1 // © 2016 and later: Unicode, Inc. and others. 2 // License & terms of use: http://www.unicode.org/copyright.html 4 ******************************************************************************* 6 * Copyright (C) 2002-2014, International Business Machines 7 * Corporation and others. All Rights Reserved. 9 ******************************************************************************* 12 * tab size: 8 (not used) 15 * created on: 2002mar07 16 * created by: Markus W. Scherer 18 * C version of UnicodeSet. 35 #if U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API 37 #include <string_view> 235 #ifndef U_HIDE_DEPRECATED_API 241 #endif// U_HIDE_DEPRECATED_API 282 /********************************************************************* 284 *********************************************************************/ 349 #if U_SHOW_CPLUSPLUS_API 461 constUChar *pattern, int32_t patternLength,
528 constUChar *prop, int32_t propLength,
529 constUChar *value, int32_t valueLength,
562 UChar* result, int32_t resultCapacity,
563 UBool escapeUnprintable,
1062 UChar* str, int32_t strCapacity,
1208 /********************************************************************* 1209 * Serialized set API 1210 *********************************************************************/ 1325 #if U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API 1327 namespaceU_HEADER_ONLY_NAMESPACE {
1329 // Note: Not U_COMMON_API, and not a subclass of UMemory, because this is a header-only class, 1330 // not intended to be used via export from the ICU DLL. 1343 // No need to compare rangeCount & end given private constructor 1344 // and assuming we don't compare iterators across the set being modified. 1345 // And comparing rangeIndex is redundant with comparing c. 1346 // We might even skip comparing uset. 1347 // Unless we want operator==() to be "correct" for more than iteration. 1348 return uset == other.uset && c == other.c;
1364 }
elseif (rangeIndex < rangeCount) {
1366 int32_t result =
uset_getItem(uset, rangeIndex, &c, &end,
nullptr, 0, &errorCode);
1367 if (
U_SUCCESS(errorCode) && result == 0) {
1392 : uset(pUset), rangeIndex(nRangeIndex), rangeCount(nRangeCount),
1394 // Fetch the first range. 1527 // No need to compare rangeCount given private constructor 1528 // and assuming we don't compare iterators across the set being modified. 1529 // We might even skip comparing uset. 1530 // Unless we want operator==() to be "correct" for more than iteration. 1531 return uset == other.uset && rangeIndex == other.rangeIndex;
1539 if (rangeIndex < rangeCount) {
1542 int32_t result =
uset_getItem(uset, rangeIndex, &start, &end,
nullptr, 0, &errorCode);
1543 if (
U_SUCCESS(errorCode) && result == 0) {
1573 : uset(pUset), rangeIndex(nRangeIndex), rangeCount(nRangeCount) {}
1640 // No need to compare count given private constructor 1641 // and assuming we don't compare iterators across the set being modified. 1642 // We might even skip comparing uset. 1643 // Unless we want operator==() to be "correct" for more than iteration. 1644 return uset == other.uset && index == other.index;
1652 if (index < count) {
1655 // assert uchars != nullptr; 1656 return {uprv_char16PtrFromUChar(uchars),
static_cast<size_t>(length)};
1684 : uset(pUset), index(nIndex), count(nCount) {}
1739 #ifndef U_HIDE_DRAFT_API 1751 // No need to compare rangeCount & end given private constructor 1752 // and assuming we don't compare iterators across the set being modified. 1753 // We might even skip comparing uset. 1754 // Unless we want operator==() to be "correct" for more than iteration. 1755 return uset == other.uset && c == other.c && index == other.index;
1764 return c <= 0xffff ?
1765 std::u16string({
static_cast<char16_t
>(c)}) :
1767 }
elseif (index < totalCount) {
1770 // assert uchars != nullptr; 1771 return {uprv_char16PtrFromUChar(uchars),
static_cast<size_t>(length)};
1784 }
elseif (index < rangeCount) {
1786 int32_t result =
uset_getItem(uset, index, &c, &end,
nullptr, 0, &errorCode);
1787 if (
U_SUCCESS(errorCode) && result == 0) {
1793 // assert index == rangeCount; 1794 // Switch from the last range to the first string. 1816 : uset(pUset), index(nIndex), rangeCount(nRangeCount), totalCount(nTotalCount),
1818 if (index < rangeCount) {
1819 // Fetch the first range. 1822 // Otherwise don't move beyond the (index - rangeCount)-th string. 1889 returnUSetElementIterator(uset, rangeCount + stringCount, rangeCount, rangeCount + stringCount);
1894 int32_t rangeCount, stringCount;
1897 #endif// U_HIDE_DRAFT_API 1899 }
// namespace U_HEADER_ONLY_NAMESPACE 1901 #endif// U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API C++ API: char16_t pointer wrappers with implicit conversion from bit-compatible raw pointer types.
"Smart pointer" class, closes a USet via uset_close().
Iterator returned by USetCodePoints.
bool operator==(const USetCodePointIterator &other) const
UChar32 operator*() const
bool operator!=(const USetCodePointIterator &other) const
USetCodePointIterator operator++(int)
Post-increment.
USetCodePointIterator(const USetCodePointIterator &other)=default
USetCodePointIterator & operator++()
Pre-increment.
C++ "range" for iterating over the code points of a USet.
USetCodePoints(const USetCodePoints &other)=default
USetCodePointIterator end() const
USetCodePoints(const USet *pUset)
Constructs a C++ "range" object over the code points of the USet.
USetCodePointIterator begin() const
Iterator returned by USetElements.
USetElementIterator & operator++()
Pre-increment.
USetElementIterator operator++(int)
Post-increment.
bool operator==(const USetElementIterator &other) const
bool operator!=(const USetElementIterator &other) const
USetElementIterator(const USetElementIterator &other)=default
std::u16string operator*() const
A C++ "range" for iterating over all of the elements of a USet.
USetElements(const USetElements &other)=default
USetElementIterator end() const
USetElements(const USet *pUset)
Constructs a C++ "range" object over all of the elements of the USet.
USetElementIterator begin() const
Iterator returned by USetRanges.
USetRangeIterator & operator++()
Pre-increment.
CodePointRange operator*() const
bool operator==(const USetRangeIterator &other) const
USetRangeIterator operator++(int)
Post-increment.
bool operator!=(const USetRangeIterator &other) const
USetRangeIterator(const USetRangeIterator &other)=default
C++ "range" for iterating over the code point ranges of a USet.
USetRangeIterator end() const
USetRangeIterator begin() const
USetRanges(const USet *pUset)
Constructs a C++ "range" object over the code point ranges of the USet.
USetRanges(const USetRanges &other)=default
Iterator returned by USetStrings.
USetStringIterator(const USetStringIterator &other)=default
bool operator!=(const USetStringIterator &other) const
std::u16string_view operator*() const
USetStringIterator operator++(int)
Post-increment.
USetStringIterator & operator++()
Pre-increment.
bool operator==(const USetStringIterator &other) const
C++ "range" for iterating over the empty and multi-character strings of a USet.
USetStrings(const USetStrings &other)=default
USetStringIterator begin() const
USetStringIterator end() const
USetStrings(const USet *pUset)
Constructs a C++ "range" object over the strings of the USet.
C++ API: "Smart pointers" for use with and in ICU4C C++ code.
#define U_DEFINE_LOCAL_OPEN_POINTER(LocalPointerClassName, Type, closeFunction)
"Smart pointer" definition macro, deletes objects via the closeFunction.
A serialized form of a Unicode set.
uint16_t staticArray[USET_SERIALIZED_STATIC_ARRAY_CAPACITY]
A small buffer for the array to reduce memory allocations.
int32_t bmpLength
The length of the array that contains BMP characters.
const uint16_t * array
The serialized Unicode Set.
int32_t length
The total length of the array.
bool operator!=(const iterator &other) const
UChar32 operator*() const
UChar32 c
The current code point in the range.
iterator operator++(int)
Post-increment.
iterator & operator++()
Pre-increment.
bool operator==(const iterator &other) const
A contiguous range of code points in a USet/UnicodeSet.
CodePointRange(UChar32 start, UChar32 end)
UChar32 rangeEnd
Inclusive end of a USet/UnicodeSet range of code points.
CodePointRange(const CodePointRange &other)=default
UChar32 rangeStart
Start of a USet/UnicodeSet range of code points.
C API: Unicode Properties.
UProperty
Selection constants for Unicode properties.
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
int8_t UBool
The ICU boolean type, a signed-byte integer.
#define U_CAPI
This is used to declare a function as a public ICU C API.
char16_t UChar
The base type for UTF-16 code units and pointers.
#define U_SENTINEL
This value is intended for sentinel values for APIs that (take or) return single code points (UChar32...
U_CAPI UBool uset_containsAll(const USet *set1, const USet *set2)
Returns true if set1 contains all the characters and strings of set2.
U_CAPI UBool uset_hasStrings(const USet *set)
U_CAPI int32_t uset_getStringCount(const USet *set)
U_CAPI UBool uset_serializedContains(const USerializedSet *set, UChar32 c)
Returns true if the given USerializedSet contains the given character.
U_CAPI UBool uset_getSerializedSet(USerializedSet *fillSet, const uint16_t *src, int32_t srcLength)
Given a serialized array, fill in the given serialized set object.
U_CAPI void uset_set(USet *set, UChar32 start, UChar32 end)
Causes the USet object to represent the range start - end.
U_CAPI void uset_addString(USet *set, const UChar *str, int32_t strLen)
Adds the given string to the given USet.
U_CAPI UBool uset_containsString(const USet *set, const UChar *str, int32_t strLen)
Returns true if the given USet contains the given string.
@ USET_IGNORE_SPACE
Ignore white space within patterns unless quoted or escaped.
@ USET_ADD_CASE_MAPPINGS
Adds all case mappings for each element in the set.
@ USET_CASE_INSENSITIVE
Enable case insensitive matching.
@ USET_SIMPLE_CASE_INSENSITIVE
Enable case insensitive matching.
U_CAPI UBool uset_isEmpty(const USet *set)
Returns true if the given USet contains no characters and no strings.
U_CAPI int32_t uset_getItemCount(const USet *set)
Returns the number of items in this set.
U_CAPI int32_t uset_getSerializedRangeCount(const USerializedSet *set)
Returns the number of disjoint ranges of characters contained in the given serialized set.
U_CAPI UChar32 uset_charAt(const USet *set, int32_t charIndex)
Returns the character at the given index within this set, where the set is ordered by ascending code ...
U_CAPI USet * uset_open(UChar32 start, UChar32 end)
Creates a USet object that contains the range of characters start..end, inclusive.
U_CAPI void uset_retainAll(USet *set, const USet *retain)
Retains only the elements in this set that are contained in the specified set.
U_CAPI void uset_setSerializedToOne(USerializedSet *fillSet, UChar32 c)
Set the USerializedSet to contain the given character (and nothing else).
U_CAPI void uset_complementString(USet *set, const UChar *str, int32_t length)
Complements the specified string in this set.
U_CAPI void uset_applyIntPropertyValue(USet *set, UProperty prop, int32_t value, UErrorCode *ec)
Modifies the set to contain those code points which have the given value for the given binary or enum...
U_CAPI void uset_complementRange(USet *set, UChar32 start, UChar32 end)
Complements the specified range in this set.
U_CAPI UBool uset_resemblesPattern(const UChar *pattern, int32_t patternLength, int32_t pos)
Return true if the given position, in the given pattern, appears to be the start of a UnicodeSet patt...
U_CAPI void uset_addRange(USet *set, UChar32 start, UChar32 end)
Adds the given range of characters to the given USet.
U_CAPI void uset_add(USet *set, UChar32 c)
Adds the given character to the given USet.
U_CAPI void uset_complementAllCodePoints(USet *set, const UChar *str, int32_t length)
Complements EACH of the characters in this string.
U_CAPI int32_t uset_size(const USet *set)
Returns the number of characters and strings contained in this set.
U_CAPI int32_t uset_span(const USet *set, const UChar *s, int32_t length, USetSpanCondition spanCondition)
Returns the length of the initial substring of the input string which consists only of characters and...
U_CAPI void uset_removeAllStrings(USet *set)
Remove all strings from this set.
U_CAPI void uset_clear(USet *set)
Removes all of the elements from this set.
U_CAPI void uset_removeAllCodePoints(USet *set, const UChar *str, int32_t length)
Removes EACH of the characters in this string.
USetSpanCondition
Argument values for whether span() and similar functions continue while the current character is cont...
@ USET_SPAN_NOT_CONTAINED
Continues a span() while there is no set element at the current position.
@ USET_SPAN_CONTAINED
Spans the longest substring that is a concatenation of set elements (characters or strings).
@ USET_SPAN_CONDITION_COUNT
One more than the last span condition.
@ USET_SPAN_SIMPLE
Continues a span() while there is a set element at the current position.
U_CAPI void uset_remove(USet *set, UChar32 c)
Removes the given character from the given USet.
U_CAPI void uset_removeRange(USet *set, UChar32 start, UChar32 end)
Removes the given range of characters from the given USet.
U_CAPI UBool uset_equals(const USet *set1, const USet *set2)
Returns true if set1 contains all of the characters and strings of set2, and vis versa.
U_CAPI UBool uset_isFrozen(const USet *set)
Determines whether the set has been frozen (made immutable) or not.
U_CAPI int32_t uset_getRangeCount(const USet *set)
U_CAPI USet * uset_clone(const USet *set)
Returns a copy of this object.
U_CAPI UBool uset_containsAllCodePoints(const USet *set, const UChar *str, int32_t strLen)
Returns true if this set contains all the characters of the given string.
U_CAPI USet * uset_openPattern(const UChar *pattern, int32_t patternLength, UErrorCode *ec)
Creates a set from the given pattern.
U_CAPI int32_t uset_serialize(const USet *set, uint16_t *dest, int32_t destCapacity, UErrorCode *pErrorCode)
Serializes this set into an array of 16-bit integers.
U_CAPI int32_t uset_indexOf(const USet *set, UChar32 c)
Returns the index of the given character within this set, where the set is ordered by ascending code ...
U_CAPI void uset_freeze(USet *set)
Freeze the set (make it immutable).
U_CAPI int32_t uset_spanUTF8(const USet *set, const char *s, int32_t length, USetSpanCondition spanCondition)
Returns the length of the initial substring of the input string which consists only of characters and...
U_CAPI const UChar * uset_getString(const USet *set, int32_t index, int32_t *pLength)
Returns the index-th string (empty or multi-character) in the set.
struct USerializedSet USerializedSet
A serialized form of a Unicode set.
U_CAPI void uset_retain(USet *set, UChar32 start, UChar32 end)
Retain only the elements in this set that are contained in the specified range.
U_CAPI int32_t uset_toPattern(const USet *set, UChar *result, int32_t resultCapacity, UBool escapeUnprintable, UErrorCode *ec)
Returns a string representation of this set.
U_CAPI int32_t uset_spanBackUTF8(const USet *set, const char *s, int32_t length, USetSpanCondition spanCondition)
Returns the start of the trailing substring of the input string which consists only of characters and...
U_CAPI void uset_complement(USet *set)
This is equivalent to uset_complementRange(set, 0, 0x10FFFF).
@ USET_SERIALIZED_STATIC_ARRAY_CAPACITY
Capacity of USerializedSet::staticArray.
U_CAPI UBool uset_containsSome(const USet *set1, const USet *set2)
Returns true if set1 contains some of the characters and strings of set2.
U_CAPI int32_t uset_spanBack(const USet *set, const UChar *s, int32_t length, USetSpanCondition spanCondition)
Returns the start of the trailing substring of the input string which consists only of characters and...
U_CAPI UBool uset_containsNone(const USet *set1, const USet *set2)
Returns true if set1 contains none of the characters and strings of set2.
U_CAPI void uset_retainAllCodePoints(USet *set, const UChar *str, int32_t length)
Retains EACH of the characters in this string.
U_CAPI UBool uset_contains(const USet *set, UChar32 c)
Returns true if the given USet contains the given character.
U_CAPI int32_t uset_applyPattern(USet *set, const UChar *pattern, int32_t patternLength, uint32_t options, UErrorCode *status)
Modifies the set to represent the set specified by the given pattern.
U_CAPI void uset_complementAll(USet *set, const USet *complement)
Complements in this set all elements contained in the specified set.
U_CAPI void uset_applyPropertyAlias(USet *set, const UChar *prop, int32_t propLength, const UChar *value, int32_t valueLength, UErrorCode *ec)
Modifies the set to contain those code points which have the given value for the given property.
U_CAPI void uset_close(USet *set)
Disposes of the storage used by a USet object.
struct USet USet
USet is the C API type corresponding to C++ class UnicodeSet.
U_CAPI USet * uset_openPatternOptions(const UChar *pattern, int32_t patternLength, uint32_t options, UErrorCode *ec)
Creates a set from the given pattern.
U_CAPI USet * uset_openEmpty(void)
Create an empty USet object.
U_CAPI void uset_addAll(USet *set, const USet *additionalSet)
Adds all of the elements in the specified set to this set if they're not already present.
U_CAPI void uset_removeString(USet *set, const UChar *str, int32_t strLen)
Removes the given string to the given USet.
U_CAPI void uset_compact(USet *set)
Reallocate this objects internal structures to take up the least possible space, without changing thi...
U_CAPI int32_t uset_getItem(const USet *set, int32_t itemIndex, UChar32 *start, UChar32 *end, UChar *str, int32_t strCapacity, UErrorCode *ec)
Returns an item of this set.
U_CAPI void uset_closeOver(USet *set, int32_t attributes)
Close this set over the given attribute.
U_CAPI void uset_addAllCodePoints(USet *set, const UChar *str, int32_t strLen)
Adds each of the characters in this string to the set.
U_CAPI USet * uset_cloneAsThawed(const USet *set)
Clone the set and make the clone mutable.
U_CAPI void uset_retainString(USet *set, const UChar *str, int32_t length)
Retains only the specified string from this set if it is present.
U_CAPI void uset_removeAll(USet *set, const USet *removeSet)
Removes from this set all of its elements that are contained in the specified set.
U_CAPI UBool uset_containsRange(const USet *set, UChar32 start, UChar32 end)
Returns true if the given USet contains all characters c where start <= c && c <= end.
U_CAPI UBool uset_getSerializedRange(const USerializedSet *set, int32_t rangeIndex, UChar32 *pStart, UChar32 *pEnd)
Returns a range of characters contained in the given serialized set.
C API: 16-bit Unicode handling macros.
#define U16_TRAIL(supplementary)
Get the trail surrogate (0xdc00..0xdfff) for a supplementary code point (0x10000.....
#define U16_LEAD(supplementary)
Get the lead surrogate (0xd800..0xdbff) for a supplementary code point (0x10000..0x10ffff).
Basic definitions for ICU, for both C and C++ APIs.
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
@ U_ZERO_ERROR
No error, no warning.
#define U_SUCCESS(x)
Does the error code indicate success?
