0644 |
|
unumberrangeformatter.h | File | 15.35 KB | 0644 |
|
unumsys.h | File | 7.26 KB | 0644 |
|
uobject.h | File | 10.68 KB | 0644 |
|
upluralrules.h | File | 8.79 KB | 0644 |
|
uregex.h | File | 71.99 KB | 0644 |
|
uregion.h | File | 9.81 KB | 0644 |
|
ureldatefmt.h | File | 17.04 KB | 0644 |
|
urename.h | File | 135.69 KB | 0644 |
|
urep.h | File | 5.38 KB | 0644 |
|
ures.h | File | 36.54 KB | 0644 |
|
uscript.h | File | 27.8 KB | 0644 |
|
usearch.h | File | 39.21 KB | 0644 |
|
uset.h | File | 44.19 KB | 0644 |
|
usetiter.h | File | 9.63 KB | 0644 |
|
ushape.h | File | 18 KB | 0644 |
|
uspoof.h | File | 65.84 KB | 0644 |
|
usprep.h | File | 8.19 KB | 0644 |
|
ustdio.h | File | 38.56 KB | 0644 |
|
ustream.h | File | 1.89 KB | 0644 |
|
ustring.h | File | 72.36 KB | 0644 |
|
ustringtrie.h | File | 3.15 KB | 0644 |
|
utext.h | File | 58.08 KB | 0644 |
|
utf.h | File | 7.87 KB | 0644 |
|
utf16.h | File | 23.35 KB | 0644 |
|
utf32.h | File | 763 B | 0644 |
|
utf8.h | File | 30.83 KB | 0644 |
|
utf_old.h | File | 45.8 KB | 0644 |
|
utmscale.h | File | 13.78 KB | 0644 |
|
utrace.h | File | 17.18 KB | 0644 |
|
utrans.h | File | 25.54 KB | 0644 |
|
utypes.h | File | 31.06 KB | 0644 |
|
uvernum.h | File | 6.33 KB | 0644 |
|
uversion.h | File | 5.99 KB | 0644 |
|
vtzone.h | File | 20.68 KB | 0644 |
|