33 #if U_SHOW_CPLUSPLUS_API
55 #if !UCONFIG_NO_BREAK_ITERATION
74 char16_t *dest, int32_t destCapacity,
75 const char16_t *src, int32_t srcLength,
82 class StringCharacterIterator;
83 class UnicodeStringAppendable;
97 #define US_INV icu::UnicodeString::kInvariant
116 #if !U_CHAR16_IS_TYPEDEF
117 # define UNICODE_STRING(cs, _length) icu::UnicodeString(true, u ## cs, _length)
119 # define UNICODE_STRING(cs, _length) icu::UnicodeString(true, (const char16_t*)u ## cs, _length)
135 #define UNICODE_STRING_SIMPLE(cs) UNICODE_STRING(cs, -1)
144 #ifndef UNISTR_FROM_CHAR_EXPLICIT
145 # if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION)
147 # define UNISTR_FROM_CHAR_EXPLICIT explicit
150 # define UNISTR_FROM_CHAR_EXPLICIT
164 #ifndef UNISTR_FROM_STRING_EXPLICIT
165 # if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION)
167 # define UNISTR_FROM_STRING_EXPLICIT explicit
170 # define UNISTR_FROM_STRING_EXPLICIT
207 #ifndef UNISTR_OBJECT_SIZE
208 # define UNISTR_OBJECT_SIZE 64
403 inline int8_t compare(int32_t start,
424 inline int8_t compare(int32_t start,
428 int32_t srcLength)
const;
443 int32_t srcLength)
const;
459 inline int8_t compare(int32_t start,
461 const char16_t *srcChars)
const;
480 inline int8_t compare(int32_t start,
482 const char16_t *srcChars,
484 int32_t srcLength)
const;
503 inline int8_t compareBetween(int32_t start,
507 int32_t srcLimit)
const;
526 inline int8_t compareCodePointOrder(
const UnicodeString& text)
const;
547 inline int8_t compareCodePointOrder(int32_t start,
572 inline int8_t compareCodePointOrder(int32_t start,
576 int32_t srcLength)
const;
597 int32_t srcLength)
const;
618 inline int8_t compareCodePointOrder(int32_t start,
620 const char16_t *srcChars)
const;
643 inline int8_t compareCodePointOrder(int32_t start,
645 const char16_t *srcChars,
647 int32_t srcLength)
const;
670 inline int8_t compareCodePointOrderBetween(int32_t start,
674 int32_t srcLimit)
const;
694 inline int8_t caseCompare(
const UnicodeString& text, uint32_t options)
const;
716 inline int8_t caseCompare(int32_t start,
719 uint32_t options)
const;
743 inline int8_t caseCompare(int32_t start,
748 uint32_t options)
const;
771 uint32_t options)
const;
793 inline int8_t caseCompare(int32_t start,
795 const char16_t *srcChars,
796 uint32_t options)
const;
820 inline int8_t caseCompare(int32_t start,
822 const char16_t *srcChars,
825 uint32_t options)
const;
849 inline int8_t caseCompareBetween(int32_t start,
854 uint32_t options)
const;
877 int32_t srcLength)
const;
888 int32_t srcLength)
const;
899 inline UBool startsWith(
const char16_t *srcChars,
901 int32_t srcLength)
const;
924 int32_t srcLength)
const;
935 int32_t srcLength)
const;
947 inline UBool endsWith(
const char16_t *srcChars,
949 int32_t srcLength)
const;
974 int32_t start)
const;
989 int32_t length)
const;
1011 int32_t length)
const;
1024 inline int32_t indexOf(
const char16_t *srcChars,
1026 int32_t start)
const;
1043 int32_t length)
const;
1065 int32_t length)
const;
1074 inline int32_t indexOf(char16_t c)
const;
1084 inline int32_t indexOf(
UChar32 c)
const;
1094 inline int32_t indexOf(char16_t c,
1095 int32_t start)
const;
1106 inline int32_t indexOf(
UChar32 c,
1107 int32_t start)
const;
1119 inline int32_t indexOf(char16_t c,
1121 int32_t length)
const;
1134 inline int32_t indexOf(
UChar32 c,
1136 int32_t length)
const;
1146 inline int32_t lastIndexOf(
const UnicodeString& text)
const;
1158 int32_t start)
const;
1173 int32_t length)
const;
1195 int32_t length)
const;
1207 inline int32_t lastIndexOf(
const char16_t *srcChars,
1209 int32_t start)
const;
1226 int32_t length)
const;
1248 int32_t length)
const;
1257 inline int32_t lastIndexOf(char16_t c)
const;
1267 inline int32_t lastIndexOf(
UChar32 c)
const;
1277 inline int32_t lastIndexOf(char16_t c,
1278 int32_t start)
const;
1289 inline int32_t lastIndexOf(
UChar32 c,
1290 int32_t start)
const;
1302 inline int32_t lastIndexOf(char16_t c,
1304 int32_t length)
const;
1317 inline int32_t lastIndexOf(
UChar32 c,
1319 int32_t length)
const;
1332 inline char16_t
charAt(int32_t offset)
const;
1341 inline char16_t operator[] (int32_t offset)
const;
1460 inline void extract(int32_t start,
1463 int32_t dstStart = 0)
const;
1499 inline void extract(int32_t start,
1517 int32_t dstStart = 0)
const;
1553 int32_t startLength,
1555 int32_t targetCapacity,
1558 #if U_CHARSET_IS_UTF8 || !UCONFIG_NO_CONVERSION
1580 int32_t startLength,
1582 uint32_t targetLength)
const;
1586 #if !UCONFIG_NO_CONVERSION
1613 inline int32_t extract(int32_t start,
1614 int32_t startLength,
1616 const char* codepage =
nullptr)
const;
1648 int32_t startLength,
1650 uint32_t targetLength,
1651 const char *codepage)
const;
1728 template<
typename StringClass>
1762 inline int32_t
length()
const;
1811 inline UBool isEmpty()
const;
1822 inline int32_t getCapacity()
const;
1831 inline int32_t hashCode()
const;
1845 inline UBool isBogus()
const;
1923 friend inline void U_EXPORT2
2043 int32_t textLength);
2066 int32_t buffCapacity);
2267 const char16_t *srcChars,
2367 const char16_t *srcChars,
2480 virtual void copy(int32_t start, int32_t limit, int32_t dest)
override;
2597 char16_t padChar = 0x0020);
2611 char16_t padChar = 0x0020);
2619 inline UBool truncate(int32_t targetLength);
2682 #if !UCONFIG_NO_BREAK_ITERATION
2891 inline const char16_t *getBuffer()
const;
2984 #if !U_CHAR16_IS_TYPEDEF
2999 #if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN)
3035 int32_t textLength);
3037 #if !U_CHAR16_IS_TYPEDEF
3049 #if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN)
3069 inline UnicodeString(
const std::nullptr_t text, int32_t textLength);
3095 int32_t textLength);
3117 #if !U_CHAR16_IS_TYPEDEF
3130 #if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN)
3152 inline UnicodeString(std::nullptr_t buffer, int32_t buffLength, int32_t buffCapacity);
3154 #if U_CHARSET_IS_UTF8 || !UCONFIG_NO_CONVERSION
3189 #if !UCONFIG_NO_CONVERSION
3227 UnicodeString(
const char *codepageData, int32_t dataLength,
const char *codepage);
3251 const char *src, int32_t srcLength,
3481 toUTF8(int32_t start, int32_t len,
3482 char *target, int32_t capacity)
const;
3491 doEqualsSubstring(int32_t start,
3495 int32_t srcLength)
const;
3497 UBool doEqualsSubstring(int32_t start,
3499 const char16_t *srcChars,
3501 int32_t srcLength)
const;
3504 doCompare(int32_t start,
3508 int32_t srcLength)
const;
3510 int8_t doCompare(int32_t start,
3512 const char16_t *srcChars,
3514 int32_t srcLength)
const;
3517 doCompareCodePointOrder(int32_t start,
3521 int32_t srcLength)
const;
3523 int8_t doCompareCodePointOrder(int32_t start,
3525 const char16_t *srcChars,
3527 int32_t srcLength)
const;
3530 doCaseCompare(int32_t start,
3535 uint32_t options)
const;
3538 doCaseCompare(int32_t start,
3540 const char16_t *srcChars,
3543 uint32_t options)
const;
3545 int32_t doIndexOf(char16_t c,
3547 int32_t length)
const;
3551 int32_t length)
const;
3553 int32_t doLastIndexOf(char16_t c,
3555 int32_t length)
const;
3557 int32_t doLastIndexOf(
UChar32 c,
3559 int32_t length)
const;
3561 void doExtract(int32_t start,
3564 int32_t dstStart)
const;
3566 inline void doExtract(int32_t start,
3570 inline char16_t doCharAt(int32_t offset)
const;
3580 const char16_t *srcChars,
3585 UnicodeString& doAppend(
const char16_t *srcChars, int32_t srcStart, int32_t srcLength);
3591 int32_t doHashCode()
const;
3595 inline char16_t* getArrayStart();
3596 inline const char16_t* getArrayStart()
const;
3598 inline UBool hasShortLength()
const;
3599 inline int32_t getShortLength()
const;
3603 inline UBool isWritable()
const;
3606 inline UBool isBufferWritable()
const;
3609 inline void setZeroLength();
3610 inline void setShortLength(int32_t len);
3611 inline void setLength(int32_t len);
3612 inline void setToEmpty();
3613 inline void setArray(char16_t *array, int32_t len, int32_t capacity);
3620 UBool allocate(int32_t capacity);
3623 void releaseArray();
3635 inline void pinIndex(int32_t& start)
const;
3636 inline void pinIndices(int32_t& start,
3637 int32_t& length)
const;
3639 #if !UCONFIG_NO_CONVERSION
3642 int32_t doExtract(int32_t start, int32_t length,
3643 char *dest, int32_t destCapacity,
3657 void doCodepageCreate(
const char *codepageData,
3659 const char *codepage);
3666 doCodepageCreate(
const char *codepageData,
3684 UBool cloneArrayIfNeeded(int32_t newCapacity = -1,
3685 int32_t growCapacity = -1,
3686 UBool doCopyArray =
true,
3687 int32_t** pBufferToDelete =
nullptr,
3688 UBool forceClone =
false);
3696 caseMap(int32_t caseLocale, uint32_t options,
3704 int32_t removeRef();
3705 int32_t refCount()
const;
3715 kInvalidUChar=0xffff,
3721 kUsingStackBuffer=2,
3723 kBufferIsReadonly=8,
3726 kAllStorageFlags=0x1f,
3729 kLength1=1<<kLengthShift,
3730 kMaxShortLength=0x3ff,
3731 kLengthIsLarge=0xffe0,
3734 kShortString=kUsingStackBuffer,
3735 kLongString=kRefCounted,
3736 kReadonlyAlias=kBufferIsReadonly,
3742 union StackBufferOrFields;
3743 friend union StackBufferOrFields;
3786 union StackBufferOrFields {
3790 int16_t fLengthAndFlags;
3791 char16_t fBuffer[US_STACKBUF_SIZE];
3794 int16_t fLengthAndFlags;
3824 UnicodeString::pinIndex(int32_t& start)
const
3829 }
else if(start > length()) {
3835 UnicodeString::pinIndices(int32_t& start,
3836 int32_t& _length)
const
3839 int32_t len = length();
3842 }
else if(start > len) {
3847 }
else if(_length > (len - start)) {
3848 _length = (len - start);
3853 UnicodeString::getArrayStart() {
3854 return (fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) ?
3855 fUnion.fStackFields.fBuffer : fUnion.fFields.fArray;
3858 inline const char16_t*
3859 UnicodeString::getArrayStart()
const {
3860 return (fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) ?
3861 fUnion.fStackFields.fBuffer : fUnion.fFields.fArray;
3869 UnicodeString::UnicodeString() {
3870 fUnion.fStackFields.fLengthAndFlags=kShortString;
3873 inline UnicodeString::UnicodeString(
const std::nullptr_t ) {
3874 fUnion.fStackFields.fLengthAndFlags=kShortString;
3877 inline UnicodeString::UnicodeString(
const std::nullptr_t , int32_t ) {
3878 fUnion.fStackFields.fLengthAndFlags=kShortString;
3881 inline UnicodeString::UnicodeString(std::nullptr_t , int32_t , int32_t ) {
3882 fUnion.fStackFields.fLengthAndFlags=kShortString;
3889 UnicodeString::hasShortLength()
const {
3890 return fUnion.fFields.fLengthAndFlags>=0;
3894 UnicodeString::getShortLength()
const {
3897 return fUnion.fFields.fLengthAndFlags>>kLengthShift;
3901 UnicodeString::length()
const {
3902 return hasShortLength() ? getShortLength() : fUnion.fFields.fLength;
3906 UnicodeString::getCapacity()
const {
3907 return (fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) ?
3908 US_STACKBUF_SIZE : fUnion.fFields.fCapacity;
3912 UnicodeString::hashCode()
const
3913 {
return doHashCode(); }
3916 UnicodeString::isBogus()
const
3917 {
return (
UBool)(fUnion.fFields.fLengthAndFlags & kIsBogus); }
3920 UnicodeString::isWritable()
const
3921 {
return (
UBool)!(fUnion.fFields.fLengthAndFlags&(kOpenGetBuffer|kIsBogus)); }
3924 UnicodeString::isBufferWritable()
const
3927 !(fUnion.fFields.fLengthAndFlags&(kOpenGetBuffer|kIsBogus|kBufferIsReadonly)) &&
3928 (!(fUnion.fFields.fLengthAndFlags&kRefCounted) || refCount()==1));
3931 inline const char16_t *
3932 UnicodeString::getBuffer()
const {
3933 if(fUnion.fFields.fLengthAndFlags&(kIsBogus|kOpenGetBuffer)) {
3935 }
else if(fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) {
3936 return fUnion.fStackFields.fBuffer;
3938 return fUnion.fFields.fArray;
3946 UnicodeString::doCompare(int32_t start,
3950 int32_t srcLength)
const
3953 return (int8_t)!isBogus();
3955 srcText.pinIndices(srcStart, srcLength);
3956 return doCompare(start, thisLength, srcText.getArrayStart(), srcStart, srcLength);
3961 UnicodeString::doEqualsSubstring(int32_t start,
3963 const UnicodeString& srcText,
3965 int32_t srcLength)
const
3967 if(srcText.isBogus()) {
3970 srcText.pinIndices(srcStart, srcLength);
3971 return !isBogus() && doEqualsSubstring(start, thisLength, srcText.getArrayStart(), srcStart, srcLength);
3981 int32_t len = length(), textLength = text.
length();
3982 return !text.
isBogus() && len == textLength && doEquals(text, len);
3988 {
return (!
operator==(text)); }
3992 {
return doCompare(0, length(), text, 0, text.
length()) == 1; }
3996 {
return doCompare(0, length(), text, 0, text.
length()) == -1; }
4000 {
return doCompare(0, length(), text, 0, text.
length()) != -1; }
4004 {
return doCompare(0, length(), text, 0, text.
length()) != 1; }
4008 {
return doCompare(0, length(), text, 0, text.
length()); }
4011 UnicodeString::compare(int32_t start,
4014 {
return doCompare(start, _length, srcText, 0, srcText.
length()); }
4018 int32_t srcLength)
const
4019 {
return doCompare(0, length(), srcChars, 0, srcLength); }
4022 UnicodeString::compare(int32_t start,
4026 int32_t srcLength)
const
4027 {
return doCompare(start, _length, srcText, srcStart, srcLength); }
4030 UnicodeString::compare(int32_t start,
4032 const char16_t *srcChars)
const
4033 {
return doCompare(start, _length, srcChars, 0, _length); }
4036 UnicodeString::compare(int32_t start,
4038 const char16_t *srcChars,
4040 int32_t srcLength)
const
4041 {
return doCompare(start, _length, srcChars, srcStart, srcLength); }
4044 UnicodeString::compareBetween(int32_t start,
4048 int32_t srcLimit)
const
4049 {
return doCompare(start, limit - start,
4050 srcText, srcStart, srcLimit - srcStart); }
4053 UnicodeString::doCompareCodePointOrder(int32_t start,
4057 int32_t srcLength)
const
4060 return (int8_t)!isBogus();
4062 srcText.pinIndices(srcStart, srcLength);
4063 return doCompareCodePointOrder(start, thisLength, srcText.getArrayStart(), srcStart, srcLength);
4069 {
return doCompareCodePointOrder(0, length(), text, 0, text.
length()); }
4072 UnicodeString::compareCodePointOrder(int32_t start,
4075 {
return doCompareCodePointOrder(start, _length, srcText, 0, srcText.
length()); }
4079 int32_t srcLength)
const
4080 {
return doCompareCodePointOrder(0, length(), srcChars, 0, srcLength); }
4083 UnicodeString::compareCodePointOrder(int32_t start,
4087 int32_t srcLength)
const
4088 {
return doCompareCodePointOrder(start, _length, srcText, srcStart, srcLength); }
4091 UnicodeString::compareCodePointOrder(int32_t start,
4093 const char16_t *srcChars)
const
4094 {
return doCompareCodePointOrder(start, _length, srcChars, 0, _length); }
4097 UnicodeString::compareCodePointOrder(int32_t start,
4099 const char16_t *srcChars,
4101 int32_t srcLength)
const
4102 {
return doCompareCodePointOrder(start, _length, srcChars, srcStart, srcLength); }
4105 UnicodeString::compareCodePointOrderBetween(int32_t start,
4109 int32_t srcLimit)
const
4110 {
return doCompareCodePointOrder(start, limit - start,
4111 srcText, srcStart, srcLimit - srcStart); }
4114 UnicodeString::doCaseCompare(int32_t start,
4119 uint32_t options)
const
4122 return (int8_t)!isBogus();
4124 srcText.pinIndices(srcStart, srcLength);
4125 return doCaseCompare(start, thisLength, srcText.getArrayStart(), srcStart, srcLength, options);
4131 return doCaseCompare(0, length(), text, 0, text.
length(), options);
4135 UnicodeString::caseCompare(int32_t start,
4138 uint32_t options)
const {
4139 return doCaseCompare(start, _length, srcText, 0, srcText.
length(), options);
4145 uint32_t options)
const {
4146 return doCaseCompare(0, length(), srcChars, 0, srcLength, options);
4150 UnicodeString::caseCompare(int32_t start,
4155 uint32_t options)
const {
4156 return doCaseCompare(start, _length, srcText, srcStart, srcLength, options);
4160 UnicodeString::caseCompare(int32_t start,
4162 const char16_t *srcChars,
4163 uint32_t options)
const {
4164 return doCaseCompare(start, _length, srcChars, 0, _length, options);
4168 UnicodeString::caseCompare(int32_t start,
4170 const char16_t *srcChars,
4173 uint32_t options)
const {
4174 return doCaseCompare(start, _length, srcChars, srcStart, srcLength, options);
4178 UnicodeString::caseCompareBetween(int32_t start,
4183 uint32_t options)
const {
4184 return doCaseCompare(start, limit - start, srcText, srcStart, srcLimit - srcStart, options);
4192 int32_t _length)
const
4195 srcText.pinIndices(srcStart, srcLength);
4197 return indexOf(srcText.getArrayStart(), srcStart, srcLength, start, _length);
4205 {
return indexOf(text, 0, text.
length(), 0, length()); }
4209 int32_t start)
const {
4211 return indexOf(text, 0, text.
length(), start, length() - start);
4217 int32_t _length)
const
4218 {
return indexOf(text, 0, text.
length(), start, _length); }
4221 UnicodeString::indexOf(
const char16_t *srcChars,
4223 int32_t start)
const {
4225 return indexOf(srcChars, 0, srcLength, start, length() - start);
4232 int32_t _length)
const
4233 {
return indexOf(srcChars, 0, srcLength, start, _length); }
4236 UnicodeString::indexOf(char16_t c,
4238 int32_t _length)
const
4239 {
return doIndexOf(c, start, _length); }
4244 int32_t _length)
const
4245 {
return doIndexOf(c, start, _length); }
4248 UnicodeString::indexOf(char16_t c)
const
4249 {
return doIndexOf(c, 0, length()); }
4253 {
return indexOf(c, 0, length()); }
4256 UnicodeString::indexOf(char16_t c,
4257 int32_t start)
const {
4259 return doIndexOf(c, start, length() - start);
4264 int32_t start)
const {
4266 return indexOf(c, start, length() - start);
4273 int32_t _length)
const
4274 {
return lastIndexOf(srcChars, 0, srcLength, start, _length); }
4277 UnicodeString::lastIndexOf(
const char16_t *srcChars,
4279 int32_t start)
const {
4281 return lastIndexOf(srcChars, 0, srcLength, start, length() - start);
4289 int32_t _length)
const
4292 srcText.pinIndices(srcStart, srcLength);
4294 return lastIndexOf(srcText.getArrayStart(), srcStart, srcLength, start, _length);
4303 int32_t _length)
const
4304 {
return lastIndexOf(text, 0, text.
length(), start, _length); }
4308 int32_t start)
const {
4310 return lastIndexOf(text, 0, text.
length(), start, length() - start);
4315 {
return lastIndexOf(text, 0, text.
length(), 0, length()); }
4318 UnicodeString::lastIndexOf(char16_t c,
4320 int32_t _length)
const
4321 {
return doLastIndexOf(c, start, _length); }
4326 int32_t _length)
const {
4327 return doLastIndexOf(c, start, _length);
4331 UnicodeString::lastIndexOf(char16_t c)
const
4332 {
return doLastIndexOf(c, 0, length()); }
4336 return lastIndexOf(c, 0, length());
4340 UnicodeString::lastIndexOf(char16_t c,
4341 int32_t start)
const {
4343 return doLastIndexOf(c, start, length() - start);
4348 int32_t start)
const {
4350 return lastIndexOf(c, start, length() - start);
4355 {
return doEqualsSubstring(0, text.
length(), text, 0, text.
length()); }
4360 int32_t srcLength)
const
4361 {
return doEqualsSubstring(0, srcLength, srcText, srcStart, srcLength); }
4368 return doEqualsSubstring(0, srcLength, srcChars, 0, srcLength);
4372 UnicodeString::startsWith(
const char16_t *srcChars, int32_t srcStart, int32_t srcLength)
const {
4376 return doEqualsSubstring(0, srcLength, srcChars, srcStart, srcLength);
4381 {
return doEqualsSubstring(length() - text.
length(), text.
length(),
4382 text, 0, text.
length()); }
4387 int32_t srcLength)
const {
4388 srcText.pinIndices(srcStart, srcLength);
4389 return doEqualsSubstring(length() - srcLength, srcLength,
4390 srcText, srcStart, srcLength);
4395 int32_t srcLength)
const {
4399 return doEqualsSubstring(length() - srcLength, srcLength, srcChars, 0, srcLength);
4403 UnicodeString::endsWith(
const char16_t *srcChars,
4405 int32_t srcLength)
const {
4409 return doEqualsSubstring(length() - srcLength, srcLength,
4410 srcChars, srcStart, srcLength);
4417 UnicodeString::replace(int32_t start,
4420 {
return doReplace(start, _length, srcText, 0, srcText.
length()); }
4423 UnicodeString::replace(int32_t start,
4428 {
return doReplace(start, _length, srcText, srcStart, srcLength); }
4431 UnicodeString::replace(int32_t start,
4435 {
return doReplace(start, _length, srcChars, 0, srcLength); }
4438 UnicodeString::replace(int32_t start,
4440 const char16_t *srcChars,
4443 {
return doReplace(start, _length, srcChars, srcStart, srcLength); }
4446 UnicodeString::replace(int32_t start,
4449 {
return doReplace(start, _length, &srcChar, 0, 1); }
4452 UnicodeString::replaceBetween(int32_t start,
4455 {
return doReplace(start, limit - start, srcText, 0, srcText.
length()); }
4458 UnicodeString::replaceBetween(int32_t start,
4463 {
return doReplace(start, limit - start, srcText, srcStart, srcLimit - srcStart); }
4468 {
return findAndReplace(0, length(), oldText, 0, oldText.
length(),
4469 newText, 0, newText.
length()); }
4472 UnicodeString::findAndReplace(int32_t start,
4476 {
return findAndReplace(start, _length, oldText, 0, oldText.
length(),
4477 newText, 0, newText.
length()); }
4483 UnicodeString::doExtract(int32_t start,
4486 { target.
replace(0, target.
length(), *
this, start, _length); }
4489 UnicodeString::extract(int32_t start,
4492 int32_t targetStart)
const
4493 { doExtract(start, _length, target, targetStart); }
4496 UnicodeString::extract(int32_t start,
4499 { doExtract(start, _length, target); }
4501 #if !UCONFIG_NO_CONVERSION
4504 UnicodeString::extract(int32_t start,
4507 const char *codepage)
const
4511 return extract(start, _length, dst, dst !=
nullptr ? 0xffffffff : 0, codepage);
4517 UnicodeString::extractBetween(int32_t start,
4520 int32_t dstStart)
const {
4523 doExtract(start, limit - start, dst, dstStart);
4527 UnicodeString::tempSubStringBetween(int32_t start, int32_t limit)
const {
4528 return tempSubString(start, limit - start);
4532 UnicodeString::doCharAt(int32_t offset)
const
4534 if((uint32_t)offset < (uint32_t)length()) {
4535 return getArrayStart()[offset];
4537 return kInvalidUChar;
4542 UnicodeString::charAt(int32_t offset)
const
4543 {
return doCharAt(offset); }
4546 UnicodeString::operator[] (int32_t offset)
const
4547 {
return doCharAt(offset); }
4550 UnicodeString::isEmpty()
const {
4552 return (fUnion.fFields.fLengthAndFlags>>kLengthShift) == 0;
4559 UnicodeString::setZeroLength() {
4560 fUnion.fFields.fLengthAndFlags &= kAllStorageFlags;
4564 UnicodeString::setShortLength(int32_t len) {
4566 fUnion.fFields.fLengthAndFlags =
4567 (int16_t)((fUnion.fFields.fLengthAndFlags & kAllStorageFlags) | (len << kLengthShift));
4571 UnicodeString::setLength(int32_t len) {
4572 if(len <= kMaxShortLength) {
4573 setShortLength(len);
4575 fUnion.fFields.fLengthAndFlags |= kLengthIsLarge;
4576 fUnion.fFields.fLength = len;
4581 UnicodeString::setToEmpty() {
4582 fUnion.fFields.fLengthAndFlags = kShortString;
4586 UnicodeString::setArray(char16_t *array, int32_t len, int32_t capacity) {
4588 fUnion.fFields.fArray = array;
4589 fUnion.fFields.fCapacity = capacity;
4592 inline UnicodeString&
4593 UnicodeString::operator= (char16_t ch)
4594 {
return doReplace(0, length(), &ch, 0, 1); }
4598 {
return replace(0, length(), ch); }
4606 return doReplace(0, length(), srcText, srcStart, srcLength);
4614 srcText.pinIndex(srcStart);
4615 return doReplace(0, length(), srcText, srcStart, srcText.
length() - srcStart);
4621 return copyFrom(srcText);
4625 UnicodeString::setTo(
const char16_t *srcChars,
4629 return doReplace(0, length(), srcChars, 0, srcLength);
4633 UnicodeString::setTo(char16_t srcChar)
4636 return doReplace(0, length(), &srcChar, 0, 1);
4643 return replace(0, length(), srcChar);
4650 {
return doAppend(srcText, srcStart, srcLength); }
4654 {
return doAppend(srcText, 0, srcText.
length()); }
4657 UnicodeString::append(
const char16_t *srcChars,
4660 {
return doAppend(srcChars, srcStart, srcLength); }
4665 {
return doAppend(srcChars, 0, srcLength); }
4668 UnicodeString::append(char16_t srcChar)
4669 {
return doAppend(&srcChar, 0, 1); }
4672 UnicodeString::operator+= (char16_t ch)
4673 {
return doAppend(&ch, 0, 1); }
4682 {
return doAppend(srcText, 0, srcText.
length()); }
4685 UnicodeString::insert(int32_t start,
4689 {
return doReplace(start, 0, srcText, srcStart, srcLength); }
4692 UnicodeString::insert(int32_t start,
4694 {
return doReplace(start, 0, srcText, 0, srcText.
length()); }
4697 UnicodeString::insert(int32_t start,
4698 const char16_t *srcChars,
4701 {
return doReplace(start, 0, srcChars, srcStart, srcLength); }
4704 UnicodeString::insert(int32_t start,
4707 {
return doReplace(start, 0, srcChars, 0, srcLength); }
4710 UnicodeString::insert(int32_t start,
4712 {
return doReplace(start, 0, &srcChar, 0, 1); }
4715 UnicodeString::insert(int32_t start,
4717 {
return replace(start, 0, srcChar); }
4721 UnicodeString::remove()
4733 UnicodeString::remove(int32_t start,
4736 if(start <= 0 && _length ==
INT32_MAX) {
4740 return doReplace(start, _length,
nullptr, 0, 0);
4744 UnicodeString::removeBetween(int32_t start,
4746 {
return doReplace(start, limit - start,
nullptr, 0, 0); }
4749 UnicodeString::retainBetween(int32_t start, int32_t limit) {
4751 return doReplace(0, start,
nullptr, 0, 0);
4755 UnicodeString::truncate(int32_t targetLength)
4757 if(isBogus() && targetLength == 0) {
4761 }
else if((uint32_t)targetLength < (uint32_t)length()) {
4762 setLength(targetLength);
4770 UnicodeString::reverse()
4771 {
return doReverse(0, length()); }
4774 UnicodeString::reverse(int32_t start,
4776 {
return doReverse(start, _length); }
C++ API: Interface for writing bytes, and implementation classes.
C++ API: char16_t pointer wrappers with implicit conversion from bit-compatible raw pointer types.
The BreakIterator class implements methods for finding the location of boundaries in text.
A ByteSink can be filled with bytes.
char16_t * wrapper with implicit conversion from distinct but bit-compatible pointer types.
const char16_t * wrapper with implicit conversion from distinct but bit-compatible pointer types.
Records lengths of string edits but not replacement text.
A Locale object represents a specific geographical, political, or cultural region.
Replaceable is an abstract base class representing a string of characters that supports the replaceme...
virtual void extractBetween(int32_t start, int32_t limit, UnicodeString &target) const =0
Copies characters in the range [start, limit) into the UnicodeString target.
char16_t charAt(int32_t offset) const
Returns the 16-bit code unit at the given offset into the text.
int32_t length() const
Returns the number of 16-bit code units in the text.
Implementation of ByteSink that writes to a "string".
A string-like object that points to a sized piece of memory.
An Appendable implementation which writes to a UnicodeString.
UnicodeString is a string class that stores Unicode characters directly and provides similar function...
int32_t indexOf(const char16_t *srcChars, int32_t srcStart, int32_t srcLength, int32_t start, int32_t length) const
Locate in this the first occurrence in the range [start, start + length) of the characters in srcChar...
UnicodeString(const UnicodeString &that)
Copy constructor.
void swap(UnicodeString &other) noexcept
Swap strings.
virtual char16_t getCharAt(int32_t offset) const override
The change in Replaceable to use virtual getCharAt() allows UnicodeString::charAt() to be inline agai...
virtual int32_t getLength() const override
Implement Replaceable::getLength() (see jitterbug 1027).
UnicodeString & foldCase(uint32_t options=0)
Case-folds the characters in this string.
UChar32 unescapeAt(int32_t &offset) const
Unescape a single escape sequence and return the represented character.
UnicodeString(const wchar_t *text, int32_t textLength)
wchar_t * constructor.
virtual void handleReplaceBetween(int32_t start, int32_t limit, const UnicodeString &text) override
Replace a substring of this object with the given text.
UBool hasMoreChar32Than(int32_t start, int32_t length, int32_t number) const
Check if the length char16_t code units of the string contain more Unicode code points than a certain...
UnicodeString(const UnicodeString &src, int32_t srcStart, int32_t srcLength)
'Substring' constructor from subrange of source string.
UnicodeString & operator=(UnicodeString &&src) noexcept
Move assignment operator; might leave src in bogus state.
virtual ~UnicodeString()
Destructor.
UnicodeString(const char *codepageData, int32_t dataLength, const char *codepage)
char* constructor.
UnicodeString & toLower()
Convert the characters in this to lower case following the conventions of the default locale.
UnicodeString(const char *codepageData, const char *codepage)
char* constructor.
UnicodeString(UBool isTerminated, ConstChar16Ptr text, int32_t textLength)
Readonly-aliasing char16_t* constructor.
UnicodeString & fastCopyFrom(const UnicodeString &src)
Almost the same as the assignment operator.
UnicodeString & toTitle(BreakIterator *titleIter)
Titlecase this string, convenience function using the default locale.
EInvariant
Constant to be used in the UnicodeString(char *, int32_t, EInvariant) constructor which constructs a ...
UnicodeString unescape() const
Unescape a string of characters and return a string containing the result.
UnicodeString(const UnicodeString &src, int32_t srcStart)
'Substring' constructor from tail of source string.
int32_t getChar32Limit(int32_t offset) const
Adjust a random-access offset so that it points behind a Unicode character.
UnicodeString(char16_t *buffer, int32_t buffLength, int32_t buffCapacity)
Writable-aliasing char16_t* constructor.
UnicodeString(int32_t capacity, UChar32 c, int32_t count)
Construct a UnicodeString with capacity to hold capacity char16_ts.
UNISTR_FROM_STRING_EXPLICIT UnicodeString(const uint16_t *text)
uint16_t * constructor.
int32_t extract(int32_t start, int32_t startLength, char *target, uint32_t targetLength, const char *codepage) const
Copy the characters in the range [start, start + length) into an array of characters in a specified c...
UnicodeString & findAndReplace(int32_t start, int32_t length, const UnicodeString &oldText, int32_t oldStart, int32_t oldLength, const UnicodeString &newText, int32_t newStart, int32_t newLength)
Replace all occurrences of characters in oldText in the range [oldStart, oldStart + oldLength) with t...
virtual void copy(int32_t start, int32_t limit, int32_t dest) override
Copy a substring of this object, retaining attribute (out-of-band) information.
virtual void extractBetween(int32_t start, int32_t limit, UnicodeString &target) const override
Copy the characters in the range [start, limit) into the UnicodeString target.
UnicodeString & replace(int32_t start, int32_t length, const UnicodeString &srcText, int32_t srcStart, int32_t srcLength)
Replace the characters in the range [start, start + length) with the characters in srcText in the ran...
UBool padLeading(int32_t targetLength, char16_t padChar=0x0020)
Pad the start of this UnicodeString with the character padChar.
int32_t getChar32Start(int32_t offset) const
Adjust a random-access offset so that it points to the beginning of a Unicode character.
UChar32 char32At(int32_t offset) const
Return the code point that contains the code unit at offset offset.
UnicodeString(const char *src, int32_t textLength, enum EInvariant inv)
Constructs a Unicode string from an invariant-character char * string.
UnicodeString & operator=(const UnicodeString &srcText)
Assignment operator.
UnicodeString & append(UChar32 srcChar)
Append the code point srcChar to the UnicodeString object.
StringClass & toUTF8String(StringClass &result) const
Convert the UnicodeString to UTF-8 and append the result to a standard string.
UnicodeString & toLower(const Locale &locale)
Convert the characters in this to lower case following the conventions of a specific locale.
UnicodeString & toTitle(BreakIterator *titleIter, const Locale &locale)
Titlecase this string.
UnicodeString tempSubString(int32_t start=0, int32_t length=INT32_MAX) const
Create a temporary substring for the specified range.
int32_t extract(Char16Ptr dest, int32_t destCapacity, UErrorCode &errorCode) const
Copy the contents of the string into dest.
int32_t length() const
Return the length of the UnicodeString object.
virtual UChar32 getChar32At(int32_t offset) const override
The change in Replaceable to use virtual getChar32At() allows UnicodeString::char32At() to be inline ...
int32_t extract(int32_t start, int32_t startLength, char *target, uint32_t targetLength) const
Copy the characters in the range [start, start + length) into an array of characters in the platform'...
static UnicodeString fromUTF8(StringPiece utf8)
Create a UnicodeString from a UTF-8 string.
UNISTR_FROM_CHAR_EXPLICIT UnicodeString(char16_t ch)
Single char16_t (code unit) constructor.
int32_t lastIndexOf(const char16_t *srcChars, int32_t srcStart, int32_t srcLength, int32_t start, int32_t length) const
Locate in this the last occurrence in the range [start, start + length) of the characters in srcChars...
UnicodeString(wchar_t *buffer, int32_t buffLength, int32_t buffCapacity)
Writable-aliasing wchar_t * constructor.
void setToBogus()
Make this UnicodeString object invalid.
friend void swap(UnicodeString &s1, UnicodeString &s2) noexcept
Non-member UnicodeString swap function.
int32_t moveIndex32(int32_t index, int32_t delta) const
Move the code unit index along the string by delta code points.
static UnicodeString fromUTF32(const UChar32 *utf32, int32_t length)
Create a UnicodeString from a UTF-32 string.
int32_t countChar32(int32_t start=0, int32_t length=INT32_MAX) const
Count Unicode code points in the length char16_t code units of the string.
UNISTR_FROM_CHAR_EXPLICIT UnicodeString(UChar32 ch)
Single UChar32 (code point) constructor.
UnicodeString & setTo(UBool isTerminated, ConstChar16Ptr text, int32_t textLength)
Aliasing setTo() function, analogous to the readonly-aliasing char16_t* constructor.
UnicodeString & toTitle(BreakIterator *titleIter, const Locale &locale, uint32_t options)
Titlecase this string, with options.
virtual UnicodeString * clone() const override
Clone this object, an instance of a subclass of Replaceable.
int32_t extract(char *dest, int32_t destCapacity, UConverter *cnv, UErrorCode &errorCode) const
Convert the UnicodeString into a codepage string using an existing UConverter.
UnicodeString(UnicodeString &&src) noexcept
Move constructor; might leave src in bogus state.
UnicodeString(const char16_t *text, int32_t textLength)
char16_t* constructor.
UnicodeString & replace(int32_t start, int32_t length, UChar32 srcChar)
Replace the characters in the range [start, start + length) with the code point srcChar.
UnicodeString(const char *src, int32_t srcLength, UConverter *cnv, UErrorCode &errorCode)
char * / UConverter constructor.
UnicodeString(const char *codepageData, int32_t dataLength)
char* constructor.
static UClassID getStaticClassID()
ICU "poor man's RTTI", returns a UClassID for this class.
UnicodeString(uint16_t *buffer, int32_t buffLength, int32_t buffCapacity)
Writable-aliasing uint16_t * constructor.
int32_t toUTF32(UChar32 *utf32, int32_t capacity, UErrorCode &errorCode) const
Convert the UnicodeString to UTF-32.
UBool isBogus() const
Determine if this object contains a valid string.
const char16_t * getTerminatedBuffer()
Get a read-only pointer to the internal buffer, making sure that it is NUL-terminated.
UnicodeString(const uint16_t *text, int32_t textLength)
uint16_t * constructor.
UNISTR_FROM_STRING_EXPLICIT UnicodeString(const char16_t *text)
char16_t* constructor.
virtual UClassID getDynamicClassID() const override
ICU "poor man's RTTI", returns a UClassID for the actual class.
int32_t extract(int32_t start, int32_t startLength, char *target, int32_t targetCapacity, enum EInvariant inv) const
Copy the characters in the range [start, start + startLength) into an array of characters.
UNISTR_FROM_STRING_EXPLICIT UnicodeString(const wchar_t *text)
wchar_t * constructor.
UnicodeString & setCharAt(int32_t offset, char16_t ch)
Set the character at the specified offset to the specified character.
UnicodeString & setTo(char16_t *buffer, int32_t buffLength, int32_t buffCapacity)
Aliasing setTo() function, analogous to the writable-aliasing char16_t* constructor.
UBool padTrailing(int32_t targetLength, char16_t padChar=0x0020)
Pad the end of this UnicodeString with the character padChar.
UNISTR_FROM_STRING_EXPLICIT UnicodeString(const char *codepageData)
char* constructor.
void releaseBuffer(int32_t newLength=-1)
Release a read/write buffer on a UnicodeString object with an "open" getBuffer(minCapacity).
char16_t * getBuffer(int32_t minCapacity)
Get a read/write pointer to the internal buffer.
void toUTF8(ByteSink &sink) const
Convert the UnicodeString to UTF-8 and write the result to a ByteSink.
virtual UBool hasMetaData() const override
Replaceable API.
UnicodeString & toUpper()
Convert the characters in this to UPPER CASE following the conventions of the default locale.
UnicodeString & toUpper(const Locale &locale)
Convert the characters in this to UPPER CASE following the conventions of a specific locale.
UnicodeString & trim()
Trims leading and trailing whitespace from this UnicodeString.
U_CAPI int32_t u_strlen(const UChar *s)
U_EXPORT UBool operator==(const StringPiece &x, const StringPiece &y)
Global operator == for StringPiece.
bool operator!=(const StringPiece &x, const StringPiece &y)
Global operator != for StringPiece.
const UChar * toUCharPtr(const char16_t *p)
Converts from const char16_t * to const UChar *.
U_COMMON_API UnicodeString operator+(const UnicodeString &s1, const UnicodeString &s2)
Create a new UnicodeString with the concatenation of two others.
C++ API: Replaceable String.
C++ API: Central ICU header for including the C++ standard <string> header and for related definition...
C++ API: StringPiece: Read-only byte string wrapper class.
struct UConverter UConverter
#define UCONFIG_NO_BREAK_ITERATION
This switch turns off break iteration.
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
#define INT32_MAX
The largest value a 32 bit signed integer can hold.
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_SIZEOF_UCHAR
Number of bytes in a UChar (always 2).
#define UNISTR_FROM_CHAR_EXPLICIT
This can be defined to be empty or "explicit".
int32_t UStringCaseMapper(int32_t caseLocale, uint32_t options, icu::BreakIterator *iter, char16_t *dest, int32_t destCapacity, const char16_t *src, int32_t srcLength, icu::Edits *edits, UErrorCode &errorCode)
Internal string case mapping function type.
#define UNISTR_FROM_STRING_EXPLICIT
This can be defined to be empty or "explicit".
#define UNISTR_OBJECT_SIZE
Desired sizeof(UnicodeString) in bytes.
void * UClassID
UClassID is used to identify classes without using the compiler's RTTI.
Basic definitions for ICU, for both C and C++ APIs.
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
#define U_COMMON_API
Set to export library symbols from inside the common library, and to import them from outside.