ICU 60.1  60.1
uniset.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 * Copyright (C) 1999-2016, International Business Machines Corporation
6 * and others. All Rights Reserved.
7 ***************************************************************************
8 * Date Name Description
9 * 10/20/99 alan Creation.
10 ***************************************************************************
11 */
12 
13 #ifndef UNICODESET_H
14 #define UNICODESET_H
15 
16 #include "unicode/unifilt.h"
17 #include "unicode/unistr.h"
18 #include "unicode/uset.h"
19 
26 
27 // Forward Declarations.
28 void U_CALLCONV UnicodeSet_initInclusion(int32_t src, UErrorCode &status);
30 class BMPSet;
31 class ParsePosition;
32 class RBBIRuleScanner;
33 class SymbolTable;
34 class UnicodeSetStringSpan;
35 class UVector;
36 class RuleCharacterIterator;
37 
278 class U_COMMON_API UnicodeSet U_FINAL : public UnicodeFilter {
279 
280  int32_t len; // length of list used; 0 <= len <= capacity
281  int32_t capacity; // capacity of list
282  UChar32* list; // MUST be terminated with HIGH
283  BMPSet *bmpSet; // The set is frozen iff either bmpSet or stringSpan is not NULL.
284  UChar32* buffer; // internal buffer, may be NULL
285  int32_t bufferCapacity; // capacity of buffer
286  int32_t patLen;
287 
297  char16_t *pat;
298  UVector* strings; // maintained in sorted order
299  UnicodeSetStringSpan *stringSpan;
300 
301 private:
302  enum { // constants
303  kIsBogus = 1 // This set is bogus (i.e. not valid)
304  };
305  uint8_t fFlags; // Bit flag (see constants above)
306 public:
316  inline UBool isBogus(void) const;
317 
334  void setToBogus();
335 
336 public:
337 
338  enum {
343  MIN_VALUE = 0,
344 
349  MAX_VALUE = 0x10ffff
350  };
351 
352  //----------------------------------------------------------------
353  // Constructors &c
354  //----------------------------------------------------------------
355 
356 public:
357 
362  UnicodeSet();
363 
372  UnicodeSet(UChar32 start, UChar32 end);
373 
374 #ifndef U_HIDE_INTERNAL_API
375 
379  kSerialized /* result of serialize() */
380  };
381 
392  UnicodeSet(const uint16_t buffer[], int32_t bufferLen,
393  ESerialization serialization, UErrorCode &status);
394 #endif /* U_HIDE_INTERNAL_API */
395 
404  UnicodeSet(const UnicodeString& pattern,
405  UErrorCode& status);
406 
407 #ifndef U_HIDE_INTERNAL_API
408 
420  UnicodeSet(const UnicodeString& pattern,
421  uint32_t options,
422  const SymbolTable* symbols,
423  UErrorCode& status);
424 #endif /* U_HIDE_INTERNAL_API */
425 
439  UnicodeSet(const UnicodeString& pattern, ParsePosition& pos,
440  uint32_t options,
441  const SymbolTable* symbols,
442  UErrorCode& status);
443 
448  UnicodeSet(const UnicodeSet& o);
449 
454  virtual ~UnicodeSet();
455 
461  UnicodeSet& operator=(const UnicodeSet& o);
462 
474  virtual UBool operator==(const UnicodeSet& o) const;
475 
481  UBool operator!=(const UnicodeSet& o) const;
482 
492  virtual UnicodeFunctor* clone() const;
493 
501  virtual int32_t hashCode(void) const;
502 
511  inline static UnicodeSet *fromUSet(USet *uset);
512 
521  inline static const UnicodeSet *fromUSet(const USet *uset);
522 
530  inline USet *toUSet();
531 
532 
540  inline const USet * toUSet() const;
541 
542 
543  //----------------------------------------------------------------
544  // Freezable API
545  //----------------------------------------------------------------
546 
555  inline UBool isFrozen() const;
556 
570  UnicodeFunctor *freeze();
571 
580  UnicodeFunctor *cloneAsThawed() const;
581 
582  //----------------------------------------------------------------
583  // Public API
584  //----------------------------------------------------------------
585 
596  UnicodeSet& set(UChar32 start, UChar32 end);
597 
603  static UBool resemblesPattern(const UnicodeString& pattern,
604  int32_t pos);
605 
618  UnicodeSet& applyPattern(const UnicodeString& pattern,
619  UErrorCode& status);
620 
621 #ifndef U_HIDE_INTERNAL_API
622 
638  UnicodeSet& applyPattern(const UnicodeString& pattern,
639  uint32_t options,
640  const SymbolTable* symbols,
641  UErrorCode& status);
642 #endif /* U_HIDE_INTERNAL_API */
643 
675  UnicodeSet& applyPattern(const UnicodeString& pattern,
676  ParsePosition& pos,
677  uint32_t options,
678  const SymbolTable* symbols,
679  UErrorCode& status);
680 
694  virtual UnicodeString& toPattern(UnicodeString& result,
695  UBool escapeUnprintable = FALSE) const;
696 
719  UnicodeSet& applyIntPropertyValue(UProperty prop,
720  int32_t value,
721  UErrorCode& ec);
722 
752  UnicodeSet& applyPropertyAlias(const UnicodeString& prop,
753  const UnicodeString& value,
754  UErrorCode& ec);
755 
764  virtual int32_t size(void) const;
765 
772  virtual UBool isEmpty(void) const;
773 
781  virtual UBool contains(UChar32 c) const;
782 
791  virtual UBool contains(UChar32 start, UChar32 end) const;
792 
800  UBool contains(const UnicodeString& s) const;
801 
809  virtual UBool containsAll(const UnicodeSet& c) const;
810 
818  UBool containsAll(const UnicodeString& s) const;
819 
828  UBool containsNone(UChar32 start, UChar32 end) const;
829 
837  UBool containsNone(const UnicodeSet& c) const;
838 
846  UBool containsNone(const UnicodeString& s) const;
847 
856  inline UBool containsSome(UChar32 start, UChar32 end) const;
857 
865  inline UBool containsSome(const UnicodeSet& s) const;
866 
874  inline UBool containsSome(const UnicodeString& s) const;
875 
894  int32_t span(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const;
895 
908  inline int32_t span(const UnicodeString &s, int32_t start, USetSpanCondition spanCondition) const;
909 
927  int32_t spanBack(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const;
928 
942  inline int32_t spanBack(const UnicodeString &s, int32_t limit, USetSpanCondition spanCondition) const;
943 
962  int32_t spanUTF8(const char *s, int32_t length, USetSpanCondition spanCondition) const;
963 
981  int32_t spanBackUTF8(const char *s, int32_t length, USetSpanCondition spanCondition) const;
982 
987  virtual UMatchDegree matches(const Replaceable& text,
988  int32_t& offset,
989  int32_t limit,
990  UBool incremental);
991 
992 private:
1015  static int32_t matchRest(const Replaceable& text,
1016  int32_t start, int32_t limit,
1017  const UnicodeString& s);
1018 
1028  int32_t findCodePoint(UChar32 c) const;
1029 
1030 public:
1031 
1039  virtual void addMatchSetTo(UnicodeSet& toUnionTo) const;
1040 
1049  int32_t indexOf(UChar32 c) const;
1050 
1060  UChar32 charAt(int32_t index) const;
1061 
1076  virtual UnicodeSet& add(UChar32 start, UChar32 end);
1077 
1085  UnicodeSet& add(UChar32 c);
1086 
1098  UnicodeSet& add(const UnicodeString& s);
1099 
1100  private:
1106  static int32_t getSingleCP(const UnicodeString& s);
1107 
1108  void _add(const UnicodeString& s);
1109 
1110  public:
1119  UnicodeSet& addAll(const UnicodeString& s);
1120 
1129  UnicodeSet& retainAll(const UnicodeString& s);
1130 
1139  UnicodeSet& complementAll(const UnicodeString& s);
1140 
1149  UnicodeSet& removeAll(const UnicodeString& s);
1150 
1159  static UnicodeSet* U_EXPORT2 createFrom(const UnicodeString& s);
1160 
1161 
1169  static UnicodeSet* U_EXPORT2 createFromAll(const UnicodeString& s);
1170 
1184  virtual UnicodeSet& retain(UChar32 start, UChar32 end);
1185 
1186 
1192  UnicodeSet& retain(UChar32 c);
1193 
1207  virtual UnicodeSet& remove(UChar32 start, UChar32 end);
1208 
1216  UnicodeSet& remove(UChar32 c);
1217 
1227  UnicodeSet& remove(const UnicodeString& s);
1228 
1236  virtual UnicodeSet& complement(void);
1237 
1252  virtual UnicodeSet& complement(UChar32 start, UChar32 end);
1253 
1261  UnicodeSet& complement(UChar32 c);
1262 
1273  UnicodeSet& complement(const UnicodeString& s);
1274 
1287  virtual UnicodeSet& addAll(const UnicodeSet& c);
1288 
1300  virtual UnicodeSet& retainAll(const UnicodeSet& c);
1301 
1313  virtual UnicodeSet& removeAll(const UnicodeSet& c);
1314 
1325  virtual UnicodeSet& complementAll(const UnicodeSet& c);
1326 
1333  virtual UnicodeSet& clear(void);
1334 
1360  UnicodeSet& closeOver(int32_t attribute);
1361 
1368  virtual UnicodeSet &removeAllStrings();
1369 
1377  virtual int32_t getRangeCount(void) const;
1378 
1386  virtual UChar32 getRangeStart(int32_t index) const;
1387 
1395  virtual UChar32 getRangeEnd(int32_t index) const;
1396 
1445  int32_t serialize(uint16_t *dest, int32_t destCapacity, UErrorCode& ec) const;
1446 
1453  virtual UnicodeSet& compact();
1454 
1466  static UClassID U_EXPORT2 getStaticClassID(void);
1467 
1476  virtual UClassID getDynamicClassID(void) const;
1477 
1478 private:
1479 
1480  // Private API for the USet API
1481 
1482  friend class USetAccess;
1483 
1484  int32_t getStringCount() const;
1485 
1486  const UnicodeString* getString(int32_t index) const;
1487 
1488  //----------------------------------------------------------------
1489  // RuleBasedTransliterator support
1490  //----------------------------------------------------------------
1491 
1492 private:
1493 
1499  virtual UBool matchesIndexValue(uint8_t v) const;
1500 
1501 private:
1502  friend class RBBIRuleScanner;
1503 
1504  //----------------------------------------------------------------
1505  // Implementation: Clone as thawed (see ICU4J Freezable)
1506  //----------------------------------------------------------------
1507 
1508  UnicodeSet(const UnicodeSet& o, UBool /* asThawed */);
1509 
1510  //----------------------------------------------------------------
1511  // Implementation: Pattern parsing
1512  //----------------------------------------------------------------
1513 
1514  void applyPatternIgnoreSpace(const UnicodeString& pattern,
1515  ParsePosition& pos,
1516  const SymbolTable* symbols,
1517  UErrorCode& status);
1518 
1519  void applyPattern(RuleCharacterIterator& chars,
1520  const SymbolTable* symbols,
1521  UnicodeString& rebuiltPat,
1522  uint32_t options,
1523  UnicodeSet& (UnicodeSet::*caseClosure)(int32_t attribute),
1524  UErrorCode& ec);
1525 
1526  //----------------------------------------------------------------
1527  // Implementation: Utility methods
1528  //----------------------------------------------------------------
1529 
1530  void ensureCapacity(int32_t newLen, UErrorCode& ec);
1531 
1532  void ensureBufferCapacity(int32_t newLen, UErrorCode& ec);
1533 
1534  void swapBuffers(void);
1535 
1536  UBool allocateStrings(UErrorCode &status);
1537 
1538  UnicodeString& _toPattern(UnicodeString& result,
1539  UBool escapeUnprintable) const;
1540 
1541  UnicodeString& _generatePattern(UnicodeString& result,
1542  UBool escapeUnprintable) const;
1543 
1544  static void _appendToPat(UnicodeString& buf, const UnicodeString& s, UBool escapeUnprintable);
1545 
1546  static void _appendToPat(UnicodeString& buf, UChar32 c, UBool escapeUnprintable);
1547 
1548  //----------------------------------------------------------------
1549  // Implementation: Fundamental operators
1550  //----------------------------------------------------------------
1551 
1552  void exclusiveOr(const UChar32* other, int32_t otherLen, int8_t polarity);
1553 
1554  void add(const UChar32* other, int32_t otherLen, int8_t polarity);
1555 
1556  void retain(const UChar32* other, int32_t otherLen, int8_t polarity);
1557 
1563  static UBool resemblesPropertyPattern(const UnicodeString& pattern,
1564  int32_t pos);
1565 
1566  static UBool resemblesPropertyPattern(RuleCharacterIterator& chars,
1567  int32_t iterOpts);
1568 
1608  UnicodeSet& applyPropertyPattern(const UnicodeString& pattern,
1609  ParsePosition& ppos,
1610  UErrorCode &ec);
1611 
1612  void applyPropertyPattern(RuleCharacterIterator& chars,
1613  UnicodeString& rebuiltPat,
1614  UErrorCode& ec);
1615 
1616  friend void U_CALLCONV UnicodeSet_initInclusion(int32_t src, UErrorCode &status);
1617  static const UnicodeSet* getInclusions(int32_t src, UErrorCode &status);
1618 
1623  typedef UBool (*Filter)(UChar32 codePoint, void* context);
1624 
1634  void applyFilter(Filter filter,
1635  void* context,
1636  int32_t src,
1637  UErrorCode &status);
1638 
1642  void setPattern(const UnicodeString& newPat);
1646  void releasePattern();
1647 
1648  friend class UnicodeSetIterator;
1649 };
1650 
1651 
1652 
1653 inline UBool UnicodeSet::operator!=(const UnicodeSet& o) const {
1654  return !operator==(o);
1655 }
1656 
1657 inline UBool UnicodeSet::isFrozen() const {
1658  return (UBool)(bmpSet!=NULL || stringSpan!=NULL);
1659 }
1660 
1661 inline UBool UnicodeSet::containsSome(UChar32 start, UChar32 end) const {
1662  return !containsNone(start, end);
1663 }
1664 
1666  return !containsNone(s);
1667 }
1668 
1670  return !containsNone(s);
1671 }
1672 
1673 inline UBool UnicodeSet::isBogus() const {
1674  return (UBool)(fFlags & kIsBogus);
1675 }
1676 
1678  return reinterpret_cast<UnicodeSet *>(uset);
1679 }
1680 
1681 inline const UnicodeSet *UnicodeSet::fromUSet(const USet *uset) {
1682  return reinterpret_cast<const UnicodeSet *>(uset);
1683 }
1684 
1686  return reinterpret_cast<USet *>(this);
1687 }
1688 
1689 inline const USet *UnicodeSet::toUSet() const {
1690  return reinterpret_cast<const USet *>(this);
1691 }
1692 
1693 inline int32_t UnicodeSet::span(const UnicodeString &s, int32_t start, USetSpanCondition spanCondition) const {
1694  int32_t sLength=s.length();
1695  if(start<0) {
1696  start=0;
1697  } else if(start>sLength) {
1698  start=sLength;
1699  }
1700  return start+span(s.getBuffer()+start, sLength-start, spanCondition);
1701 }
1702 
1703 inline int32_t UnicodeSet::spanBack(const UnicodeString &s, int32_t limit, USetSpanCondition spanCondition) const {
1704  int32_t sLength=s.length();
1705  if(limit<0) {
1706  limit=0;
1707  } else if(limit>sLength) {
1708  limit=sLength;
1709  }
1710  return spanBack(s.getBuffer(), limit, spanCondition);
1711 }
1712 
1714 
1715 #endif
static UClassID getStaticClassID()
ICU "poor man&#39;s RTTI", returns a UClassID for this class.
int32_t spanBack(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const
Returns the start of the trailing substring of the input string which consists only of characters and...
UMatchDegree
Constants returned by UnicodeMatcher::matches() indicating the degree of match.
Definition: unimatch.h:32
C++ API: Unicode String.
U_EXPORT UBool operator==(const StringPiece &x, const StringPiece &y)
Global operator == for StringPiece.
UnicodeSetIterator iterates over the contents of a UnicodeSet.
Definition: usetiter.h:63
UBool isBogus(void) const
Determine if this object contains a valid set.
Definition: uniset.h:1673
#define U_CALLCONV
Similar to U_CDECL_BEGIN/U_CDECL_END, this qualifier is necessary in callback function typedefs to ma...
Definition: platform.h:848
UBool operator!=(const UnicodeSet &o) const
Compares the specified object with this set for equality.
Definition: uniset.h:1653
void * UClassID
UClassID is used to identify classes without using the compiler&#39;s RTTI.
Definition: uobject.h:93
static UnicodeSet * fromUSet(USet *uset)
Get a UnicodeSet pointer from a USet.
Definition: uniset.h:1677
virtual UBool matchesIndexValue(uint8_t v) const =0
Returns TRUE if this matcher will match a character c, where c & 0xFF == v, at offset, in the forward direction (with limit > offset).
C API: Unicode Set.
An interface that defines both lookup protocol and parsing of symbolic names.
Definition: symtable.h:56
virtual UClassID getDynamicClassID(void) const =0
Returns a unique class ID polymorphically.
virtual UnicodeFunctor * clone() const =0
Return a copy of this object.
Replaceable is an abstract base class representing a string of characters that supports the replaceme...
Definition: rep.h:73
#define U_NAMESPACE_BEGIN
This is used to begin a declaration of a public ICU C++ API.
Definition: uversion.h:131
UnicodeFilter defines a protocol for selecting a subset of the full range (U+0000 to U+10FFFF) of Uni...
Definition: unifilt.h:61
virtual void addMatchSetTo(UnicodeSet &toUnionTo) const =0
Union the set of all characters that may be matched by this object into the given set...
UBool operator!=(const StringPiece &x, const StringPiece &y)
Global operator != for StringPiece.
Definition: stringpiece.h:218
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:396
#define NULL
Define NULL if necessary, to nullptr for C++ and to ((void *)0) for C.
Definition: utypes.h:188
UnicodeFunctor is an abstract base class for objects that perform match and/or replace operations on ...
Definition: unifunct.h:35
virtual UMatchDegree matches(const Replaceable &text, int32_t &offset, int32_t limit, UBool incremental)
Implement UnicodeMatcher API.
A mutable set of Unicode characters and multicharacter strings.
Definition: uniset.h:278
USetSpanCondition
Argument values for whether span() and similar functions continue while the current character is cont...
Definition: uset.h:152
#define U_NAMESPACE_END
This is used to end a declaration of a public ICU C++ API.
Definition: uversion.h:132
struct USet USet
Definition: ucnv.h:69
UProperty
Selection constants for Unicode properties.
Definition: uchar.h:165
UErrorCode
Error code to replace exception handling, so that the code is compatible with all C++ compilers...
Definition: utypes.h:396
int32_t length(void) const
Return the length of the UnicodeString object.
Definition: unistr.h:3942
ParsePosition is a simple class used by Format and its subclasses to keep track of the current positi...
Definition: parsepos.h:49
char16_t * getBuffer(int32_t minCapacity)
Get a read/write pointer to the internal buffer.
void UnicodeSet_initInclusion(int32_t src, UErrorCode &status)
virtual UnicodeString & toPattern(UnicodeString &result, UBool escapeUnprintable=FALSE) const =0
Returns a string representation of this matcher.
UBool containsSome(UChar32 start, UChar32 end) const
Returns true if this set contains one or more of the characters in the given range.
Definition: uniset.h:1661
virtual UBool contains(UChar32 c) const =0
Returns true for characters that are in the selected subset.
#define FALSE
The FALSE value of a UBool.
Definition: umachine.h:244
#define U_COMMON_API
Set to export library symbols from inside the common library, and to import them from outside...
Definition: utypes.h:359
UnicodeString is a string class that stores Unicode characters directly and provides similar function...
Definition: unistr.h:286
UBool isFrozen() const
Determines whether the set has been frozen (made immutable) or not.
Definition: uniset.h:1657
USet * toUSet()
Produce a USet * pointer for this UnicodeSet.
Definition: uniset.h:1685
C++ API: Unicode Filter.
int32_t span(const char16_t *s, int32_t length, USetSpanCondition spanCondition) const
Returns the length of the initial substring of the input string which consists only of characters and...
int8_t UBool
The ICU boolean type.
Definition: umachine.h:236