ICU 75.1  75.1
uset.h
Go to the documentation of this file.
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 *******************************************************************************
5 *
6 * Copyright (C) 2002-2014, International Business Machines
7 * Corporation and others. All Rights Reserved.
8 *
9 *******************************************************************************
10 * file name: uset.h
11 * encoding: UTF-8
12 * tab size: 8 (not used)
13 * indentation:4
14 *
15 * created on: 2002mar07
16 * created by: Markus W. Scherer
17 *
18 * C version of UnicodeSet.
19 */
20 
21 
29 #ifndef __USET_H__
30 #define __USET_H__
31 
32 #include "unicode/utypes.h"
33 #include "unicode/uchar.h"
34 
35 #if U_SHOW_CPLUSPLUS_API
36 #include "unicode/localpointer.h"
37 #endif // U_SHOW_CPLUSPLUS_API
38 
39 #ifndef USET_DEFINED
40 
41 #ifndef U_IN_DOXYGEN
42 #define USET_DEFINED
43 #endif
50 typedef struct USet USet;
51 #endif
52 
64 enum {
70 
98 
111 
125 };
126 
182 typedef enum USetSpanCondition {
231 #ifndef U_HIDE_DEPRECATED_API
237 #endif // U_HIDE_DEPRECATED_API
239 
240 enum {
248 };
249 
255 typedef struct USerializedSet {
260  const uint16_t *array;
265  int32_t bmpLength;
270  int32_t length;
277 
278 /*********************************************************************
279  * USet API
280  *********************************************************************/
281 
289 U_CAPI USet* U_EXPORT2
291 
302 U_CAPI USet* U_EXPORT2
304 
314 U_CAPI USet* U_EXPORT2
315 uset_openPattern(const UChar* pattern, int32_t patternLength,
316  UErrorCode* ec);
317 
331 U_CAPI USet* U_EXPORT2
332 uset_openPatternOptions(const UChar* pattern, int32_t patternLength,
333  uint32_t options,
334  UErrorCode* ec);
335 
342 U_CAPI void U_EXPORT2
344 
345 #if U_SHOW_CPLUSPLUS_API
346 
347 U_NAMESPACE_BEGIN
348 
359 
360 U_NAMESPACE_END
361 
362 #endif
363 
373 U_CAPI USet * U_EXPORT2
374 uset_clone(const USet *set);
375 
385 U_CAPI UBool U_EXPORT2
386 uset_isFrozen(const USet *set);
387 
402 U_CAPI void U_EXPORT2
404 
415 U_CAPI USet * U_EXPORT2
417 
427 U_CAPI void U_EXPORT2
429  UChar32 start, UChar32 end);
430 
455 U_CAPI int32_t U_EXPORT2
457  const UChar *pattern, int32_t patternLength,
458  uint32_t options,
459  UErrorCode *status);
460 
483 U_CAPI void U_EXPORT2
485  UProperty prop, int32_t value, UErrorCode* ec);
486 
522 U_CAPI void U_EXPORT2
524  const UChar *prop, int32_t propLength,
525  const UChar *value, int32_t valueLength,
526  UErrorCode* ec);
527 
537 U_CAPI UBool U_EXPORT2
538 uset_resemblesPattern(const UChar *pattern, int32_t patternLength,
539  int32_t pos);
540 
556 U_CAPI int32_t U_EXPORT2
557 uset_toPattern(const USet* set,
558  UChar* result, int32_t resultCapacity,
559  UBool escapeUnprintable,
560  UErrorCode* ec);
561 
570 U_CAPI void U_EXPORT2
572 
585 U_CAPI void U_EXPORT2
586 uset_addAll(USet* set, const USet *additionalSet);
587 
597 U_CAPI void U_EXPORT2
598 uset_addRange(USet* set, UChar32 start, UChar32 end);
599 
609 U_CAPI void U_EXPORT2
610 uset_addString(USet* set, const UChar* str, int32_t strLen);
611 
621 U_CAPI void U_EXPORT2
622 uset_addAllCodePoints(USet* set, const UChar *str, int32_t strLen);
623 
632 U_CAPI void U_EXPORT2
634 
644 U_CAPI void U_EXPORT2
646 
656 U_CAPI void U_EXPORT2
657 uset_removeString(USet* set, const UChar* str, int32_t strLen);
658 
668 U_CAPI void U_EXPORT2
669 uset_removeAllCodePoints(USet *set, const UChar *str, int32_t length);
670 
682 U_CAPI void U_EXPORT2
683 uset_removeAll(USet* set, const USet* removeSet);
684 
697 U_CAPI void U_EXPORT2
698 uset_retain(USet* set, UChar32 start, UChar32 end);
699 
711 U_CAPI void U_EXPORT2
712 uset_retainString(USet *set, const UChar *str, int32_t length);
713 
723 U_CAPI void U_EXPORT2
724 uset_retainAllCodePoints(USet *set, const UChar *str, int32_t length);
725 
738 U_CAPI void U_EXPORT2
739 uset_retainAll(USet* set, const USet* retain);
740 
749 U_CAPI void U_EXPORT2
751 
765 U_CAPI void U_EXPORT2
767 
781 U_CAPI void U_EXPORT2
783 
794 U_CAPI void U_EXPORT2
795 uset_complementString(USet *set, const UChar *str, int32_t length);
796 
806 U_CAPI void U_EXPORT2
807 uset_complementAllCodePoints(USet *set, const UChar *str, int32_t length);
808 
820 U_CAPI void U_EXPORT2
821 uset_complementAll(USet* set, const USet* complement);
822 
830 U_CAPI void U_EXPORT2
832 
861 U_CAPI void U_EXPORT2
862 uset_closeOver(USet* set, int32_t attributes);
863 
870 U_CAPI void U_EXPORT2
872 
880 U_CAPI UBool U_EXPORT2
881 uset_isEmpty(const USet* set);
882 
888 U_CAPI UBool U_EXPORT2
889 uset_hasStrings(const USet *set);
890 
899 U_CAPI UBool U_EXPORT2
900 uset_contains(const USet* set, UChar32 c);
901 
911 U_CAPI UBool U_EXPORT2
912 uset_containsRange(const USet* set, UChar32 start, UChar32 end);
913 
922 U_CAPI UBool U_EXPORT2
923 uset_containsString(const USet* set, const UChar* str, int32_t strLen);
924 
935 U_CAPI int32_t U_EXPORT2
936 uset_indexOf(const USet* set, UChar32 c);
937 
953 U_CAPI UChar32 U_EXPORT2
954 uset_charAt(const USet* set, int32_t charIndex);
955 
969 U_CAPI int32_t U_EXPORT2
970 uset_size(const USet* set);
971 
980 U_CAPI int32_t U_EXPORT2
982 
991 U_CAPI int32_t U_EXPORT2
993 
1022 U_CAPI int32_t U_EXPORT2
1023 uset_getItem(const USet* set, int32_t itemIndex,
1024  UChar32* start, UChar32* end,
1025  UChar* str, int32_t strCapacity,
1026  UErrorCode* ec);
1027 
1036 U_CAPI UBool U_EXPORT2
1037 uset_containsAll(const USet* set1, const USet* set2);
1038 
1049 U_CAPI UBool U_EXPORT2
1050 uset_containsAllCodePoints(const USet* set, const UChar *str, int32_t strLen);
1051 
1060 U_CAPI UBool U_EXPORT2
1061 uset_containsNone(const USet* set1, const USet* set2);
1062 
1071 U_CAPI UBool U_EXPORT2
1072 uset_containsSome(const USet* set1, const USet* set2);
1073 
1093 U_CAPI int32_t U_EXPORT2
1094 uset_span(const USet *set, const UChar *s, int32_t length, USetSpanCondition spanCondition);
1095 
1114 U_CAPI int32_t U_EXPORT2
1115 uset_spanBack(const USet *set, const UChar *s, int32_t length, USetSpanCondition spanCondition);
1116 
1136 U_CAPI int32_t U_EXPORT2
1137 uset_spanUTF8(const USet *set, const char *s, int32_t length, USetSpanCondition spanCondition);
1138 
1157 U_CAPI int32_t U_EXPORT2
1158 uset_spanBackUTF8(const USet *set, const char *s, int32_t length, USetSpanCondition spanCondition);
1159 
1168 U_CAPI UBool U_EXPORT2
1169 uset_equals(const USet* set1, const USet* set2);
1170 
1171 /*********************************************************************
1172  * Serialized set API
1173  *********************************************************************/
1174 
1224 U_CAPI int32_t U_EXPORT2
1225 uset_serialize(const USet* set, uint16_t* dest, int32_t destCapacity, UErrorCode* pErrorCode);
1226 
1235 U_CAPI UBool U_EXPORT2
1236 uset_getSerializedSet(USerializedSet* fillSet, const uint16_t* src, int32_t srcLength);
1237 
1245 U_CAPI void U_EXPORT2
1247 
1256 U_CAPI UBool U_EXPORT2
1258 
1268 U_CAPI int32_t U_EXPORT2
1270 
1284 U_CAPI UBool U_EXPORT2
1285 uset_getSerializedRange(const USerializedSet* set, int32_t rangeIndex,
1286  UChar32* pStart, UChar32* pEnd);
1287 
1288 #endif
"Smart pointer" class, closes a USet via uset_close().
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.
Definition: localpointer.h:550
A serialized form of a Unicode set.
Definition: uset.h:255
uint16_t staticArray[USET_SERIALIZED_STATIC_ARRAY_CAPACITY]
A small buffer for the array to reduce memory allocations.
Definition: uset.h:275
int32_t bmpLength
The length of the array that contains BMP characters.
Definition: uset.h:265
const uint16_t * array
The serialized Unicode Set.
Definition: uset.h:260
int32_t length
The total length of the array.
Definition: uset.h:270
C API: Unicode Properties.
UProperty
Selection constants for Unicode properties.
Definition: uchar.h:196
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:427
int8_t UBool
The ICU boolean type, a signed-byte integer.
Definition: umachine.h:247
#define U_CAPI
This is used to declare a function as a public ICU C API.
Definition: umachine.h:110
char16_t UChar
The base type for UTF-16 code units and pointers.
Definition: umachine.h:378
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 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.
Definition: uset.h:69
@ USET_ADD_CASE_MAPPINGS
Adds all case mappings for each element in the set.
Definition: uset.h:110
@ USET_CASE_INSENSITIVE
Enable case insensitive matching.
Definition: uset.h:97
@ USET_SIMPLE_CASE_INSENSITIVE
Enable case insensitive matching.
Definition: uset.h:124
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...
Definition: uset.h:182
@ USET_SPAN_NOT_CONTAINED
Continues a span() while there is no set element at the current position.
Definition: uset.h:195
@ USET_SPAN_CONTAINED
Spans the longest substring that is a concatenation of set elements (characters or strings).
Definition: uset.h:210
@ USET_SPAN_CONDITION_COUNT
One more than the last span condition.
Definition: uset.h:236
@ USET_SPAN_SIMPLE
Continues a span() while there is a set element at the current position.
Definition: uset.h:230
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...
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.
Definition: uset.h:247
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.
Definition: uset.h:50
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.
Basic definitions for ICU, for both C and C++ APIs.
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition: utypes.h:415