ICU 69.1  69.1
unistr.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) 1998-2016, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 **********************************************************************
8 *
9 * File unistr.h
10 *
11 * Modification History:
12 *
13 * Date Name Description
14 * 09/25/98 stephen Creation.
15 * 11/11/98 stephen Changed per 11/9 code review.
16 * 04/20/99 stephen Overhauled per 4/16 code review.
17 * 11/18/99 aliu Made to inherit from Replaceable. Added method
18 * handleReplaceBetween(); other methods unchanged.
19 * 06/25/01 grhoten Remove dependency on iostream.
20 ******************************************************************************
21 */
22 
23 #ifndef UNISTR_H
24 #define UNISTR_H
25 
31 #include "unicode/utypes.h"
32 
33 #if U_SHOW_CPLUSPLUS_API
34 
35 #include <cstddef>
36 #include "unicode/char16ptr.h"
37 #include "unicode/rep.h"
38 #include "unicode/std_string.h"
39 #include "unicode/stringpiece.h"
40 #include "unicode/bytestream.h"
41 
42 struct UConverter; // unicode/ucnv.h
43 
44 #ifndef USTRING_H
50 U_CAPI int32_t U_EXPORT2 u_strlen(const UChar *s);
51 #endif
52 
53 U_NAMESPACE_BEGIN
54 
55 #if !UCONFIG_NO_BREAK_ITERATION
56 class BreakIterator; // unicode/brkiter.h
57 #endif
58 class Edits;
59 
60 U_NAMESPACE_END
61 
62 // Not #ifndef U_HIDE_INTERNAL_API because UnicodeString needs the UStringCaseMapper.
69 typedef int32_t U_CALLCONV
70 UStringCaseMapper(int32_t caseLocale, uint32_t options,
72  icu::BreakIterator *iter,
73 #endif
74  char16_t *dest, int32_t destCapacity,
75  const char16_t *src, int32_t srcLength,
76  icu::Edits *edits,
77  UErrorCode &errorCode);
78 
79 U_NAMESPACE_BEGIN
80 
81 class Locale; // unicode/locid.h
82 class StringCharacterIterator;
83 class UnicodeStringAppendable; // unicode/appendable.h
84 
85 /* The <iostream> include has been moved to unicode/ustream.h */
86 
97 #define US_INV icu::UnicodeString::kInvariant
98 
116 #if !U_CHAR16_IS_TYPEDEF
117 # define UNICODE_STRING(cs, _length) icu::UnicodeString(true, u ## cs, _length)
118 #else
119 # define UNICODE_STRING(cs, _length) icu::UnicodeString(true, (const char16_t*)u ## cs, _length)
120 #endif
121 
135 #define UNICODE_STRING_SIMPLE(cs) UNICODE_STRING(cs, -1)
136 
144 #ifndef UNISTR_FROM_CHAR_EXPLICIT
145 # if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION)
146  // Auto-"explicit" in ICU library code.
147 # define UNISTR_FROM_CHAR_EXPLICIT explicit
148 # else
149  // Empty by default for source code compatibility.
150 # define UNISTR_FROM_CHAR_EXPLICIT
151 # endif
152 #endif
153 
164 #ifndef UNISTR_FROM_STRING_EXPLICIT
165 # if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION)
166  // Auto-"explicit" in ICU library code.
167 # define UNISTR_FROM_STRING_EXPLICIT explicit
168 # else
169  // Empty by default for source code compatibility.
170 # define UNISTR_FROM_STRING_EXPLICIT
171 # endif
172 #endif
173 
207 #ifndef UNISTR_OBJECT_SIZE
208 # define UNISTR_OBJECT_SIZE 64
209 #endif
210 
296 {
297 public:
298 
307  enum EInvariant {
312  kInvariant
313  };
314 
315  //========================================
316  // Read-only operations
317  //========================================
318 
319  /* Comparison - bitwise only - for international comparison use collation */
320 
328  inline UBool operator== (const UnicodeString& text) const;
329 
337  inline UBool operator!= (const UnicodeString& text) const;
338 
346  inline UBool operator> (const UnicodeString& text) const;
347 
355  inline UBool operator< (const UnicodeString& text) const;
356 
364  inline UBool operator>= (const UnicodeString& text) const;
365 
373  inline UBool operator<= (const UnicodeString& text) const;
374 
386  inline int8_t compare(const UnicodeString& text) const;
387 
403  inline int8_t compare(int32_t start,
404  int32_t length,
405  const UnicodeString& text) const;
406 
424  inline int8_t compare(int32_t start,
425  int32_t length,
426  const UnicodeString& srcText,
427  int32_t srcStart,
428  int32_t srcLength) const;
429 
442  inline int8_t compare(ConstChar16Ptr srcChars,
443  int32_t srcLength) const;
444 
459  inline int8_t compare(int32_t start,
460  int32_t length,
461  const char16_t *srcChars) const;
462 
480  inline int8_t compare(int32_t start,
481  int32_t length,
482  const char16_t *srcChars,
483  int32_t srcStart,
484  int32_t srcLength) const;
485 
503  inline int8_t compareBetween(int32_t start,
504  int32_t limit,
505  const UnicodeString& srcText,
506  int32_t srcStart,
507  int32_t srcLimit) const;
508 
526  inline int8_t compareCodePointOrder(const UnicodeString& text) const;
527 
547  inline int8_t compareCodePointOrder(int32_t start,
548  int32_t length,
549  const UnicodeString& srcText) const;
550 
572  inline int8_t compareCodePointOrder(int32_t start,
573  int32_t length,
574  const UnicodeString& srcText,
575  int32_t srcStart,
576  int32_t srcLength) const;
577 
596  inline int8_t compareCodePointOrder(ConstChar16Ptr srcChars,
597  int32_t srcLength) const;
598 
618  inline int8_t compareCodePointOrder(int32_t start,
619  int32_t length,
620  const char16_t *srcChars) const;
621 
643  inline int8_t compareCodePointOrder(int32_t start,
644  int32_t length,
645  const char16_t *srcChars,
646  int32_t srcStart,
647  int32_t srcLength) const;
648 
670  inline int8_t compareCodePointOrderBetween(int32_t start,
671  int32_t limit,
672  const UnicodeString& srcText,
673  int32_t srcStart,
674  int32_t srcLimit) const;
675 
694  inline int8_t caseCompare(const UnicodeString& text, uint32_t options) const;
695 
716  inline int8_t caseCompare(int32_t start,
717  int32_t length,
718  const UnicodeString& srcText,
719  uint32_t options) const;
720 
743  inline int8_t caseCompare(int32_t start,
744  int32_t length,
745  const UnicodeString& srcText,
746  int32_t srcStart,
747  int32_t srcLength,
748  uint32_t options) const;
749 
769  inline int8_t caseCompare(ConstChar16Ptr srcChars,
770  int32_t srcLength,
771  uint32_t options) const;
772 
793  inline int8_t caseCompare(int32_t start,
794  int32_t length,
795  const char16_t *srcChars,
796  uint32_t options) const;
797 
820  inline int8_t caseCompare(int32_t start,
821  int32_t length,
822  const char16_t *srcChars,
823  int32_t srcStart,
824  int32_t srcLength,
825  uint32_t options) const;
826 
849  inline int8_t caseCompareBetween(int32_t start,
850  int32_t limit,
851  const UnicodeString& srcText,
852  int32_t srcStart,
853  int32_t srcLimit,
854  uint32_t options) const;
855 
863  inline UBool startsWith(const UnicodeString& text) const;
864 
875  inline UBool startsWith(const UnicodeString& srcText,
876  int32_t srcStart,
877  int32_t srcLength) const;
878 
887  inline UBool startsWith(ConstChar16Ptr srcChars,
888  int32_t srcLength) const;
889 
899  inline UBool startsWith(const char16_t *srcChars,
900  int32_t srcStart,
901  int32_t srcLength) const;
902 
910  inline UBool endsWith(const UnicodeString& text) const;
911 
922  inline UBool endsWith(const UnicodeString& srcText,
923  int32_t srcStart,
924  int32_t srcLength) const;
925 
934  inline UBool endsWith(ConstChar16Ptr srcChars,
935  int32_t srcLength) const;
936 
947  inline UBool endsWith(const char16_t *srcChars,
948  int32_t srcStart,
949  int32_t srcLength) const;
950 
951 
952  /* Searching - bitwise only */
953 
962  inline int32_t indexOf(const UnicodeString& text) const;
963 
973  inline int32_t indexOf(const UnicodeString& text,
974  int32_t start) const;
975 
987  inline int32_t indexOf(const UnicodeString& text,
988  int32_t start,
989  int32_t length) const;
990 
1007  inline int32_t indexOf(const UnicodeString& srcText,
1008  int32_t srcStart,
1009  int32_t srcLength,
1010  int32_t start,
1011  int32_t length) const;
1012 
1024  inline int32_t indexOf(const char16_t *srcChars,
1025  int32_t srcLength,
1026  int32_t start) const;
1027 
1040  inline int32_t indexOf(ConstChar16Ptr srcChars,
1041  int32_t srcLength,
1042  int32_t start,
1043  int32_t length) const;
1044 
1061  int32_t indexOf(const char16_t *srcChars,
1062  int32_t srcStart,
1063  int32_t srcLength,
1064  int32_t start,
1065  int32_t length) const;
1066 
1074  inline int32_t indexOf(char16_t c) const;
1075 
1084  inline int32_t indexOf(UChar32 c) const;
1085 
1094  inline int32_t indexOf(char16_t c,
1095  int32_t start) const;
1096 
1106  inline int32_t indexOf(UChar32 c,
1107  int32_t start) const;
1108 
1119  inline int32_t indexOf(char16_t c,
1120  int32_t start,
1121  int32_t length) const;
1122 
1134  inline int32_t indexOf(UChar32 c,
1135  int32_t start,
1136  int32_t length) const;
1137 
1146  inline int32_t lastIndexOf(const UnicodeString& text) const;
1147 
1157  inline int32_t lastIndexOf(const UnicodeString& text,
1158  int32_t start) const;
1159 
1171  inline int32_t lastIndexOf(const UnicodeString& text,
1172  int32_t start,
1173  int32_t length) const;
1174 
1191  inline int32_t lastIndexOf(const UnicodeString& srcText,
1192  int32_t srcStart,
1193  int32_t srcLength,
1194  int32_t start,
1195  int32_t length) const;
1196 
1207  inline int32_t lastIndexOf(const char16_t *srcChars,
1208  int32_t srcLength,
1209  int32_t start) const;
1210 
1223  inline int32_t lastIndexOf(ConstChar16Ptr srcChars,
1224  int32_t srcLength,
1225  int32_t start,
1226  int32_t length) const;
1227 
1244  int32_t lastIndexOf(const char16_t *srcChars,
1245  int32_t srcStart,
1246  int32_t srcLength,
1247  int32_t start,
1248  int32_t length) const;
1249 
1257  inline int32_t lastIndexOf(char16_t c) const;
1258 
1267  inline int32_t lastIndexOf(UChar32 c) const;
1268 
1277  inline int32_t lastIndexOf(char16_t c,
1278  int32_t start) const;
1279 
1289  inline int32_t lastIndexOf(UChar32 c,
1290  int32_t start) const;
1291 
1302  inline int32_t lastIndexOf(char16_t c,
1303  int32_t start,
1304  int32_t length) const;
1305 
1317  inline int32_t lastIndexOf(UChar32 c,
1318  int32_t start,
1319  int32_t length) const;
1320 
1321 
1322  /* Character access */
1323 
1332  inline char16_t charAt(int32_t offset) const;
1333 
1341  inline char16_t operator[] (int32_t offset) const;
1342 
1354  UChar32 char32At(int32_t offset) const;
1355 
1371  int32_t getChar32Start(int32_t offset) const;
1372 
1389  int32_t getChar32Limit(int32_t offset) const;
1390 
1441  int32_t moveIndex32(int32_t index, int32_t delta) const;
1442 
1443  /* Substring extraction */
1444 
1460  inline void extract(int32_t start,
1461  int32_t length,
1462  Char16Ptr dst,
1463  int32_t dstStart = 0) const;
1464 
1486  int32_t
1487  extract(Char16Ptr dest, int32_t destCapacity,
1488  UErrorCode &errorCode) const;
1489 
1499  inline void extract(int32_t start,
1500  int32_t length,
1501  UnicodeString& target) const;
1502 
1514  inline void extractBetween(int32_t start,
1515  int32_t limit,
1516  char16_t *dst,
1517  int32_t dstStart = 0) const;
1518 
1527  virtual void extractBetween(int32_t start,
1528  int32_t limit,
1529  UnicodeString& target) const;
1530 
1552  int32_t extract(int32_t start,
1553  int32_t startLength,
1554  char *target,
1555  int32_t targetCapacity,
1556  enum EInvariant inv) const;
1557 
1558 #if U_CHARSET_IS_UTF8 || !UCONFIG_NO_CONVERSION
1559 
1579  int32_t extract(int32_t start,
1580  int32_t startLength,
1581  char *target,
1582  uint32_t targetLength) const;
1583 
1584 #endif
1585 
1586 #if !UCONFIG_NO_CONVERSION
1587 
1613  inline int32_t extract(int32_t start,
1614  int32_t startLength,
1615  char *target,
1616  const char *codepage = 0) const;
1617 
1647  int32_t extract(int32_t start,
1648  int32_t startLength,
1649  char *target,
1650  uint32_t targetLength,
1651  const char *codepage) const;
1652 
1670  int32_t extract(char *dest, int32_t destCapacity,
1671  UConverter *cnv,
1672  UErrorCode &errorCode) const;
1673 
1674 #endif
1675 
1689  UnicodeString tempSubString(int32_t start=0, int32_t length=INT32_MAX) const;
1690 
1701  inline UnicodeString tempSubStringBetween(int32_t start, int32_t limit=INT32_MAX) const;
1702 
1714  void toUTF8(ByteSink &sink) const;
1715 
1728  template<typename StringClass>
1729  StringClass &toUTF8String(StringClass &result) const {
1730  StringByteSink<StringClass> sbs(&result, length());
1731  toUTF8(sbs);
1732  return result;
1733  }
1734 
1750  int32_t toUTF32(UChar32 *utf32, int32_t capacity, UErrorCode &errorCode) const;
1751 
1752  /* Length operations */
1753 
1762  inline int32_t length(void) const;
1763 
1777  int32_t
1778  countChar32(int32_t start=0, int32_t length=INT32_MAX) const;
1779 
1803  UBool
1804  hasMoreChar32Than(int32_t start, int32_t length, int32_t number) const;
1805 
1811  inline UBool isEmpty(void) const;
1812 
1822  inline int32_t getCapacity(void) const;
1823 
1824  /* Other operations */
1825 
1831  inline int32_t hashCode(void) const;
1832 
1845  inline UBool isBogus(void) const;
1846 
1847 
1848  //========================================
1849  // Write operations
1850  //========================================
1851 
1852  /* Assignment operations */
1853 
1873 
1900 
1910 
1917 
1924  friend inline void U_EXPORT2
1926  s1.swap(s2);
1927  }
1928 
1936  inline UnicodeString& operator= (char16_t ch);
1937 
1945  inline UnicodeString& operator= (UChar32 ch);
1946 
1958  inline UnicodeString& setTo(const UnicodeString& srcText,
1959  int32_t srcStart);
1960 
1974  inline UnicodeString& setTo(const UnicodeString& srcText,
1975  int32_t srcStart,
1976  int32_t srcLength);
1977 
1986  inline UnicodeString& setTo(const UnicodeString& srcText);
1987 
1996  inline UnicodeString& setTo(const char16_t *srcChars,
1997  int32_t srcLength);
1998 
2007  inline UnicodeString& setTo(char16_t srcChar);
2008 
2017  inline UnicodeString& setTo(UChar32 srcChar);
2018 
2042  UnicodeString &setTo(UBool isTerminated,
2043  ConstChar16Ptr text,
2044  int32_t textLength);
2045 
2065  UnicodeString &setTo(char16_t *buffer,
2066  int32_t buffLength,
2067  int32_t buffCapacity);
2068 
2108  void setToBogus();
2109 
2117  UnicodeString& setCharAt(int32_t offset,
2118  char16_t ch);
2119 
2120 
2121  /* Append operations */
2122 
2130  inline UnicodeString& operator+= (char16_t ch);
2131 
2139  inline UnicodeString& operator+= (UChar32 ch);
2140 
2148  inline UnicodeString& operator+= (const UnicodeString& srcText);
2149 
2164  inline UnicodeString& append(const UnicodeString& srcText,
2165  int32_t srcStart,
2166  int32_t srcLength);
2167 
2175  inline UnicodeString& append(const UnicodeString& srcText);
2176 
2190  inline UnicodeString& append(const char16_t *srcChars,
2191  int32_t srcStart,
2192  int32_t srcLength);
2193 
2203  inline UnicodeString& append(ConstChar16Ptr srcChars,
2204  int32_t srcLength);
2205 
2212  inline UnicodeString& append(char16_t srcChar);
2213 
2221 
2222 
2223  /* Insert operations */
2224 
2238  inline UnicodeString& insert(int32_t start,
2239  const UnicodeString& srcText,
2240  int32_t srcStart,
2241  int32_t srcLength);
2242 
2251  inline UnicodeString& insert(int32_t start,
2252  const UnicodeString& srcText);
2253 
2267  inline UnicodeString& insert(int32_t start,
2268  const char16_t *srcChars,
2269  int32_t srcStart,
2270  int32_t srcLength);
2271 
2281  inline UnicodeString& insert(int32_t start,
2282  ConstChar16Ptr srcChars,
2283  int32_t srcLength);
2284 
2293  inline UnicodeString& insert(int32_t start,
2294  char16_t srcChar);
2295 
2304  inline UnicodeString& insert(int32_t start,
2305  UChar32 srcChar);
2306 
2307 
2308  /* Replace operations */
2309 
2327  inline UnicodeString& replace(int32_t start,
2328  int32_t length,
2329  const UnicodeString& srcText,
2330  int32_t srcStart,
2331  int32_t srcLength);
2332 
2345  inline UnicodeString& replace(int32_t start,
2346  int32_t length,
2347  const UnicodeString& srcText);
2348 
2366  inline UnicodeString& replace(int32_t start,
2367  int32_t length,
2368  const char16_t *srcChars,
2369  int32_t srcStart,
2370  int32_t srcLength);
2371 
2384  inline UnicodeString& replace(int32_t start,
2385  int32_t length,
2386  ConstChar16Ptr srcChars,
2387  int32_t srcLength);
2388 
2400  inline UnicodeString& replace(int32_t start,
2401  int32_t length,
2402  char16_t srcChar);
2403 
2415  UnicodeString& replace(int32_t start, int32_t length, UChar32 srcChar);
2416 
2426  inline UnicodeString& replaceBetween(int32_t start,
2427  int32_t limit,
2428  const UnicodeString& srcText);
2429 
2444  inline UnicodeString& replaceBetween(int32_t start,
2445  int32_t limit,
2446  const UnicodeString& srcText,
2447  int32_t srcStart,
2448  int32_t srcLimit);
2449 
2457  virtual void handleReplaceBetween(int32_t start,
2458  int32_t limit,
2459  const UnicodeString& text);
2460 
2466  virtual UBool hasMetaData() const;
2467 
2481  virtual void copy(int32_t start, int32_t limit, int32_t dest);
2482 
2483  /* Search and replace operations */
2484 
2493  inline UnicodeString& findAndReplace(const UnicodeString& oldText,
2494  const UnicodeString& newText);
2495 
2507  inline UnicodeString& findAndReplace(int32_t start,
2508  int32_t length,
2509  const UnicodeString& oldText,
2510  const UnicodeString& newText);
2511 
2530  int32_t length,
2531  const UnicodeString& oldText,
2532  int32_t oldStart,
2533  int32_t oldLength,
2534  const UnicodeString& newText,
2535  int32_t newStart,
2536  int32_t newLength);
2537 
2538 
2539  /* Remove operations */
2540 
2549  inline UnicodeString& remove();
2550 
2559  inline UnicodeString& remove(int32_t start,
2560  int32_t length = (int32_t)INT32_MAX);
2561 
2570  inline UnicodeString& removeBetween(int32_t start,
2571  int32_t limit = (int32_t)INT32_MAX);
2572 
2582  inline UnicodeString &retainBetween(int32_t start, int32_t limit = INT32_MAX);
2583 
2584  /* Length operations */
2585 
2597  UBool padLeading(int32_t targetLength,
2598  char16_t padChar = 0x0020);
2599 
2611  UBool padTrailing(int32_t targetLength,
2612  char16_t padChar = 0x0020);
2613 
2620  inline UBool truncate(int32_t targetLength);
2621 
2628 
2629 
2630  /* Miscellaneous operations */
2631 
2637  inline UnicodeString& reverse(void);
2638 
2647  inline UnicodeString& reverse(int32_t start,
2648  int32_t length);
2649 
2657 
2665  UnicodeString& toUpper(const Locale& locale);
2666 
2674 
2682  UnicodeString& toLower(const Locale& locale);
2683 
2684 #if !UCONFIG_NO_BREAK_ITERATION
2685 
2713 
2741  UnicodeString &toTitle(BreakIterator *titleIter, const Locale &locale);
2742 
2773  UnicodeString &toTitle(BreakIterator *titleIter, const Locale &locale, uint32_t options);
2774 
2775 #endif
2776 
2790  UnicodeString &foldCase(uint32_t options=0 /*U_FOLD_CASE_DEFAULT*/);
2791 
2792  //========================================
2793  // Access to the internal buffer
2794  //========================================
2795 
2839  char16_t *getBuffer(int32_t minCapacity);
2840 
2861  void releaseBuffer(int32_t newLength=-1);
2862 
2893  inline const char16_t *getBuffer() const;
2894 
2928  const char16_t *getTerminatedBuffer();
2929 
2930  //========================================
2931  // Constructors
2932  //========================================
2933 
2937  inline UnicodeString();
2938 
2950  UnicodeString(int32_t capacity, UChar32 c, int32_t count);
2951 
2962 
2973 
2985 
2986 #if !U_CHAR16_IS_TYPEDEF
2998  UnicodeString(ConstChar16Ptr(text)) {}
2999 #endif
3000 
3001 #if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN)
3014  UnicodeString(ConstChar16Ptr(text)) {}
3015 #endif
3016 
3027  UNISTR_FROM_STRING_EXPLICIT inline UnicodeString(const std::nullptr_t text);
3028 
3036  UnicodeString(const char16_t *text,
3037  int32_t textLength);
3038 
3039 #if !U_CHAR16_IS_TYPEDEF
3047  UnicodeString(const uint16_t *text, int32_t textLength) :
3048  UnicodeString(ConstChar16Ptr(text), textLength) {}
3049 #endif
3050 
3051 #if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN)
3060  UnicodeString(const wchar_t *text, int32_t textLength) :
3061  UnicodeString(ConstChar16Ptr(text), textLength) {}
3062 #endif
3063 
3071  inline UnicodeString(const std::nullptr_t text, int32_t textLength);
3072 
3095  UnicodeString(UBool isTerminated,
3096  ConstChar16Ptr text,
3097  int32_t textLength);
3098 
3117  UnicodeString(char16_t *buffer, int32_t buffLength, int32_t buffCapacity);
3118 
3119 #if !U_CHAR16_IS_TYPEDEF
3128  UnicodeString(uint16_t *buffer, int32_t buffLength, int32_t buffCapacity) :
3129  UnicodeString(Char16Ptr(buffer), buffLength, buffCapacity) {}
3130 #endif
3131 
3132 #if U_SIZEOF_WCHAR_T==2 || defined(U_IN_DOXYGEN)
3142  UnicodeString(wchar_t *buffer, int32_t buffLength, int32_t buffCapacity) :
3143  UnicodeString(Char16Ptr(buffer), buffLength, buffCapacity) {}
3144 #endif
3145 
3154  inline UnicodeString(std::nullptr_t buffer, int32_t buffLength, int32_t buffCapacity);
3155 
3156 #if U_CHARSET_IS_UTF8 || !UCONFIG_NO_CONVERSION
3157 
3177  UNISTR_FROM_STRING_EXPLICIT UnicodeString(const char *codepageData);
3178 
3187  UnicodeString(const char *codepageData, int32_t dataLength);
3188 
3189 #endif
3190 
3191 #if !UCONFIG_NO_CONVERSION
3192 
3210  UnicodeString(const char *codepageData, const char *codepage);
3211 
3229  UnicodeString(const char *codepageData, int32_t dataLength, const char *codepage);
3230 
3253  const char *src, int32_t srcLength,
3254  UConverter *cnv,
3255  UErrorCode &errorCode);
3256 
3257 #endif
3258 
3282  UnicodeString(const char *src, int32_t textLength, enum EInvariant inv);
3283 
3284 
3302 
3310 
3317  UnicodeString(const UnicodeString& src, int32_t srcStart);
3318 
3326  UnicodeString(const UnicodeString& src, int32_t srcStart, int32_t srcLength);
3327 
3341  virtual UnicodeString *clone() const;
3342 
3346  virtual ~UnicodeString();
3347 
3362 
3374  static UnicodeString fromUTF32(const UChar32 *utf32, int32_t length);
3375 
3376  /* Miscellaneous operations */
3377 
3413 
3433  UChar32 unescapeAt(int32_t &offset) const;
3434 
3440  static UClassID U_EXPORT2 getStaticClassID();
3441 
3447  virtual UClassID getDynamicClassID() const;
3448 
3449  //========================================
3450  // Implementation methods
3451  //========================================
3452 
3453 protected:
3458  virtual int32_t getLength() const;
3459 
3465  virtual char16_t getCharAt(int32_t offset) const;
3466 
3472  virtual UChar32 getChar32At(int32_t offset) const;
3473 
3474 private:
3475  // For char* constructors. Could be made public.
3476  UnicodeString &setToUTF8(StringPiece utf8);
3477  // For extract(char*).
3478  // We could make a toUTF8(target, capacity, errorCode) public but not
3479  // this version: New API will be cleaner if we make callers create substrings
3480  // rather than having start+length on every method,
3481  // and it should take a UErrorCode&.
3482  int32_t
3483  toUTF8(int32_t start, int32_t len,
3484  char *target, int32_t capacity) const;
3485 
3490  UBool doEquals(const UnicodeString &text, int32_t len) const;
3491 
3492  inline int8_t
3493  doCompare(int32_t start,
3494  int32_t length,
3495  const UnicodeString& srcText,
3496  int32_t srcStart,
3497  int32_t srcLength) const;
3498 
3499  int8_t doCompare(int32_t start,
3500  int32_t length,
3501  const char16_t *srcChars,
3502  int32_t srcStart,
3503  int32_t srcLength) const;
3504 
3505  inline int8_t
3506  doCompareCodePointOrder(int32_t start,
3507  int32_t length,
3508  const UnicodeString& srcText,
3509  int32_t srcStart,
3510  int32_t srcLength) const;
3511 
3512  int8_t doCompareCodePointOrder(int32_t start,
3513  int32_t length,
3514  const char16_t *srcChars,
3515  int32_t srcStart,
3516  int32_t srcLength) const;
3517 
3518  inline int8_t
3519  doCaseCompare(int32_t start,
3520  int32_t length,
3521  const UnicodeString &srcText,
3522  int32_t srcStart,
3523  int32_t srcLength,
3524  uint32_t options) const;
3525 
3526  int8_t
3527  doCaseCompare(int32_t start,
3528  int32_t length,
3529  const char16_t *srcChars,
3530  int32_t srcStart,
3531  int32_t srcLength,
3532  uint32_t options) const;
3533 
3534  int32_t doIndexOf(char16_t c,
3535  int32_t start,
3536  int32_t length) const;
3537 
3538  int32_t doIndexOf(UChar32 c,
3539  int32_t start,
3540  int32_t length) const;
3541 
3542  int32_t doLastIndexOf(char16_t c,
3543  int32_t start,
3544  int32_t length) const;
3545 
3546  int32_t doLastIndexOf(UChar32 c,
3547  int32_t start,
3548  int32_t length) const;
3549 
3550  void doExtract(int32_t start,
3551  int32_t length,
3552  char16_t *dst,
3553  int32_t dstStart) const;
3554 
3555  inline void doExtract(int32_t start,
3556  int32_t length,
3557  UnicodeString& target) const;
3558 
3559  inline char16_t doCharAt(int32_t offset) const;
3560 
3561  UnicodeString& doReplace(int32_t start,
3562  int32_t length,
3563  const UnicodeString& srcText,
3564  int32_t srcStart,
3565  int32_t srcLength);
3566 
3567  UnicodeString& doReplace(int32_t start,
3568  int32_t length,
3569  const char16_t *srcChars,
3570  int32_t srcStart,
3571  int32_t srcLength);
3572 
3573  UnicodeString& doAppend(const UnicodeString& src, int32_t srcStart, int32_t srcLength);
3574  UnicodeString& doAppend(const char16_t *srcChars, int32_t srcStart, int32_t srcLength);
3575 
3576  UnicodeString& doReverse(int32_t start,
3577  int32_t length);
3578 
3579  // calculate hash code
3580  int32_t doHashCode(void) const;
3581 
3582  // get pointer to start of array
3583  // these do not check for kOpenGetBuffer, unlike the public getBuffer() function
3584  inline char16_t* getArrayStart(void);
3585  inline const char16_t* getArrayStart(void) const;
3586 
3587  inline UBool hasShortLength() const;
3588  inline int32_t getShortLength() const;
3589 
3590  // A UnicodeString object (not necessarily its current buffer)
3591  // is writable unless it isBogus() or it has an "open" getBuffer(minCapacity).
3592  inline UBool isWritable() const;
3593 
3594  // Is the current buffer writable?
3595  inline UBool isBufferWritable() const;
3596 
3597  // None of the following does releaseArray().
3598  inline void setZeroLength();
3599  inline void setShortLength(int32_t len);
3600  inline void setLength(int32_t len);
3601  inline void setToEmpty();
3602  inline void setArray(char16_t *array, int32_t len, int32_t capacity); // sets length but not flags
3603 
3604  // allocate the array; result may be the stack buffer
3605  // sets refCount to 1 if appropriate
3606  // sets fArray, fCapacity, and flags
3607  // sets length to 0
3608  // returns boolean for success or failure
3609  UBool allocate(int32_t capacity);
3610 
3611  // release the array if owned
3612  void releaseArray(void);
3613 
3614  // turn a bogus string into an empty one
3615  void unBogus();
3616 
3617  // implements assignment operator, copy constructor, and fastCopyFrom()
3618  UnicodeString &copyFrom(const UnicodeString &src, UBool fastCopy=false);
3619 
3620  // Copies just the fields without memory management.
3621  void copyFieldsFrom(UnicodeString &src, UBool setSrcToBogus) U_NOEXCEPT;
3622 
3623  // Pin start and limit to acceptable values.
3624  inline void pinIndex(int32_t& start) const;
3625  inline void pinIndices(int32_t& start,
3626  int32_t& length) const;
3627 
3628 #if !UCONFIG_NO_CONVERSION
3629 
3630  /* Internal extract() using UConverter. */
3631  int32_t doExtract(int32_t start, int32_t length,
3632  char *dest, int32_t destCapacity,
3633  UConverter *cnv,
3634  UErrorCode &errorCode) const;
3635 
3636  /*
3637  * Real constructor for converting from codepage data.
3638  * It assumes that it is called with !fRefCounted.
3639  *
3640  * If `codepage==0`, then the default converter
3641  * is used for the platform encoding.
3642  * If `codepage` is an empty string (`""`),
3643  * then a simple conversion is performed on the codepage-invariant
3644  * subset ("invariant characters") of the platform encoding. See utypes.h.
3645  */
3646  void doCodepageCreate(const char *codepageData,
3647  int32_t dataLength,
3648  const char *codepage);
3649 
3650  /*
3651  * Worker function for creating a UnicodeString from
3652  * a codepage string using a UConverter.
3653  */
3654  void
3655  doCodepageCreate(const char *codepageData,
3656  int32_t dataLength,
3657  UConverter *converter,
3658  UErrorCode &status);
3659 
3660 #endif
3661 
3662  /*
3663  * This function is called when write access to the array
3664  * is necessary.
3665  *
3666  * We need to make a copy of the array if
3667  * the buffer is read-only, or
3668  * the buffer is refCounted (shared), and refCount>1, or
3669  * the buffer is too small.
3670  *
3671  * Return false if memory could not be allocated.
3672  */
3673  UBool cloneArrayIfNeeded(int32_t newCapacity = -1,
3674  int32_t growCapacity = -1,
3675  UBool doCopyArray = true,
3676  int32_t **pBufferToDelete = 0,
3677  UBool forceClone = false);
3678 
3684  UnicodeString &
3685  caseMap(int32_t caseLocale, uint32_t options,
3687  BreakIterator *iter,
3688 #endif
3689  UStringCaseMapper *stringCaseMapper);
3690 
3691  // ref counting
3692  void addRef(void);
3693  int32_t removeRef(void);
3694  int32_t refCount(void) const;
3695 
3696  // constants
3697  enum {
3703  US_STACKBUF_SIZE=(int32_t)(UNISTR_OBJECT_SIZE-sizeof(void *)-2)/U_SIZEOF_UCHAR,
3704  kInvalidUChar=0xffff, // U+FFFF returned by charAt(invalid index)
3705  kInvalidHashCode=0, // invalid hash code
3706  kEmptyHashCode=1, // hash code for empty string
3707 
3708  // bit flag values for fLengthAndFlags
3709  kIsBogus=1, // this string is bogus, i.e., not valid or NULL
3710  kUsingStackBuffer=2,// using fUnion.fStackFields instead of fUnion.fFields
3711  kRefCounted=4, // there is a refCount field before the characters in fArray
3712  kBufferIsReadonly=8,// do not write to this buffer
3713  kOpenGetBuffer=16, // getBuffer(minCapacity) was called (is "open"),
3714  // and releaseBuffer(newLength) must be called
3715  kAllStorageFlags=0x1f,
3716 
3717  kLengthShift=5, // remaining 11 bits for non-negative short length, or negative if long
3718  kLength1=1<<kLengthShift,
3719  kMaxShortLength=0x3ff, // max non-negative short length (leaves top bit 0)
3720  kLengthIsLarge=0xffe0, // short length < 0, real length is in fUnion.fFields.fLength
3721 
3722  // combined values for convenience
3723  kShortString=kUsingStackBuffer,
3724  kLongString=kRefCounted,
3725  kReadonlyAlias=kBufferIsReadonly,
3726  kWritableAlias=0
3727  };
3728 
3729  friend class UnicodeStringAppendable;
3730 
3731  union StackBufferOrFields; // forward declaration necessary before friend declaration
3732  friend union StackBufferOrFields; // make US_STACKBUF_SIZE visible inside fUnion
3733 
3734  /*
3735  * The following are all the class fields that are stored
3736  * in each UnicodeString object.
3737  * Note that UnicodeString has virtual functions,
3738  * therefore there is an implicit vtable pointer
3739  * as the first real field.
3740  * The fields should be aligned such that no padding is necessary.
3741  * On 32-bit machines, the size should be 32 bytes,
3742  * on 64-bit machines (8-byte pointers), it should be 40 bytes.
3743  *
3744  * We use a hack to achieve this.
3745  *
3746  * With at least some compilers, each of the following is forced to
3747  * a multiple of sizeof(pointer) [the largest field base unit here is a data pointer],
3748  * rounded up with additional padding if the fields do not already fit that requirement:
3749  * - sizeof(class UnicodeString)
3750  * - offsetof(UnicodeString, fUnion)
3751  * - sizeof(fUnion)
3752  * - sizeof(fStackFields)
3753  *
3754  * We optimize for the longest possible internal buffer for short strings.
3755  * fUnion.fStackFields begins with 2 bytes for storage flags
3756  * and the length of relatively short strings,
3757  * followed by the buffer for short string contents.
3758  * There is no padding inside fStackFields.
3759  *
3760  * Heap-allocated and aliased strings use fUnion.fFields.
3761  * Both fStackFields and fFields must begin with the same fields for flags and short length,
3762  * that is, those must have the same memory offsets inside the object,
3763  * because the flags must be inspected in order to decide which half of fUnion is being used.
3764  * We assume that the compiler does not reorder the fields.
3765  *
3766  * (Padding at the end of fFields is ok:
3767  * As long as it is no larger than fStackFields, it is not wasted space.)
3768  *
3769  * For some of the history of the UnicodeString class fields layout, see
3770  * - ICU ticket #11551 "longer UnicodeString contents in stack buffer"
3771  * - ICU ticket #11336 "UnicodeString: recombine stack buffer arrays"
3772  * - ICU ticket #8322 "why is sizeof(UnicodeString)==48?"
3773  */
3774  // (implicit) *vtable;
3775  union StackBufferOrFields {
3776  // fStackFields is used iff (fLengthAndFlags&kUsingStackBuffer) else fFields is used.
3777  // Each struct of the union must begin with fLengthAndFlags.
3778  struct {
3779  int16_t fLengthAndFlags; // bit fields: see constants above
3780  char16_t fBuffer[US_STACKBUF_SIZE]; // buffer for short strings
3781  } fStackFields;
3782  struct {
3783  int16_t fLengthAndFlags; // bit fields: see constants above
3784  int32_t fLength; // number of characters in fArray if >127; else undefined
3785  int32_t fCapacity; // capacity of fArray (in char16_ts)
3786  // array pointer last to minimize padding for machines with P128 data model
3787  // or pointer sizes that are not a power of 2
3788  char16_t *fArray; // the Unicode data
3789  } fFields;
3790  } fUnion;
3791 };
3792 
3801 U_COMMON_API UnicodeString U_EXPORT2
3803 
3804 //========================================
3805 // Inline members
3806 //========================================
3807 
3808 //========================================
3809 // Privates
3810 //========================================
3811 
3812 inline void
3813 UnicodeString::pinIndex(int32_t& start) const
3814 {
3815  // pin index
3816  if(start < 0) {
3817  start = 0;
3818  } else if(start > length()) {
3819  start = length();
3820  }
3821 }
3822 
3823 inline void
3824 UnicodeString::pinIndices(int32_t& start,
3825  int32_t& _length) const
3826 {
3827  // pin indices
3828  int32_t len = length();
3829  if(start < 0) {
3830  start = 0;
3831  } else if(start > len) {
3832  start = len;
3833  }
3834  if(_length < 0) {
3835  _length = 0;
3836  } else if(_length > (len - start)) {
3837  _length = (len - start);
3838  }
3839 }
3840 
3841 inline char16_t*
3842 UnicodeString::getArrayStart() {
3843  return (fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) ?
3844  fUnion.fStackFields.fBuffer : fUnion.fFields.fArray;
3845 }
3846 
3847 inline const char16_t*
3848 UnicodeString::getArrayStart() const {
3849  return (fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) ?
3850  fUnion.fStackFields.fBuffer : fUnion.fFields.fArray;
3851 }
3852 
3853 //========================================
3854 // Default constructor
3855 //========================================
3856 
3857 inline
3858 UnicodeString::UnicodeString() {
3859  fUnion.fStackFields.fLengthAndFlags=kShortString;
3860 }
3861 
3862 inline UnicodeString::UnicodeString(const std::nullptr_t /*text*/) {
3863  fUnion.fStackFields.fLengthAndFlags=kShortString;
3864 }
3865 
3866 inline UnicodeString::UnicodeString(const std::nullptr_t /*text*/, int32_t /*length*/) {
3867  fUnion.fStackFields.fLengthAndFlags=kShortString;
3868 }
3869 
3870 inline UnicodeString::UnicodeString(std::nullptr_t /*buffer*/, int32_t /*buffLength*/, int32_t /*buffCapacity*/) {
3871  fUnion.fStackFields.fLengthAndFlags=kShortString;
3872 }
3873 
3874 //========================================
3875 // Read-only implementation methods
3876 //========================================
3877 inline UBool
3878 UnicodeString::hasShortLength() const {
3879  return fUnion.fFields.fLengthAndFlags>=0;
3880 }
3881 
3882 inline int32_t
3883 UnicodeString::getShortLength() const {
3884  // fLengthAndFlags must be non-negative -> short length >= 0
3885  // and arithmetic or logical shift does not matter.
3886  return fUnion.fFields.fLengthAndFlags>>kLengthShift;
3887 }
3888 
3889 inline int32_t
3890 UnicodeString::length() const {
3891  return hasShortLength() ? getShortLength() : fUnion.fFields.fLength;
3892 }
3893 
3894 inline int32_t
3895 UnicodeString::getCapacity() const {
3896  return (fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) ?
3897  US_STACKBUF_SIZE : fUnion.fFields.fCapacity;
3898 }
3899 
3900 inline int32_t
3901 UnicodeString::hashCode() const
3902 { return doHashCode(); }
3903 
3904 inline UBool
3905 UnicodeString::isBogus() const
3906 { return (UBool)(fUnion.fFields.fLengthAndFlags & kIsBogus); }
3907 
3908 inline UBool
3909 UnicodeString::isWritable() const
3910 { return (UBool)!(fUnion.fFields.fLengthAndFlags&(kOpenGetBuffer|kIsBogus)); }
3911 
3912 inline UBool
3913 UnicodeString::isBufferWritable() const
3914 {
3915  return (UBool)(
3916  !(fUnion.fFields.fLengthAndFlags&(kOpenGetBuffer|kIsBogus|kBufferIsReadonly)) &&
3917  (!(fUnion.fFields.fLengthAndFlags&kRefCounted) || refCount()==1));
3918 }
3919 
3920 inline const char16_t *
3921 UnicodeString::getBuffer() const {
3922  if(fUnion.fFields.fLengthAndFlags&(kIsBogus|kOpenGetBuffer)) {
3923  return nullptr;
3924  } else if(fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) {
3925  return fUnion.fStackFields.fBuffer;
3926  } else {
3927  return fUnion.fFields.fArray;
3928  }
3929 }
3930 
3931 //========================================
3932 // Read-only alias methods
3933 //========================================
3934 inline int8_t
3935 UnicodeString::doCompare(int32_t start,
3936  int32_t thisLength,
3937  const UnicodeString& srcText,
3938  int32_t srcStart,
3939  int32_t srcLength) const
3940 {
3941  if(srcText.isBogus()) {
3942  return (int8_t)!isBogus(); // 0 if both are bogus, 1 otherwise
3943  } else {
3944  srcText.pinIndices(srcStart, srcLength);
3945  return doCompare(start, thisLength, srcText.getArrayStart(), srcStart, srcLength);
3946  }
3947 }
3948 
3949 inline UBool
3951 {
3952  if(isBogus()) {
3953  return text.isBogus();
3954  } else {
3955  int32_t len = length(), textLength = text.length();
3956  return !text.isBogus() && len == textLength && doEquals(text, len);
3957  }
3958 }
3959 
3960 inline UBool
3962 { return (! operator==(text)); }
3963 
3964 inline UBool
3965 UnicodeString::operator> (const UnicodeString& text) const
3966 { return doCompare(0, length(), text, 0, text.length()) == 1; }
3967 
3968 inline UBool
3969 UnicodeString::operator< (const UnicodeString& text) const
3970 { return doCompare(0, length(), text, 0, text.length()) == -1; }
3971 
3972 inline UBool
3973 UnicodeString::operator>= (const UnicodeString& text) const
3974 { return doCompare(0, length(), text, 0, text.length()) != -1; }
3975 
3976 inline UBool
3977 UnicodeString::operator<= (const UnicodeString& text) const
3978 { return doCompare(0, length(), text, 0, text.length()) != 1; }
3979 
3980 inline int8_t
3981 UnicodeString::compare(const UnicodeString& text) const
3982 { return doCompare(0, length(), text, 0, text.length()); }
3983 
3984 inline int8_t
3985 UnicodeString::compare(int32_t start,
3986  int32_t _length,
3987  const UnicodeString& srcText) const
3988 { return doCompare(start, _length, srcText, 0, srcText.length()); }
3989 
3990 inline int8_t
3991 UnicodeString::compare(ConstChar16Ptr srcChars,
3992  int32_t srcLength) const
3993 { return doCompare(0, length(), srcChars, 0, srcLength); }
3994 
3995 inline int8_t
3996 UnicodeString::compare(int32_t start,
3997  int32_t _length,
3998  const UnicodeString& srcText,
3999  int32_t srcStart,
4000  int32_t srcLength) const
4001 { return doCompare(start, _length, srcText, srcStart, srcLength); }
4002 
4003 inline int8_t
4004 UnicodeString::compare(int32_t start,
4005  int32_t _length,
4006  const char16_t *srcChars) const
4007 { return doCompare(start, _length, srcChars, 0, _length); }
4008 
4009 inline int8_t
4010 UnicodeString::compare(int32_t start,
4011  int32_t _length,
4012  const char16_t *srcChars,
4013  int32_t srcStart,
4014  int32_t srcLength) const
4015 { return doCompare(start, _length, srcChars, srcStart, srcLength); }
4016 
4017 inline int8_t
4018 UnicodeString::compareBetween(int32_t start,
4019  int32_t limit,
4020  const UnicodeString& srcText,
4021  int32_t srcStart,
4022  int32_t srcLimit) const
4023 { return doCompare(start, limit - start,
4024  srcText, srcStart, srcLimit - srcStart); }
4025 
4026 inline int8_t
4027 UnicodeString::doCompareCodePointOrder(int32_t start,
4028  int32_t thisLength,
4029  const UnicodeString& srcText,
4030  int32_t srcStart,
4031  int32_t srcLength) const
4032 {
4033  if(srcText.isBogus()) {
4034  return (int8_t)!isBogus(); // 0 if both are bogus, 1 otherwise
4035  } else {
4036  srcText.pinIndices(srcStart, srcLength);
4037  return doCompareCodePointOrder(start, thisLength, srcText.getArrayStart(), srcStart, srcLength);
4038  }
4039 }
4040 
4041 inline int8_t
4042 UnicodeString::compareCodePointOrder(const UnicodeString& text) const
4043 { return doCompareCodePointOrder(0, length(), text, 0, text.length()); }
4044 
4045 inline int8_t
4046 UnicodeString::compareCodePointOrder(int32_t start,
4047  int32_t _length,
4048  const UnicodeString& srcText) const
4049 { return doCompareCodePointOrder(start, _length, srcText, 0, srcText.length()); }
4050 
4051 inline int8_t
4052 UnicodeString::compareCodePointOrder(ConstChar16Ptr srcChars,
4053  int32_t srcLength) const
4054 { return doCompareCodePointOrder(0, length(), srcChars, 0, srcLength); }
4055 
4056 inline int8_t
4057 UnicodeString::compareCodePointOrder(int32_t start,
4058  int32_t _length,
4059  const UnicodeString& srcText,
4060  int32_t srcStart,
4061  int32_t srcLength) const
4062 { return doCompareCodePointOrder(start, _length, srcText, srcStart, srcLength); }
4063 
4064 inline int8_t
4065 UnicodeString::compareCodePointOrder(int32_t start,
4066  int32_t _length,
4067  const char16_t *srcChars) const
4068 { return doCompareCodePointOrder(start, _length, srcChars, 0, _length); }
4069 
4070 inline int8_t
4071 UnicodeString::compareCodePointOrder(int32_t start,
4072  int32_t _length,
4073  const char16_t *srcChars,
4074  int32_t srcStart,
4075  int32_t srcLength) const
4076 { return doCompareCodePointOrder(start, _length, srcChars, srcStart, srcLength); }
4077 
4078 inline int8_t
4079 UnicodeString::compareCodePointOrderBetween(int32_t start,
4080  int32_t limit,
4081  const UnicodeString& srcText,
4082  int32_t srcStart,
4083  int32_t srcLimit) const
4084 { return doCompareCodePointOrder(start, limit - start,
4085  srcText, srcStart, srcLimit - srcStart); }
4086 
4087 inline int8_t
4088 UnicodeString::doCaseCompare(int32_t start,
4089  int32_t thisLength,
4090  const UnicodeString &srcText,
4091  int32_t srcStart,
4092  int32_t srcLength,
4093  uint32_t options) const
4094 {
4095  if(srcText.isBogus()) {
4096  return (int8_t)!isBogus(); // 0 if both are bogus, 1 otherwise
4097  } else {
4098  srcText.pinIndices(srcStart, srcLength);
4099  return doCaseCompare(start, thisLength, srcText.getArrayStart(), srcStart, srcLength, options);
4100  }
4101 }
4102 
4103 inline int8_t
4104 UnicodeString::caseCompare(const UnicodeString &text, uint32_t options) const {
4105  return doCaseCompare(0, length(), text, 0, text.length(), options);
4106 }
4107 
4108 inline int8_t
4109 UnicodeString::caseCompare(int32_t start,
4110  int32_t _length,
4111  const UnicodeString &srcText,
4112  uint32_t options) const {
4113  return doCaseCompare(start, _length, srcText, 0, srcText.length(), options);
4114 }
4115 
4116 inline int8_t
4117 UnicodeString::caseCompare(ConstChar16Ptr srcChars,
4118  int32_t srcLength,
4119  uint32_t options) const {
4120  return doCaseCompare(0, length(), srcChars, 0, srcLength, options);
4121 }
4122 
4123 inline int8_t
4124 UnicodeString::caseCompare(int32_t start,
4125  int32_t _length,
4126  const UnicodeString &srcText,
4127  int32_t srcStart,
4128  int32_t srcLength,
4129  uint32_t options) const {
4130  return doCaseCompare(start, _length, srcText, srcStart, srcLength, options);
4131 }
4132 
4133 inline int8_t
4134 UnicodeString::caseCompare(int32_t start,
4135  int32_t _length,
4136  const char16_t *srcChars,
4137  uint32_t options) const {
4138  return doCaseCompare(start, _length, srcChars, 0, _length, options);
4139 }
4140 
4141 inline int8_t
4142 UnicodeString::caseCompare(int32_t start,
4143  int32_t _length,
4144  const char16_t *srcChars,
4145  int32_t srcStart,
4146  int32_t srcLength,
4147  uint32_t options) const {
4148  return doCaseCompare(start, _length, srcChars, srcStart, srcLength, options);
4149 }
4150 
4151 inline int8_t
4152 UnicodeString::caseCompareBetween(int32_t start,
4153  int32_t limit,
4154  const UnicodeString &srcText,
4155  int32_t srcStart,
4156  int32_t srcLimit,
4157  uint32_t options) const {
4158  return doCaseCompare(start, limit - start, srcText, srcStart, srcLimit - srcStart, options);
4159 }
4160 
4161 inline int32_t
4162 UnicodeString::indexOf(const UnicodeString& srcText,
4163  int32_t srcStart,
4164  int32_t srcLength,
4165  int32_t start,
4166  int32_t _length) const
4167 {
4168  if(!srcText.isBogus()) {
4169  srcText.pinIndices(srcStart, srcLength);
4170  if(srcLength > 0) {
4171  return indexOf(srcText.getArrayStart(), srcStart, srcLength, start, _length);
4172  }
4173  }
4174  return -1;
4175 }
4176 
4177 inline int32_t
4178 UnicodeString::indexOf(const UnicodeString& text) const
4179 { return indexOf(text, 0, text.length(), 0, length()); }
4180 
4181 inline int32_t
4182 UnicodeString::indexOf(const UnicodeString& text,
4183  int32_t start) const {
4184  pinIndex(start);
4185  return indexOf(text, 0, text.length(), start, length() - start);
4186 }
4187 
4188 inline int32_t
4189 UnicodeString::indexOf(const UnicodeString& text,
4190  int32_t start,
4191  int32_t _length) const
4192 { return indexOf(text, 0, text.length(), start, _length); }
4193 
4194 inline int32_t
4195 UnicodeString::indexOf(const char16_t *srcChars,
4196  int32_t srcLength,
4197  int32_t start) const {
4198  pinIndex(start);
4199  return indexOf(srcChars, 0, srcLength, start, length() - start);
4200 }
4201 
4202 inline int32_t
4203 UnicodeString::indexOf(ConstChar16Ptr srcChars,
4204  int32_t srcLength,
4205  int32_t start,
4206  int32_t _length) const
4207 { return indexOf(srcChars, 0, srcLength, start, _length); }
4208 
4209 inline int32_t
4210 UnicodeString::indexOf(char16_t c,
4211  int32_t start,
4212  int32_t _length) const
4213 { return doIndexOf(c, start, _length); }
4214 
4215 inline int32_t
4216 UnicodeString::indexOf(UChar32 c,
4217  int32_t start,
4218  int32_t _length) const
4219 { return doIndexOf(c, start, _length); }
4220 
4221 inline int32_t
4222 UnicodeString::indexOf(char16_t c) const
4223 { return doIndexOf(c, 0, length()); }
4224 
4225 inline int32_t
4226 UnicodeString::indexOf(UChar32 c) const
4227 { return indexOf(c, 0, length()); }
4228 
4229 inline int32_t
4230 UnicodeString::indexOf(char16_t c,
4231  int32_t start) const {
4232  pinIndex(start);
4233  return doIndexOf(c, start, length() - start);
4234 }
4235 
4236 inline int32_t
4237 UnicodeString::indexOf(UChar32 c,
4238  int32_t start) const {
4239  pinIndex(start);
4240  return indexOf(c, start, length() - start);
4241 }
4242 
4243 inline int32_t
4244 UnicodeString::lastIndexOf(ConstChar16Ptr srcChars,
4245  int32_t srcLength,
4246  int32_t start,
4247  int32_t _length) const
4248 { return lastIndexOf(srcChars, 0, srcLength, start, _length); }
4249 
4250 inline int32_t
4251 UnicodeString::lastIndexOf(const char16_t *srcChars,
4252  int32_t srcLength,
4253  int32_t start) const {
4254  pinIndex(start);
4255  return lastIndexOf(srcChars, 0, srcLength, start, length() - start);
4256 }
4257 
4258 inline int32_t
4259 UnicodeString::lastIndexOf(const UnicodeString& srcText,
4260  int32_t srcStart,
4261  int32_t srcLength,
4262  int32_t start,
4263  int32_t _length) const
4264 {
4265  if(!srcText.isBogus()) {
4266  srcText.pinIndices(srcStart, srcLength);
4267  if(srcLength > 0) {
4268  return lastIndexOf(srcText.getArrayStart(), srcStart, srcLength, start, _length);
4269  }
4270  }
4271  return -1;
4272 }
4273 
4274 inline int32_t
4275 UnicodeString::lastIndexOf(const UnicodeString& text,
4276  int32_t start,
4277  int32_t _length) const
4278 { return lastIndexOf(text, 0, text.length(), start, _length); }
4279 
4280 inline int32_t
4281 UnicodeString::lastIndexOf(const UnicodeString& text,
4282  int32_t start) const {
4283  pinIndex(start);
4284  return lastIndexOf(text, 0, text.length(), start, length() - start);
4285 }
4286 
4287 inline int32_t
4288 UnicodeString::lastIndexOf(const UnicodeString& text) const
4289 { return lastIndexOf(text, 0, text.length(), 0, length()); }
4290 
4291 inline int32_t
4292 UnicodeString::lastIndexOf(char16_t c,
4293  int32_t start,
4294  int32_t _length) const
4295 { return doLastIndexOf(c, start, _length); }
4296 
4297 inline int32_t
4298 UnicodeString::lastIndexOf(UChar32 c,
4299  int32_t start,
4300  int32_t _length) const {
4301  return doLastIndexOf(c, start, _length);
4302 }
4303 
4304 inline int32_t
4305 UnicodeString::lastIndexOf(char16_t c) const
4306 { return doLastIndexOf(c, 0, length()); }
4307 
4308 inline int32_t
4309 UnicodeString::lastIndexOf(UChar32 c) const {
4310  return lastIndexOf(c, 0, length());
4311 }
4312 
4313 inline int32_t
4314 UnicodeString::lastIndexOf(char16_t c,
4315  int32_t start) const {
4316  pinIndex(start);
4317  return doLastIndexOf(c, start, length() - start);
4318 }
4319 
4320 inline int32_t
4321 UnicodeString::lastIndexOf(UChar32 c,
4322  int32_t start) const {
4323  pinIndex(start);
4324  return lastIndexOf(c, start, length() - start);
4325 }
4326 
4327 inline UBool
4328 UnicodeString::startsWith(const UnicodeString& text) const
4329 { return compare(0, text.length(), text, 0, text.length()) == 0; }
4330 
4331 inline UBool
4332 UnicodeString::startsWith(const UnicodeString& srcText,
4333  int32_t srcStart,
4334  int32_t srcLength) const
4335 { return doCompare(0, srcLength, srcText, srcStart, srcLength) == 0; }
4336 
4337 inline UBool
4338 UnicodeString::startsWith(ConstChar16Ptr srcChars, int32_t srcLength) const {
4339  if(srcLength < 0) {
4340  srcLength = u_strlen(toUCharPtr(srcChars));
4341  }
4342  return doCompare(0, srcLength, srcChars, 0, srcLength) == 0;
4343 }
4344 
4345 inline UBool
4346 UnicodeString::startsWith(const char16_t *srcChars, int32_t srcStart, int32_t srcLength) const {
4347  if(srcLength < 0) {
4348  srcLength = u_strlen(toUCharPtr(srcChars));
4349  }
4350  return doCompare(0, srcLength, srcChars, srcStart, srcLength) == 0;
4351 }
4352 
4353 inline UBool
4354 UnicodeString::endsWith(const UnicodeString& text) const
4355 { return doCompare(length() - text.length(), text.length(),
4356  text, 0, text.length()) == 0; }
4357 
4358 inline UBool
4359 UnicodeString::endsWith(const UnicodeString& srcText,
4360  int32_t srcStart,
4361  int32_t srcLength) const {
4362  srcText.pinIndices(srcStart, srcLength);
4363  return doCompare(length() - srcLength, srcLength,
4364  srcText, srcStart, srcLength) == 0;
4365 }
4366 
4367 inline UBool
4368 UnicodeString::endsWith(ConstChar16Ptr srcChars,
4369  int32_t srcLength) const {
4370  if(srcLength < 0) {
4371  srcLength = u_strlen(toUCharPtr(srcChars));
4372  }
4373  return doCompare(length() - srcLength, srcLength,
4374  srcChars, 0, srcLength) == 0;
4375 }
4376 
4377 inline UBool
4378 UnicodeString::endsWith(const char16_t *srcChars,
4379  int32_t srcStart,
4380  int32_t srcLength) const {
4381  if(srcLength < 0) {
4382  srcLength = u_strlen(toUCharPtr(srcChars + srcStart));
4383  }
4384  return doCompare(length() - srcLength, srcLength,
4385  srcChars, srcStart, srcLength) == 0;
4386 }
4387 
4388 //========================================
4389 // replace
4390 //========================================
4391 inline UnicodeString&
4392 UnicodeString::replace(int32_t start,
4393  int32_t _length,
4394  const UnicodeString& srcText)
4395 { return doReplace(start, _length, srcText, 0, srcText.length()); }
4396 
4397 inline UnicodeString&
4398 UnicodeString::replace(int32_t start,
4399  int32_t _length,
4400  const UnicodeString& srcText,
4401  int32_t srcStart,
4402  int32_t srcLength)
4403 { return doReplace(start, _length, srcText, srcStart, srcLength); }
4404 
4405 inline UnicodeString&
4406 UnicodeString::replace(int32_t start,
4407  int32_t _length,
4408  ConstChar16Ptr srcChars,
4409  int32_t srcLength)
4410 { return doReplace(start, _length, srcChars, 0, srcLength); }
4411 
4412 inline UnicodeString&
4413 UnicodeString::replace(int32_t start,
4414  int32_t _length,
4415  const char16_t *srcChars,
4416  int32_t srcStart,
4417  int32_t srcLength)
4418 { return doReplace(start, _length, srcChars, srcStart, srcLength); }
4419 
4420 inline UnicodeString&
4421 UnicodeString::replace(int32_t start,
4422  int32_t _length,
4423  char16_t srcChar)
4424 { return doReplace(start, _length, &srcChar, 0, 1); }
4425 
4426 inline UnicodeString&
4427 UnicodeString::replaceBetween(int32_t start,
4428  int32_t limit,
4429  const UnicodeString& srcText)
4430 { return doReplace(start, limit - start, srcText, 0, srcText.length()); }
4431 
4432 inline UnicodeString&
4433 UnicodeString::replaceBetween(int32_t start,
4434  int32_t limit,
4435  const UnicodeString& srcText,
4436  int32_t srcStart,
4437  int32_t srcLimit)
4438 { return doReplace(start, limit - start, srcText, srcStart, srcLimit - srcStart); }
4439 
4440 inline UnicodeString&
4441 UnicodeString::findAndReplace(const UnicodeString& oldText,
4442  const UnicodeString& newText)
4443 { return findAndReplace(0, length(), oldText, 0, oldText.length(),
4444  newText, 0, newText.length()); }
4445 
4446 inline UnicodeString&
4447 UnicodeString::findAndReplace(int32_t start,
4448  int32_t _length,
4449  const UnicodeString& oldText,
4450  const UnicodeString& newText)
4451 { return findAndReplace(start, _length, oldText, 0, oldText.length(),
4452  newText, 0, newText.length()); }
4453 
4454 // ============================
4455 // extract
4456 // ============================
4457 inline void
4458 UnicodeString::doExtract(int32_t start,
4459  int32_t _length,
4460  UnicodeString& target) const
4461 { target.replace(0, target.length(), *this, start, _length); }
4462 
4463 inline void
4464 UnicodeString::extract(int32_t start,
4465  int32_t _length,
4466  Char16Ptr target,
4467  int32_t targetStart) const
4468 { doExtract(start, _length, target, targetStart); }
4469 
4470 inline void
4471 UnicodeString::extract(int32_t start,
4472  int32_t _length,
4473  UnicodeString& target) const
4474 { doExtract(start, _length, target); }
4475 
4476 #if !UCONFIG_NO_CONVERSION
4477 
4478 inline int32_t
4479 UnicodeString::extract(int32_t start,
4480  int32_t _length,
4481  char *dst,
4482  const char *codepage) const
4483 
4484 {
4485  // This dstSize value will be checked explicitly
4486  return extract(start, _length, dst, dst!=0 ? 0xffffffff : 0, codepage);
4487 }
4488 
4489 #endif
4490 
4491 inline void
4492 UnicodeString::extractBetween(int32_t start,
4493  int32_t limit,
4494  char16_t *dst,
4495  int32_t dstStart) const {
4496  pinIndex(start);
4497  pinIndex(limit);
4498  doExtract(start, limit - start, dst, dstStart);
4499 }
4500 
4501 inline UnicodeString
4502 UnicodeString::tempSubStringBetween(int32_t start, int32_t limit) const {
4503  return tempSubString(start, limit - start);
4504 }
4505 
4506 inline char16_t
4507 UnicodeString::doCharAt(int32_t offset) const
4508 {
4509  if((uint32_t)offset < (uint32_t)length()) {
4510  return getArrayStart()[offset];
4511  } else {
4512  return kInvalidUChar;
4513  }
4514 }
4515 
4516 inline char16_t
4517 UnicodeString::charAt(int32_t offset) const
4518 { return doCharAt(offset); }
4519 
4520 inline char16_t
4521 UnicodeString::operator[] (int32_t offset) const
4522 { return doCharAt(offset); }
4523 
4524 inline UBool
4525 UnicodeString::isEmpty() const {
4526  // Arithmetic or logical right shift does not matter: only testing for 0.
4527  return (fUnion.fFields.fLengthAndFlags>>kLengthShift) == 0;
4528 }
4529 
4530 //========================================
4531 // Write implementation methods
4532 //========================================
4533 inline void
4534 UnicodeString::setZeroLength() {
4535  fUnion.fFields.fLengthAndFlags &= kAllStorageFlags;
4536 }
4537 
4538 inline void
4539 UnicodeString::setShortLength(int32_t len) {
4540  // requires 0 <= len <= kMaxShortLength
4541  fUnion.fFields.fLengthAndFlags =
4542  (int16_t)((fUnion.fFields.fLengthAndFlags & kAllStorageFlags) | (len << kLengthShift));
4543 }
4544 
4545 inline void
4546 UnicodeString::setLength(int32_t len) {
4547  if(len <= kMaxShortLength) {
4548  setShortLength(len);
4549  } else {
4550  fUnion.fFields.fLengthAndFlags |= kLengthIsLarge;
4551  fUnion.fFields.fLength = len;
4552  }
4553 }
4554 
4555 inline void
4556 UnicodeString::setToEmpty() {
4557  fUnion.fFields.fLengthAndFlags = kShortString;
4558 }
4559 
4560 inline void
4561 UnicodeString::setArray(char16_t *array, int32_t len, int32_t capacity) {
4562  setLength(len);
4563  fUnion.fFields.fArray = array;
4564  fUnion.fFields.fCapacity = capacity;
4565 }
4566 
4567 inline UnicodeString&
4568 UnicodeString::operator= (char16_t ch)
4569 { return doReplace(0, length(), &ch, 0, 1); }
4570 
4571 inline UnicodeString&
4572 UnicodeString::operator= (UChar32 ch)
4573 { return replace(0, length(), ch); }
4574 
4575 inline UnicodeString&
4576 UnicodeString::setTo(const UnicodeString& srcText,
4577  int32_t srcStart,
4578  int32_t srcLength)
4579 {
4580  unBogus();
4581  return doReplace(0, length(), srcText, srcStart, srcLength);
4582 }
4583 
4584 inline UnicodeString&
4585 UnicodeString::setTo(const UnicodeString& srcText,
4586  int32_t srcStart)
4587 {
4588  unBogus();
4589  srcText.pinIndex(srcStart);
4590  return doReplace(0, length(), srcText, srcStart, srcText.length() - srcStart);
4591 }
4592 
4593 inline UnicodeString&
4594 UnicodeString::setTo(const UnicodeString& srcText)
4595 {
4596  return copyFrom(srcText);
4597 }
4598 
4599 inline UnicodeString&
4600 UnicodeString::setTo(const char16_t *srcChars,
4601  int32_t srcLength)
4602 {
4603  unBogus();
4604  return doReplace(0, length(), srcChars, 0, srcLength);
4605 }
4606 
4607 inline UnicodeString&
4608 UnicodeString::setTo(char16_t srcChar)
4609 {
4610  unBogus();
4611  return doReplace(0, length(), &srcChar, 0, 1);
4612 }
4613 
4614 inline UnicodeString&
4615 UnicodeString::setTo(UChar32 srcChar)
4616 {
4617  unBogus();
4618  return replace(0, length(), srcChar);
4619 }
4620 
4621 inline UnicodeString&
4622 UnicodeString::append(const UnicodeString& srcText,
4623  int32_t srcStart,
4624  int32_t srcLength)
4625 { return doAppend(srcText, srcStart, srcLength); }
4626 
4627 inline UnicodeString&
4628 UnicodeString::append(const UnicodeString& srcText)
4629 { return doAppend(srcText, 0, srcText.length()); }
4630 
4631 inline UnicodeString&
4632 UnicodeString::append(const char16_t *srcChars,
4633  int32_t srcStart,
4634  int32_t srcLength)
4635 { return doAppend(srcChars, srcStart, srcLength); }
4636 
4637 inline UnicodeString&
4638 UnicodeString::append(ConstChar16Ptr srcChars,
4639  int32_t srcLength)
4640 { return doAppend(srcChars, 0, srcLength); }
4641 
4642 inline UnicodeString&
4643 UnicodeString::append(char16_t srcChar)
4644 { return doAppend(&srcChar, 0, 1); }
4645 
4646 inline UnicodeString&
4647 UnicodeString::operator+= (char16_t ch)
4648 { return doAppend(&ch, 0, 1); }
4649 
4650 inline UnicodeString&
4651 UnicodeString::operator+= (UChar32 ch) {
4652  return append(ch);
4653 }
4654 
4655 inline UnicodeString&
4656 UnicodeString::operator+= (const UnicodeString& srcText)
4657 { return doAppend(srcText, 0, srcText.length()); }
4658 
4659 inline UnicodeString&
4660 UnicodeString::insert(int32_t start,
4661  const UnicodeString& srcText,
4662  int32_t srcStart,
4663  int32_t srcLength)
4664 { return doReplace(start, 0, srcText, srcStart, srcLength); }
4665 
4666 inline UnicodeString&
4667 UnicodeString::insert(int32_t start,
4668  const UnicodeString& srcText)
4669 { return doReplace(start, 0, srcText, 0, srcText.length()); }
4670 
4671 inline UnicodeString&
4672 UnicodeString::insert(int32_t start,
4673  const char16_t *srcChars,
4674  int32_t srcStart,
4675  int32_t srcLength)
4676 { return doReplace(start, 0, srcChars, srcStart, srcLength); }
4677 
4678 inline UnicodeString&
4679 UnicodeString::insert(int32_t start,
4680  ConstChar16Ptr srcChars,
4681  int32_t srcLength)
4682 { return doReplace(start, 0, srcChars, 0, srcLength); }
4683 
4684 inline UnicodeString&
4685 UnicodeString::insert(int32_t start,
4686  char16_t srcChar)
4687 { return doReplace(start, 0, &srcChar, 0, 1); }
4688 
4689 inline UnicodeString&
4690 UnicodeString::insert(int32_t start,
4691  UChar32 srcChar)
4692 { return replace(start, 0, srcChar); }
4693 
4694 
4695 inline UnicodeString&
4696 UnicodeString::remove()
4697 {
4698  // remove() of a bogus string makes the string empty and non-bogus
4699  if(isBogus()) {
4700  setToEmpty();
4701  } else {
4702  setZeroLength();
4703  }
4704  return *this;
4705 }
4706 
4707 inline UnicodeString&
4708 UnicodeString::remove(int32_t start,
4709  int32_t _length)
4710 {
4711  if(start <= 0 && _length == INT32_MAX) {
4712  // remove(guaranteed everything) of a bogus string makes the string empty and non-bogus
4713  return remove();
4714  }
4715  return doReplace(start, _length, NULL, 0, 0);
4716 }
4717 
4718 inline UnicodeString&
4719 UnicodeString::removeBetween(int32_t start,
4720  int32_t limit)
4721 { return doReplace(start, limit - start, NULL, 0, 0); }
4722 
4723 inline UnicodeString &
4724 UnicodeString::retainBetween(int32_t start, int32_t limit) {
4725  truncate(limit);
4726  return doReplace(0, start, NULL, 0, 0);
4727 }
4728 
4729 inline UBool
4730 UnicodeString::truncate(int32_t targetLength)
4731 {
4732  if(isBogus() && targetLength == 0) {
4733  // truncate(0) of a bogus string makes the string empty and non-bogus
4734  unBogus();
4735  return false;
4736  } else if((uint32_t)targetLength < (uint32_t)length()) {
4737  setLength(targetLength);
4738  return true;
4739  } else {
4740  return false;
4741  }
4742 }
4743 
4744 inline UnicodeString&
4745 UnicodeString::reverse()
4746 { return doReverse(0, length()); }
4747 
4748 inline UnicodeString&
4749 UnicodeString::reverse(int32_t start,
4750  int32_t _length)
4751 { return doReverse(start, _length); }
4752 
4753 U_NAMESPACE_END
4754 
4755 #endif /* U_SHOW_CPLUSPLUS_API */
4756 
4757 #endif
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.
Definition: brkiter.h:106
A ByteSink can be filled with bytes.
Definition: bytestream.h:53
char16_t * wrapper with implicit conversion from distinct but bit-compatible pointer types.
Definition: char16ptr.h:42
const char16_t * wrapper with implicit conversion from distinct but bit-compatible pointer types.
Definition: char16ptr.h:149
Records lengths of string edits but not replacement text.
Definition: edits.h:80
A Locale object represents a specific geographical, political, or cultural region.
Definition: locid.h:195
Replaceable is an abstract base class representing a string of characters that supports the replaceme...
Definition: rep.h:77
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.
Definition: rep.h:251
int32_t length() const
Returns the number of 16-bit code units in the text.
Definition: rep.h:246
Implementation of ByteSink that writes to a "string".
Definition: bytestream.h:267
A string-like object that points to a sized piece of memory.
Definition: stringpiece.h:60
An Appendable implementation which writes to a UnicodeString.
Definition: appendable.h:156
UnicodeString is a string class that stores Unicode characters directly and provides similar function...
Definition: unistr.h:296
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.
UnicodeString & operator=(UnicodeString &&src) U_NOEXCEPT
Move assignment operator; might leave src in bogus state.
virtual UChar32 getChar32At(int32_t offset) const
The change in Replaceable to use virtual getChar32At() allows UnicodeString::char32At() to be inline ...
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.
Definition: unistr.h:3060
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.
virtual void extractBetween(int32_t start, int32_t limit, UnicodeString &target) const
Copy the characters in the range [start, limit) into the UnicodeString target.
virtual ~UnicodeString()
Destructor.
UnicodeString(const char *codepageData, int32_t dataLength, const char *codepage)
char* constructor.
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 ...
Definition: unistr.h:307
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.
Definition: unistr.h:2997
UnicodeString & trim(void)
Trims leading and trailing whitespace from this UnicodeString.
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 int32_t getLength() const
Implement Replaceable::getLength() (see jitterbug 1027).
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...
Definition: unistr.h:4398
virtual UnicodeString * clone() const
Clone this object, an instance of a subclass of Replaceable.
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.
Definition: unistr.h:1729
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.
int32_t length(void) const
Return the length of the UnicodeString object.
Definition: unistr.h:3890
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 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.
UnicodeString & toUpper(void)
Convert the characters in this to UPPER CASE following the conventions of the default locale.
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.
Definition: unistr.h:3142
void setToBogus()
Make this UnicodeString object invalid.
virtual void handleReplaceBetween(int32_t start, int32_t limit, const UnicodeString &text)
Replace a substring of this object with the given text.
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.
virtual UBool hasMetaData() const
Replaceable API.
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.
virtual void copy(int32_t start, int32_t limit, int32_t dest)
Copy a substring of this object, retaining attribute (out-of-band) information.
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.
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(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.
Definition: unistr.h:3128
int32_t toUTF32(UChar32 *utf32, int32_t capacity, UErrorCode &errorCode) const
Convert the UnicodeString to UTF-32.
const char16_t * getTerminatedBuffer()
Get a read-only pointer to the internal buffer, making sure that it is NUL-terminated.
UnicodeString & toLower(void)
Convert the characters in this to lower case following the conventions of the default locale.
UnicodeString(const uint16_t *text, int32_t textLength)
uint16_t * constructor.
Definition: unistr.h:3047
UNISTR_FROM_STRING_EXPLICIT UnicodeString(const char16_t *text)
char16_t* constructor.
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.
Definition: unistr.h:3013
virtual char16_t getCharAt(int32_t offset) const
The change in Replaceable to use virtual getCharAt() allows UnicodeString::charAt() to be inline agai...
UBool isBogus(void) const
Determine if this object contains a valid string.
Definition: unistr.h:3905
UnicodeString & setCharAt(int32_t offset, char16_t ch)
Set the character at the specified offset to the specified character.
friend void swap(UnicodeString &s1, UnicodeString &s2) U_NOEXCEPT
Non-member UnicodeString swap function.
Definition: unistr.h:1925
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.
UnicodeString(UnicodeString &&src) U_NOEXCEPT
Move constructor; might leave src in bogus state.
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 swap(UnicodeString &other) U_NOEXCEPT
Swap strings.
void toUTF8(ByteSink &sink) const
Convert the UnicodeString to UTF-8 and write the result to a ByteSink.
virtual UClassID getDynamicClassID() const
ICU "poor man's RTTI", returns a UClassID for the actual class.
UnicodeString & toUpper(const Locale &locale)
Convert the characters in this to UPPER CASE following the conventions of a specific locale.
U_CAPI int32_t u_strlen(const UChar *s)
U_EXPORT UBool 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 *.
Definition: char16ptr.h:260
U_COMMON_API UnicodeString operator+(const UnicodeString &s1, const UnicodeString &s2)
Create a new UnicodeString with the concatenation of two others.
UBool operator!=(const StringPiece &x, const StringPiece &y)
Global operator != for StringPiece.
Definition: stringpiece.h:335
#define U_CALLCONV
Similar to U_CDECL_BEGIN/U_CDECL_END, this qualifier is necessary in callback function typedefs to ma...
Definition: platform.h:870
#define U_NOEXCEPT
"noexcept" if supported, otherwise empty.
Definition: platform.h:529
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
Definition: ucnv_err.h:96
#define UCONFIG_NO_BREAK_ITERATION
This switch turns off break iteration.
Definition: uconfig.h:348
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:467
#define INT32_MAX
The largest value a 32 bit signed integer can hold.
Definition: umachine.h:208
int8_t UBool
The ICU boolean type, a signed-byte integer.
Definition: umachine.h:269
#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:418
#define U_SIZEOF_UCHAR
Number of bytes in a UChar.
Definition: umachine.h:358
#define UNISTR_FROM_CHAR_EXPLICIT
This can be defined to be empty or "explicit".
Definition: unistr.h:150
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.
Definition: unistr.h:70
#define UNISTR_FROM_STRING_EXPLICIT
This can be defined to be empty or "explicit".
Definition: unistr.h:170
#define UNISTR_OBJECT_SIZE
Desired sizeof(UnicodeString) in bytes.
Definition: unistr.h:208
void * UClassID
UClassID is used to identify classes without using the compiler's RTTI.
Definition: uobject.h:96
Basic definitions for ICU, for both C and C++ APIs.
#define NULL
Define NULL if necessary, to nullptr for C++ and to ((void *)0) for C.
Definition: utypes.h:188
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition: utypes.h:415
#define U_COMMON_API
Set to export library symbols from inside the common library, and to import them from outside.
Definition: utypes.h:300