File: | out/../deps/icu-small/source/common/umutex.h |
Warning: | line 89, column 12 Called C++ object pointer is null |
Press '?' to see keyboard shortcuts
Keyboard shortcuts:
1 | // © 2016 and later: Unicode, Inc. and others. | ||||
2 | // License & terms of use: http://www.unicode.org/copyright.html | ||||
3 | /* | ||||
4 | ****************************************************************************** | ||||
5 | * Copyright (C) 1999-2016, International Business Machines Corporation and | ||||
6 | * others. All Rights Reserved. | ||||
7 | ****************************************************************************** | ||||
8 | * | ||||
9 | * File unistr.cpp | ||||
10 | * | ||||
11 | * Modification History: | ||||
12 | * | ||||
13 | * Date Name Description | ||||
14 | * 09/25/98 stephen Creation. | ||||
15 | * 04/20/99 stephen Overhauled per 4/16 code review. | ||||
16 | * 07/09/99 stephen Renamed {hi,lo},{byte,word} to icu_X for HP/UX | ||||
17 | * 11/18/99 aliu Added handleReplaceBetween() to make inherit from | ||||
18 | * Replaceable. | ||||
19 | * 06/25/01 grhoten Removed the dependency on iostream | ||||
20 | ****************************************************************************** | ||||
21 | */ | ||||
22 | |||||
23 | #include "unicode/utypes.h" | ||||
24 | #include "unicode/appendable.h" | ||||
25 | #include "unicode/putil.h" | ||||
26 | #include "cstring.h" | ||||
27 | #include "cmemory.h" | ||||
28 | #include "unicode/ustring.h" | ||||
29 | #include "unicode/unistr.h" | ||||
30 | #include "unicode/utf.h" | ||||
31 | #include "unicode/utf16.h" | ||||
32 | #include "uelement.h" | ||||
33 | #include "ustr_imp.h" | ||||
34 | #include "umutex.h" | ||||
35 | #include "uassert.h" | ||||
36 | |||||
37 | #if 0 | ||||
38 | |||||
39 | #include <iostream> | ||||
40 | using namespace std; | ||||
41 | |||||
42 | //DEBUGGING | ||||
43 | void | ||||
44 | print(const UnicodeString& s, | ||||
45 | const char *name) | ||||
46 | { | ||||
47 | UChar c; | ||||
48 | cout << name << ":|"; | ||||
49 | for(int i = 0; i < s.length(); ++i) { | ||||
50 | c = s[i]; | ||||
51 | if(c>= 0x007E || c < 0x0020) | ||||
52 | cout << "[0x" << hex << s[i] << "]"; | ||||
53 | else | ||||
54 | cout << (char) s[i]; | ||||
55 | } | ||||
56 | cout << '|' << endl; | ||||
57 | } | ||||
58 | |||||
59 | void | ||||
60 | print(const UChar *s, | ||||
61 | int32_t len, | ||||
62 | const char *name) | ||||
63 | { | ||||
64 | UChar c; | ||||
65 | cout << name << ":|"; | ||||
66 | for(int i = 0; i < len; ++i) { | ||||
67 | c = s[i]; | ||||
68 | if(c>= 0x007E || c < 0x0020) | ||||
69 | cout << "[0x" << hex << s[i] << "]"; | ||||
70 | else | ||||
71 | cout << (char) s[i]; | ||||
72 | } | ||||
73 | cout << '|' << endl; | ||||
74 | } | ||||
75 | // END DEBUGGING | ||||
76 | #endif | ||||
77 | |||||
78 | // Local function definitions for now | ||||
79 | |||||
80 | // need to copy areas that may overlap | ||||
81 | static | ||||
82 | inline void | ||||
83 | us_arrayCopy(const UChar *src, int32_t srcStart, | ||||
84 | UChar *dst, int32_t dstStart, int32_t count) | ||||
85 | { | ||||
86 | if(count>0) { | ||||
87 | uprv_memmove(dst+dstStart, src+srcStart, (size_t)count*sizeof(*src))do { clang diagnostic push
clang diagnostic ignored "-Waddress" (void)0; (void)0; clang diagnostic pop :: memmove(dst+dstStart , src+srcStart, (size_t)count*sizeof(*src)); } while (false); | ||||
88 | } | ||||
89 | } | ||||
90 | |||||
91 | // u_unescapeAt() callback to get a UChar from a UnicodeString | ||||
92 | U_CDECL_BEGINextern "C" { | ||||
93 | static UChar U_CALLCONV | ||||
94 | UnicodeString_charAt(int32_t offset, void *context) { | ||||
95 | return ((icu::UnicodeString*) context)->charAt(offset); | ||||
96 | } | ||||
97 | U_CDECL_END} | ||||
98 | |||||
99 | U_NAMESPACE_BEGINnamespace icu_71 { | ||||
100 | |||||
101 | /* The Replaceable virtual destructor can't be defined in the header | ||||
102 | due to how AIX works with multiple definitions of virtual functions. | ||||
103 | */ | ||||
104 | Replaceable::~Replaceable() {} | ||||
105 | |||||
106 | UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UnicodeString)UClassID UnicodeString::getStaticClassID() { static char classID = 0; return (UClassID)&classID; } UClassID UnicodeString ::getDynamicClassID() const { return UnicodeString::getStaticClassID (); } | ||||
107 | |||||
108 | UnicodeString U_EXPORT2 | ||||
109 | operator+ (const UnicodeString &s1, const UnicodeString &s2) { | ||||
110 | return | ||||
111 | UnicodeString(s1.length()+s2.length()+1, (UChar32)0, 0). | ||||
112 | append(s1). | ||||
113 | append(s2); | ||||
114 | } | ||||
115 | |||||
116 | //======================================== | ||||
117 | // Reference Counting functions, put at top of file so that optimizing compilers | ||||
118 | // have a chance to automatically inline. | ||||
119 | //======================================== | ||||
120 | |||||
121 | void | ||||
122 | UnicodeString::addRef() { | ||||
123 | umtx_atomic_inc((u_atomic_int32_t *)fUnion.fFields.fArray - 1); | ||||
124 | } | ||||
125 | |||||
126 | int32_t | ||||
127 | UnicodeString::removeRef() { | ||||
128 | return umtx_atomic_dec((u_atomic_int32_t *)fUnion.fFields.fArray - 1); | ||||
129 | } | ||||
130 | |||||
131 | int32_t | ||||
132 | UnicodeString::refCount() const { | ||||
133 | return umtx_loadAcquire(*((u_atomic_int32_t *)fUnion.fFields.fArray - 1)); | ||||
134 | } | ||||
135 | |||||
136 | void | ||||
137 | UnicodeString::releaseArray() { | ||||
138 | if((fUnion.fFields.fLengthAndFlags & kRefCounted) && removeRef() == 0) { | ||||
139 | uprv_freeuprv_free_71((int32_t *)fUnion.fFields.fArray - 1); | ||||
140 | } | ||||
141 | } | ||||
142 | |||||
143 | |||||
144 | |||||
145 | //======================================== | ||||
146 | // Constructors | ||||
147 | //======================================== | ||||
148 | |||||
149 | // The default constructor is inline in unistr.h. | ||||
150 | |||||
151 | UnicodeString::UnicodeString(int32_t capacity, UChar32 c, int32_t count) { | ||||
152 | fUnion.fFields.fLengthAndFlags = 0; | ||||
153 | if(count <= 0 || (uint32_t)c > 0x10ffff) { | ||||
154 | // just allocate and do not do anything else | ||||
155 | allocate(capacity); | ||||
156 | } else if(c <= 0xffff) { | ||||
157 | int32_t length = count; | ||||
158 | if(capacity < length) { | ||||
159 | capacity = length; | ||||
160 | } | ||||
161 | if(allocate(capacity)) { | ||||
162 | UChar *array = getArrayStart(); | ||||
163 | UChar unit = (UChar)c; | ||||
164 | for(int32_t i = 0; i < length; ++i) { | ||||
165 | array[i] = unit; | ||||
166 | } | ||||
167 | setLength(length); | ||||
168 | } | ||||
169 | } else { // supplementary code point, write surrogate pairs | ||||
170 | if(count > (INT32_MAX(2147483647) / 2)) { | ||||
171 | // We would get more than 2G UChars. | ||||
172 | allocate(capacity); | ||||
173 | return; | ||||
174 | } | ||||
175 | int32_t length = count * 2; | ||||
176 | if(capacity < length) { | ||||
177 | capacity = length; | ||||
178 | } | ||||
179 | if(allocate(capacity)) { | ||||
180 | UChar *array = getArrayStart(); | ||||
181 | UChar lead = U16_LEAD(c)(UChar)(((c)>>10)+0xd7c0); | ||||
182 | UChar trail = U16_TRAIL(c)(UChar)(((c)&0x3ff)|0xdc00); | ||||
183 | for(int32_t i = 0; i < length; i += 2) { | ||||
184 | array[i] = lead; | ||||
185 | array[i + 1] = trail; | ||||
186 | } | ||||
187 | setLength(length); | ||||
188 | } | ||||
189 | } | ||||
190 | } | ||||
191 | |||||
192 | UnicodeString::UnicodeString(UChar ch) { | ||||
193 | fUnion.fFields.fLengthAndFlags = kLength1 | kShortString; | ||||
194 | fUnion.fStackFields.fBuffer[0] = ch; | ||||
195 | } | ||||
196 | |||||
197 | UnicodeString::UnicodeString(UChar32 ch) { | ||||
198 | fUnion.fFields.fLengthAndFlags = kShortString; | ||||
199 | int32_t i = 0; | ||||
200 | UBool isError = FALSE0; | ||||
201 | U16_APPEND(fUnion.fStackFields.fBuffer, i, US_STACKBUF_SIZE, ch, isError)do { if((uint32_t)(ch)<=0xffff) { (fUnion.fStackFields.fBuffer )[(i)++]=(uint16_t)(ch); } else if((uint32_t)(ch)<=0x10ffff && (i)+1<(US_STACKBUF_SIZE)) { (fUnion.fStackFields .fBuffer)[(i)++]=(uint16_t)(((ch)>>10)+0xd7c0); (fUnion .fStackFields.fBuffer)[(i)++]=(uint16_t)(((ch)&0x3ff)|0xdc00 ); } else { (isError)=true; } } while (false); | ||||
202 | // We test isError so that the compiler does not complain that we don't. | ||||
203 | // If isError then i==0 which is what we want anyway. | ||||
204 | if(!isError) { | ||||
205 | setShortLength(i); | ||||
206 | } | ||||
207 | } | ||||
208 | |||||
209 | UnicodeString::UnicodeString(const UChar *text) { | ||||
210 | fUnion.fFields.fLengthAndFlags = kShortString; | ||||
211 | doAppend(text, 0, -1); | ||||
212 | } | ||||
213 | |||||
214 | UnicodeString::UnicodeString(const UChar *text, | ||||
215 | int32_t textLength) { | ||||
216 | fUnion.fFields.fLengthAndFlags = kShortString; | ||||
217 | doAppend(text, 0, textLength); | ||||
218 | } | ||||
219 | |||||
220 | UnicodeString::UnicodeString(UBool isTerminated, | ||||
221 | ConstChar16Ptr textPtr, | ||||
222 | int32_t textLength) { | ||||
223 | fUnion.fFields.fLengthAndFlags = kReadonlyAlias; | ||||
224 | const UChar *text = textPtr; | ||||
225 | if(text == NULL__null) { | ||||
226 | // treat as an empty string, do not alias | ||||
227 | setToEmpty(); | ||||
228 | } else if(textLength < -1 || | ||||
229 | (textLength == -1 && !isTerminated) || | ||||
230 | (textLength >= 0 && isTerminated && text[textLength] != 0) | ||||
231 | ) { | ||||
232 | setToBogus(); | ||||
233 | } else { | ||||
234 | if(textLength == -1) { | ||||
235 | // text is terminated, or else it would have failed the above test | ||||
236 | textLength = u_strlenu_strlen_71(text); | ||||
237 | } | ||||
238 | setArray(const_cast<UChar *>(text), textLength, | ||||
239 | isTerminated ? textLength + 1 : textLength); | ||||
240 | } | ||||
241 | } | ||||
242 | |||||
243 | UnicodeString::UnicodeString(UChar *buff, | ||||
244 | int32_t buffLength, | ||||
245 | int32_t buffCapacity) { | ||||
246 | fUnion.fFields.fLengthAndFlags = kWritableAlias; | ||||
247 | if(buff == NULL__null) { | ||||
248 | // treat as an empty string, do not alias | ||||
249 | setToEmpty(); | ||||
250 | } else if(buffLength < -1 || buffCapacity < 0 || buffLength > buffCapacity) { | ||||
251 | setToBogus(); | ||||
252 | } else { | ||||
253 | if(buffLength == -1) { | ||||
254 | // fLength = u_strlen(buff); but do not look beyond buffCapacity | ||||
255 | const UChar *p = buff, *limit = buff + buffCapacity; | ||||
256 | while(p != limit && *p != 0) { | ||||
257 | ++p; | ||||
258 | } | ||||
259 | buffLength = (int32_t)(p - buff); | ||||
260 | } | ||||
261 | setArray(buff, buffLength, buffCapacity); | ||||
262 | } | ||||
263 | } | ||||
264 | |||||
265 | UnicodeString::UnicodeString(const char *src, int32_t length, EInvariant) { | ||||
266 | fUnion.fFields.fLengthAndFlags = kShortString; | ||||
267 | if(src==NULL__null) { | ||||
268 | // treat as an empty string | ||||
269 | } else { | ||||
270 | if(length<0) { | ||||
271 | length=(int32_t)uprv_strlen(src):: strlen(src); | ||||
272 | } | ||||
273 | if(cloneArrayIfNeeded(length, length, FALSE0)) { | ||||
274 | u_charsToUCharsu_charsToUChars_71(src, getArrayStart(), length); | ||||
275 | setLength(length); | ||||
276 | } else { | ||||
277 | setToBogus(); | ||||
278 | } | ||||
279 | } | ||||
280 | } | ||||
281 | |||||
282 | #if U_CHARSET_IS_UTF81 | ||||
283 | |||||
284 | UnicodeString::UnicodeString(const char *codepageData) { | ||||
285 | fUnion.fFields.fLengthAndFlags = kShortString; | ||||
286 | if(codepageData != 0) { | ||||
287 | setToUTF8(codepageData); | ||||
288 | } | ||||
289 | } | ||||
290 | |||||
291 | UnicodeString::UnicodeString(const char *codepageData, int32_t dataLength) { | ||||
292 | fUnion.fFields.fLengthAndFlags = kShortString; | ||||
293 | // if there's nothing to convert, do nothing | ||||
294 | if(codepageData == 0 || dataLength == 0 || dataLength < -1) { | ||||
295 | return; | ||||
296 | } | ||||
297 | if(dataLength == -1) { | ||||
298 | dataLength = (int32_t)uprv_strlen(codepageData):: strlen(codepageData); | ||||
299 | } | ||||
300 | setToUTF8(StringPiece(codepageData, dataLength)); | ||||
301 | } | ||||
302 | |||||
303 | // else see unistr_cnv.cpp | ||||
304 | #endif | ||||
305 | |||||
306 | UnicodeString::UnicodeString(const UnicodeString& that) { | ||||
307 | fUnion.fFields.fLengthAndFlags = kShortString; | ||||
308 | copyFrom(that); | ||||
309 | } | ||||
310 | |||||
311 | UnicodeString::UnicodeString(UnicodeString &&src) U_NOEXCEPTnoexcept { | ||||
312 | copyFieldsFrom(src, TRUE1); | ||||
313 | } | ||||
314 | |||||
315 | UnicodeString::UnicodeString(const UnicodeString& that, | ||||
316 | int32_t srcStart) { | ||||
317 | fUnion.fFields.fLengthAndFlags = kShortString; | ||||
318 | setTo(that, srcStart); | ||||
319 | } | ||||
320 | |||||
321 | UnicodeString::UnicodeString(const UnicodeString& that, | ||||
322 | int32_t srcStart, | ||||
323 | int32_t srcLength) { | ||||
324 | fUnion.fFields.fLengthAndFlags = kShortString; | ||||
325 | setTo(that, srcStart, srcLength); | ||||
326 | } | ||||
327 | |||||
328 | // Replaceable base class clone() default implementation, does not clone | ||||
329 | Replaceable * | ||||
330 | Replaceable::clone() const { | ||||
331 | return NULL__null; | ||||
332 | } | ||||
333 | |||||
334 | // UnicodeString overrides clone() with a real implementation | ||||
335 | UnicodeString * | ||||
336 | UnicodeString::clone() const { | ||||
337 | LocalPointer<UnicodeString> clonedString(new UnicodeString(*this)); | ||||
338 | return clonedString.isValid() && !clonedString->isBogus() ? clonedString.orphan() : nullptr; | ||||
339 | } | ||||
340 | |||||
341 | //======================================== | ||||
342 | // array allocation | ||||
343 | //======================================== | ||||
344 | |||||
345 | namespace { | ||||
346 | |||||
347 | const int32_t kGrowSize = 128; | ||||
348 | |||||
349 | // The number of bytes for one int32_t reference counter and capacity UChars | ||||
350 | // must fit into a 32-bit size_t (at least when on a 32-bit platform). | ||||
351 | // We also add one for the NUL terminator, to avoid reallocation in getTerminatedBuffer(), | ||||
352 | // and round up to a multiple of 16 bytes. | ||||
353 | // This means that capacity must be at most (0xfffffff0 - 4) / 2 - 1 = 0x7ffffff5. | ||||
354 | // (With more complicated checks we could go up to 0x7ffffffd without rounding up, | ||||
355 | // but that does not seem worth it.) | ||||
356 | const int32_t kMaxCapacity = 0x7ffffff5; | ||||
357 | |||||
358 | int32_t getGrowCapacity(int32_t newLength) { | ||||
359 | int32_t growSize = (newLength >> 2) + kGrowSize; | ||||
360 | if(growSize <= (kMaxCapacity - newLength)) { | ||||
361 | return newLength + growSize; | ||||
362 | } else { | ||||
363 | return kMaxCapacity; | ||||
364 | } | ||||
365 | } | ||||
366 | |||||
367 | } // namespace | ||||
368 | |||||
369 | UBool | ||||
370 | UnicodeString::allocate(int32_t capacity) { | ||||
371 | if(capacity
| ||||
372 | fUnion.fFields.fLengthAndFlags = kShortString; | ||||
373 | return TRUE1; | ||||
374 | } | ||||
375 | if(capacity <= kMaxCapacity) { | ||||
376 | ++capacity; // for the NUL | ||||
377 | // Switch to size_t which is unsigned so that we can allocate up to 4GB. | ||||
378 | // Reference counter + UChars. | ||||
379 | size_t numBytes = sizeof(int32_t) + (size_t)capacity * U_SIZEOF_UCHAR2; | ||||
380 | // Round up to a multiple of 16. | ||||
381 | numBytes = (numBytes + 15) & ~15; | ||||
382 | int32_t *array = (int32_t *) uprv_mallocuprv_malloc_71(numBytes); | ||||
383 | if(array != NULL__null) { | ||||
384 | // set initial refCount and point behind the refCount | ||||
385 | *array++ = 1; | ||||
386 | numBytes -= sizeof(int32_t); | ||||
387 | |||||
388 | // have fArray point to the first UChar | ||||
389 | fUnion.fFields.fArray = (UChar *)array; | ||||
390 | fUnion.fFields.fCapacity = (int32_t)(numBytes / U_SIZEOF_UCHAR2); | ||||
391 | fUnion.fFields.fLengthAndFlags = kLongString; | ||||
392 | return TRUE1; | ||||
393 | } | ||||
394 | } | ||||
395 | fUnion.fFields.fLengthAndFlags = kIsBogus; | ||||
396 | fUnion.fFields.fArray = 0; | ||||
397 | fUnion.fFields.fCapacity = 0; | ||||
398 | return FALSE0; | ||||
399 | } | ||||
400 | |||||
401 | //======================================== | ||||
402 | // Destructor | ||||
403 | //======================================== | ||||
404 | |||||
405 | #ifdef UNISTR_COUNT_FINAL_STRING_LENGTHS | ||||
406 | static u_atomic_int32_t finalLengthCounts[0x400]; // UnicodeString::kMaxShortLength+1 | ||||
407 | static u_atomic_int32_t beyondCount(0); | ||||
408 | |||||
409 | U_CAPIextern "C" void unistr_printLengths() { | ||||
410 | int32_t i; | ||||
411 | for(i = 0; i <= 59; ++i) { | ||||
412 | printf("%2d, %9d\n", i, (int32_t)finalLengthCounts[i]); | ||||
413 | } | ||||
414 | int32_t beyond = beyondCount; | ||||
415 | for(; i < UPRV_LENGTHOF(finalLengthCounts)(int32_t)(sizeof(finalLengthCounts)/sizeof((finalLengthCounts )[0])); ++i) { | ||||
416 | beyond += finalLengthCounts[i]; | ||||
417 | } | ||||
418 | printf(">59, %9d\n", beyond); | ||||
419 | } | ||||
420 | #endif | ||||
421 | |||||
422 | UnicodeString::~UnicodeString() | ||||
423 | { | ||||
424 | #ifdef UNISTR_COUNT_FINAL_STRING_LENGTHS | ||||
425 | // Count lengths of strings at the end of their lifetime. | ||||
426 | // Useful for discussion of a desirable stack buffer size. | ||||
427 | // Count the contents length, not the optional NUL terminator nor further capacity. | ||||
428 | // Ignore open-buffer strings and strings which alias external storage. | ||||
429 | if((fUnion.fFields.fLengthAndFlags&(kOpenGetBuffer|kReadonlyAlias|kWritableAlias)) == 0) { | ||||
430 | if(hasShortLength()) { | ||||
431 | umtx_atomic_inc(finalLengthCounts + getShortLength()); | ||||
432 | } else { | ||||
433 | umtx_atomic_inc(&beyondCount); | ||||
434 | } | ||||
435 | } | ||||
436 | #endif | ||||
437 | |||||
438 | releaseArray(); | ||||
439 | } | ||||
440 | |||||
441 | //======================================== | ||||
442 | // Factory methods | ||||
443 | //======================================== | ||||
444 | |||||
445 | UnicodeString UnicodeString::fromUTF8(StringPiece utf8) { | ||||
446 | UnicodeString result; | ||||
447 | result.setToUTF8(utf8); | ||||
448 | return result; | ||||
449 | } | ||||
450 | |||||
451 | UnicodeString UnicodeString::fromUTF32(const UChar32 *utf32, int32_t length) { | ||||
452 | UnicodeString result; | ||||
453 | int32_t capacity; | ||||
454 | // Most UTF-32 strings will be BMP-only and result in a same-length | ||||
455 | // UTF-16 string. We overestimate the capacity just slightly, | ||||
456 | // just in case there are a few supplementary characters. | ||||
457 | if(length <= US_STACKBUF_SIZE) { | ||||
458 | capacity = US_STACKBUF_SIZE; | ||||
459 | } else { | ||||
460 | capacity = length + (length >> 4) + 4; | ||||
461 | } | ||||
462 | do { | ||||
463 | UChar *utf16 = result.getBuffer(capacity); | ||||
464 | int32_t length16; | ||||
465 | UErrorCode errorCode = U_ZERO_ERROR; | ||||
466 | u_strFromUTF32WithSubu_strFromUTF32WithSub_71(utf16, result.getCapacity(), &length16, | ||||
467 | utf32, length, | ||||
468 | 0xfffd, // Substitution character. | ||||
469 | NULL__null, // Don't care about number of substitutions. | ||||
470 | &errorCode); | ||||
471 | result.releaseBuffer(length16); | ||||
472 | if(errorCode == U_BUFFER_OVERFLOW_ERROR) { | ||||
473 | capacity = length16 + 1; // +1 for the terminating NUL. | ||||
474 | continue; | ||||
475 | } else if(U_FAILURE(errorCode)) { | ||||
476 | result.setToBogus(); | ||||
477 | } | ||||
478 | break; | ||||
479 | } while(TRUE1); | ||||
480 | return result; | ||||
481 | } | ||||
482 | |||||
483 | //======================================== | ||||
484 | // Assignment | ||||
485 | //======================================== | ||||
486 | |||||
487 | UnicodeString & | ||||
488 | UnicodeString::operator=(const UnicodeString &src) { | ||||
489 | return copyFrom(src); | ||||
490 | } | ||||
491 | |||||
492 | UnicodeString & | ||||
493 | UnicodeString::fastCopyFrom(const UnicodeString &src) { | ||||
494 | return copyFrom(src, TRUE1); | ||||
495 | } | ||||
496 | |||||
497 | UnicodeString & | ||||
498 | UnicodeString::copyFrom(const UnicodeString &src, UBool fastCopy) { | ||||
499 | // if assigning to ourselves, do nothing | ||||
500 | if(this == &src) { | ||||
501 | return *this; | ||||
502 | } | ||||
503 | |||||
504 | // is the right side bogus? | ||||
505 | if(src.isBogus()) { | ||||
506 | setToBogus(); | ||||
507 | return *this; | ||||
508 | } | ||||
509 | |||||
510 | // delete the current contents | ||||
511 | releaseArray(); | ||||
512 | |||||
513 | if(src.isEmpty()) { | ||||
514 | // empty string - use the stack buffer | ||||
515 | setToEmpty(); | ||||
516 | return *this; | ||||
517 | } | ||||
518 | |||||
519 | // fLength>0 and not an "open" src.getBuffer(minCapacity) | ||||
520 | fUnion.fFields.fLengthAndFlags = src.fUnion.fFields.fLengthAndFlags; | ||||
521 | switch(src.fUnion.fFields.fLengthAndFlags & kAllStorageFlags) { | ||||
522 | case kShortString: | ||||
523 | // short string using the stack buffer, do the same | ||||
524 | uprv_memcpy(fUnion.fStackFields.fBuffer, src.fUnion.fStackFields.fBuffer,do { clang diagnostic push
clang diagnostic ignored "-Waddress" (void)0; (void)0; clang diagnostic pop :: memcpy(fUnion.fStackFields .fBuffer, src.fUnion.fStackFields.fBuffer, getShortLength() * 2); } while (false) | ||||
525 | getShortLength() * U_SIZEOF_UCHAR)do { clang diagnostic push
clang diagnostic ignored "-Waddress" (void)0; (void)0; clang diagnostic pop :: memcpy(fUnion.fStackFields .fBuffer, src.fUnion.fStackFields.fBuffer, getShortLength() * 2); } while (false); | ||||
526 | break; | ||||
527 | case kLongString: | ||||
528 | // src uses a refCounted string buffer, use that buffer with refCount | ||||
529 | // src is const, use a cast - we don't actually change it | ||||
530 | ((UnicodeString &)src).addRef(); | ||||
531 | // copy all fields, share the reference-counted buffer | ||||
532 | fUnion.fFields.fArray = src.fUnion.fFields.fArray; | ||||
533 | fUnion.fFields.fCapacity = src.fUnion.fFields.fCapacity; | ||||
534 | if(!hasShortLength()) { | ||||
535 | fUnion.fFields.fLength = src.fUnion.fFields.fLength; | ||||
536 | } | ||||
537 | break; | ||||
538 | case kReadonlyAlias: | ||||
539 | if(fastCopy) { | ||||
540 | // src is a readonly alias, do the same | ||||
541 | // -> maintain the readonly alias as such | ||||
542 | fUnion.fFields.fArray = src.fUnion.fFields.fArray; | ||||
543 | fUnion.fFields.fCapacity = src.fUnion.fFields.fCapacity; | ||||
544 | if(!hasShortLength()) { | ||||
545 | fUnion.fFields.fLength = src.fUnion.fFields.fLength; | ||||
546 | } | ||||
547 | break; | ||||
548 | } | ||||
549 | // else if(!fastCopy) fall through to case kWritableAlias | ||||
550 | // -> allocate a new buffer and copy the contents | ||||
551 | U_FALLTHROUGH[[clang::fallthrough]]; | ||||
552 | case kWritableAlias: { | ||||
553 | // src is a writable alias; we make a copy of that instead | ||||
554 | int32_t srcLength = src.length(); | ||||
555 | if(allocate(srcLength)) { | ||||
556 | u_memcpyu_memcpy_71(getArrayStart(), src.getArrayStart(), srcLength); | ||||
557 | setLength(srcLength); | ||||
558 | break; | ||||
559 | } | ||||
560 | // if there is not enough memory, then fall through to setting to bogus | ||||
561 | U_FALLTHROUGH[[clang::fallthrough]]; | ||||
562 | } | ||||
563 | default: | ||||
564 | // if src is bogus, set ourselves to bogus | ||||
565 | // do not call setToBogus() here because fArray and flags are not consistent here | ||||
566 | fUnion.fFields.fLengthAndFlags = kIsBogus; | ||||
567 | fUnion.fFields.fArray = 0; | ||||
568 | fUnion.fFields.fCapacity = 0; | ||||
569 | break; | ||||
570 | } | ||||
571 | |||||
572 | return *this; | ||||
573 | } | ||||
574 | |||||
575 | UnicodeString &UnicodeString::operator=(UnicodeString &&src) U_NOEXCEPTnoexcept { | ||||
576 | // No explicit check for self move assignment, consistent with standard library. | ||||
577 | // Self move assignment causes no crash nor leak but might make the object bogus. | ||||
578 | releaseArray(); | ||||
579 | copyFieldsFrom(src, TRUE1); | ||||
580 | return *this; | ||||
581 | } | ||||
582 | |||||
583 | // Same as move assignment except without memory management. | ||||
584 | void UnicodeString::copyFieldsFrom(UnicodeString &src, UBool setSrcToBogus) U_NOEXCEPTnoexcept { | ||||
585 | int16_t lengthAndFlags = fUnion.fFields.fLengthAndFlags = src.fUnion.fFields.fLengthAndFlags; | ||||
586 | if(lengthAndFlags & kUsingStackBuffer) { | ||||
587 | // Short string using the stack buffer, copy the contents. | ||||
588 | // Check for self assignment to prevent "overlap in memcpy" warnings, | ||||
589 | // although it should be harmless to copy a buffer to itself exactly. | ||||
590 | if(this != &src) { | ||||
591 | uprv_memcpy(fUnion.fStackFields.fBuffer, src.fUnion.fStackFields.fBuffer,do { clang diagnostic push
clang diagnostic ignored "-Waddress" (void)0; (void)0; clang diagnostic pop :: memcpy(fUnion.fStackFields .fBuffer, src.fUnion.fStackFields.fBuffer, getShortLength() * 2); } while (false) | ||||
592 | getShortLength() * U_SIZEOF_UCHAR)do { clang diagnostic push
clang diagnostic ignored "-Waddress" (void)0; (void)0; clang diagnostic pop :: memcpy(fUnion.fStackFields .fBuffer, src.fUnion.fStackFields.fBuffer, getShortLength() * 2); } while (false); | ||||
593 | } | ||||
594 | } else { | ||||
595 | // In all other cases, copy all fields. | ||||
596 | fUnion.fFields.fArray = src.fUnion.fFields.fArray; | ||||
597 | fUnion.fFields.fCapacity = src.fUnion.fFields.fCapacity; | ||||
598 | if(!hasShortLength()) { | ||||
599 | fUnion.fFields.fLength = src.fUnion.fFields.fLength; | ||||
600 | } | ||||
601 | if(setSrcToBogus) { | ||||
602 | // Set src to bogus without releasing any memory. | ||||
603 | src.fUnion.fFields.fLengthAndFlags = kIsBogus; | ||||
604 | src.fUnion.fFields.fArray = NULL__null; | ||||
605 | src.fUnion.fFields.fCapacity = 0; | ||||
606 | } | ||||
607 | } | ||||
608 | } | ||||
609 | |||||
610 | void UnicodeString::swap(UnicodeString &other) U_NOEXCEPTnoexcept { | ||||
611 | UnicodeString temp; // Empty short string: Known not to need releaseArray(). | ||||
612 | // Copy fields without resetting source values in between. | ||||
613 | temp.copyFieldsFrom(*this, FALSE0); | ||||
614 | this->copyFieldsFrom(other, FALSE0); | ||||
615 | other.copyFieldsFrom(temp, FALSE0); | ||||
616 | // Set temp to an empty string so that other's memory is not released twice. | ||||
617 | temp.fUnion.fFields.fLengthAndFlags = kShortString; | ||||
618 | } | ||||
619 | |||||
620 | //======================================== | ||||
621 | // Miscellaneous operations | ||||
622 | //======================================== | ||||
623 | |||||
624 | UnicodeString UnicodeString::unescape() const { | ||||
625 | UnicodeString result(length(), (UChar32)0, (int32_t)0); // construct with capacity | ||||
626 | if (result.isBogus()) { | ||||
627 | return result; | ||||
628 | } | ||||
629 | const UChar *array = getBuffer(); | ||||
630 | int32_t len = length(); | ||||
631 | int32_t prev = 0; | ||||
632 | for (int32_t i=0;;) { | ||||
633 | if (i == len) { | ||||
634 | result.append(array, prev, len - prev); | ||||
635 | break; | ||||
636 | } | ||||
637 | if (array[i++] == 0x5C /*'\\'*/) { | ||||
638 | result.append(array, prev, (i - 1) - prev); | ||||
639 | UChar32 c = unescapeAt(i); // advances i | ||||
640 | if (c < 0) { | ||||
641 | result.remove(); // return empty string | ||||
642 | break; // invalid escape sequence | ||||
643 | } | ||||
644 | result.append(c); | ||||
645 | prev = i; | ||||
646 | } | ||||
647 | } | ||||
648 | return result; | ||||
649 | } | ||||
650 | |||||
651 | UChar32 UnicodeString::unescapeAt(int32_t &offset) const { | ||||
652 | return u_unescapeAtu_unescapeAt_71(UnicodeString_charAt, &offset, length(), (void*)this); | ||||
653 | } | ||||
654 | |||||
655 | //======================================== | ||||
656 | // Read-only implementation | ||||
657 | //======================================== | ||||
658 | UBool | ||||
659 | UnicodeString::doEquals(const UnicodeString &text, int32_t len) const { | ||||
660 | // Requires: this & text not bogus and have same lengths. | ||||
661 | // Byte-wise comparison works for equality regardless of endianness. | ||||
662 | return uprv_memcmp(getArrayStart(), text.getArrayStart(), len * U_SIZEOF_UCHAR):: memcmp(getArrayStart(), text.getArrayStart(),len * 2) == 0; | ||||
663 | } | ||||
664 | |||||
665 | int8_t | ||||
666 | UnicodeString::doCompare( int32_t start, | ||||
667 | int32_t length, | ||||
668 | const UChar *srcChars, | ||||
669 | int32_t srcStart, | ||||
670 | int32_t srcLength) const | ||||
671 | { | ||||
672 | // compare illegal string values | ||||
673 | if(isBogus()) { | ||||
674 | return -1; | ||||
675 | } | ||||
676 | |||||
677 | // pin indices to legal values | ||||
678 | pinIndices(start, length); | ||||
679 | |||||
680 | if(srcChars == NULL__null) { | ||||
681 | // treat const UChar *srcChars==NULL as an empty string | ||||
682 | return length == 0 ? 0 : 1; | ||||
683 | } | ||||
684 | |||||
685 | // get the correct pointer | ||||
686 | const UChar *chars = getArrayStart(); | ||||
687 | |||||
688 | chars += start; | ||||
689 | srcChars += srcStart; | ||||
690 | |||||
691 | int32_t minLength; | ||||
692 | int8_t lengthResult; | ||||
693 | |||||
694 | // get the srcLength if necessary | ||||
695 | if(srcLength < 0) { | ||||
696 | srcLength = u_strlenu_strlen_71(srcChars + srcStart); | ||||
697 | } | ||||
698 | |||||
699 | // are we comparing different lengths? | ||||
700 | if(length != srcLength) { | ||||
701 | if(length < srcLength) { | ||||
702 | minLength = length; | ||||
703 | lengthResult = -1; | ||||
704 | } else { | ||||
705 | minLength = srcLength; | ||||
706 | lengthResult = 1; | ||||
707 | } | ||||
708 | } else { | ||||
709 | minLength = length; | ||||
710 | lengthResult = 0; | ||||
711 | } | ||||
712 | |||||
713 | /* | ||||
714 | * note that uprv_memcmp() returns an int but we return an int8_t; | ||||
715 | * we need to take care not to truncate the result - | ||||
716 | * one way to do this is to right-shift the value to | ||||
717 | * move the sign bit into the lower 8 bits and making sure that this | ||||
718 | * does not become 0 itself | ||||
719 | */ | ||||
720 | |||||
721 | if(minLength > 0 && chars != srcChars) { | ||||
722 | int32_t result; | ||||
723 | |||||
724 | # if U_IS_BIG_ENDIAN(1234 == 4321) | ||||
725 | // big-endian: byte comparison works | ||||
726 | result = uprv_memcmp(chars, srcChars, minLength * sizeof(UChar)):: memcmp(chars, srcChars,minLength * sizeof(UChar)); | ||||
727 | if(result != 0) { | ||||
728 | return (int8_t)(result >> 15 | 1); | ||||
729 | } | ||||
730 | # else | ||||
731 | // little-endian: compare UChar units | ||||
732 | do { | ||||
733 | result = ((int32_t)*(chars++) - (int32_t)*(srcChars++)); | ||||
734 | if(result != 0) { | ||||
735 | return (int8_t)(result >> 15 | 1); | ||||
736 | } | ||||
737 | } while(--minLength > 0); | ||||
738 | # endif | ||||
739 | } | ||||
740 | return lengthResult; | ||||
741 | } | ||||
742 | |||||
743 | /* String compare in code point order - doCompare() compares in code unit order. */ | ||||
744 | int8_t | ||||
745 | UnicodeString::doCompareCodePointOrder(int32_t start, | ||||
746 | int32_t length, | ||||
747 | const UChar *srcChars, | ||||
748 | int32_t srcStart, | ||||
749 | int32_t srcLength) const | ||||
750 | { | ||||
751 | // compare illegal string values | ||||
752 | // treat const UChar *srcChars==NULL as an empty string | ||||
753 | if(isBogus()) { | ||||
754 | return -1; | ||||
755 | } | ||||
756 | |||||
757 | // pin indices to legal values | ||||
758 | pinIndices(start, length); | ||||
759 | |||||
760 | if(srcChars == NULL__null) { | ||||
761 | srcStart = srcLength = 0; | ||||
762 | } | ||||
763 | |||||
764 | int32_t diff = uprv_strCompareuprv_strCompare_71(getArrayStart() + start, length, (srcChars!=NULL__null)?(srcChars + srcStart):NULL__null, srcLength, FALSE0, TRUE1); | ||||
765 | /* translate the 32-bit result into an 8-bit one */ | ||||
766 | if(diff!=0) { | ||||
767 | return (int8_t)(diff >> 15 | 1); | ||||
768 | } else { | ||||
769 | return 0; | ||||
770 | } | ||||
771 | } | ||||
772 | |||||
773 | int32_t | ||||
774 | UnicodeString::getLength() const { | ||||
775 | return length(); | ||||
776 | } | ||||
777 | |||||
778 | UChar | ||||
779 | UnicodeString::getCharAt(int32_t offset) const { | ||||
780 | return charAt(offset); | ||||
781 | } | ||||
782 | |||||
783 | UChar32 | ||||
784 | UnicodeString::getChar32At(int32_t offset) const { | ||||
785 | return char32At(offset); | ||||
786 | } | ||||
787 | |||||
788 | UChar32 | ||||
789 | UnicodeString::char32At(int32_t offset) const | ||||
790 | { | ||||
791 | int32_t len = length(); | ||||
792 | if((uint32_t)offset < (uint32_t)len) { | ||||
793 | const UChar *array = getArrayStart(); | ||||
794 | UChar32 c; | ||||
795 | U16_GET(array, 0, offset, len, c)do { (c)=(array)[offset]; if((((c)&0xfffff800)==0xd800)) { uint16_t __c2; if((((c)&0x400)==0)) { if((offset)+1!=(len ) && (((__c2=(array)[(offset)+1])&0xfffffc00)==0xdc00 )) { (c)=(((UChar32)((c))<<10UL)+(UChar32)(__c2)-((0xd800 <<10UL)+0xdc00-0x10000)); } } else { if((offset)>(0) && (((__c2=(array)[(offset)-1])&0xfffffc00)==0xd800 )) { (c)=(((UChar32)(__c2)<<10UL)+(UChar32)((c))-((0xd800 <<10UL)+0xdc00-0x10000)); } } } } while (false); | ||||
796 | return c; | ||||
797 | } else { | ||||
798 | return kInvalidUChar; | ||||
799 | } | ||||
800 | } | ||||
801 | |||||
802 | int32_t | ||||
803 | UnicodeString::getChar32Start(int32_t offset) const { | ||||
804 | if((uint32_t)offset < (uint32_t)length()) { | ||||
805 | const UChar *array = getArrayStart(); | ||||
806 | U16_SET_CP_START(array, 0, offset)do { if(((((array)[offset])&0xfffffc00)==0xdc00) && (offset)>(0) && ((((array)[(offset)-1])&0xfffffc00 )==0xd800)) { --(offset); } } while (false); | ||||
807 | return offset; | ||||
808 | } else { | ||||
809 | return 0; | ||||
810 | } | ||||
811 | } | ||||
812 | |||||
813 | int32_t | ||||
814 | UnicodeString::getChar32Limit(int32_t offset) const { | ||||
815 | int32_t len = length(); | ||||
816 | if((uint32_t)offset < (uint32_t)len) { | ||||
817 | const UChar *array = getArrayStart(); | ||||
818 | U16_SET_CP_LIMIT(array, 0, offset, len)do { if((0)<(offset) && ((offset)<(len) || (len )<0) && ((((array)[(offset)-1])&0xfffffc00)==0xd800 ) && ((((array)[offset])&0xfffffc00)==0xdc00)) { ++ (offset); } } while (false); | ||||
819 | return offset; | ||||
820 | } else { | ||||
821 | return len; | ||||
822 | } | ||||
823 | } | ||||
824 | |||||
825 | int32_t | ||||
826 | UnicodeString::countChar32(int32_t start, int32_t length) const { | ||||
827 | pinIndices(start, length); | ||||
828 | // if(isBogus()) then fArray==0 and start==0 - u_countChar32() checks for NULL | ||||
829 | return u_countChar32u_countChar32_71(getArrayStart()+start, length); | ||||
830 | } | ||||
831 | |||||
832 | UBool | ||||
833 | UnicodeString::hasMoreChar32Than(int32_t start, int32_t length, int32_t number) const { | ||||
834 | pinIndices(start, length); | ||||
835 | // if(isBogus()) then fArray==0 and start==0 - u_strHasMoreChar32Than() checks for NULL | ||||
836 | return u_strHasMoreChar32Thanu_strHasMoreChar32Than_71(getArrayStart()+start, length, number); | ||||
837 | } | ||||
838 | |||||
839 | int32_t | ||||
840 | UnicodeString::moveIndex32(int32_t index, int32_t delta) const { | ||||
841 | // pin index | ||||
842 | int32_t len = length(); | ||||
843 | if(index<0) { | ||||
844 | index=0; | ||||
845 | } else if(index>len) { | ||||
846 | index=len; | ||||
847 | } | ||||
848 | |||||
849 | const UChar *array = getArrayStart(); | ||||
850 | if(delta>0) { | ||||
851 | U16_FWD_N(array, index, len, delta)do { int32_t __N=(delta); while(__N>0 && ((index)< (len) || ((len)<0 && (array)[index]!=0))) { do { if (((((array)[(index)++])&0xfffffc00)==0xd800) && ( index)!=(len) && ((((array)[index])&0xfffffc00)== 0xdc00)) { ++(index); } } while (false); --__N; } } while (false ); | ||||
852 | } else { | ||||
853 | U16_BACK_N(array, 0, index, -delta)do { int32_t __N=(-delta); while(__N>0 && (index)> (0)) { do { if(((((array)[--(index)])&0xfffffc00)==0xdc00 ) && (index)>(0) && ((((array)[(index)-1]) &0xfffffc00)==0xd800)) { --(index); } } while (false); -- __N; } } while (false); | ||||
854 | } | ||||
855 | |||||
856 | return index; | ||||
857 | } | ||||
858 | |||||
859 | void | ||||
860 | UnicodeString::doExtract(int32_t start, | ||||
861 | int32_t length, | ||||
862 | UChar *dst, | ||||
863 | int32_t dstStart) const | ||||
864 | { | ||||
865 | // pin indices to legal values | ||||
866 | pinIndices(start, length); | ||||
867 | |||||
868 | // do not copy anything if we alias dst itself | ||||
869 | const UChar *array = getArrayStart(); | ||||
870 | if(array + start != dst + dstStart) { | ||||
871 | us_arrayCopy(array, start, dst, dstStart, length); | ||||
872 | } | ||||
873 | } | ||||
874 | |||||
875 | int32_t | ||||
876 | UnicodeString::extract(Char16Ptr dest, int32_t destCapacity, | ||||
877 | UErrorCode &errorCode) const { | ||||
878 | int32_t len = length(); | ||||
879 | if(U_SUCCESS(errorCode)) { | ||||
880 | if(isBogus() || destCapacity<0 || (destCapacity>0 && dest==0)) { | ||||
881 | errorCode=U_ILLEGAL_ARGUMENT_ERROR; | ||||
882 | } else { | ||||
883 | const UChar *array = getArrayStart(); | ||||
884 | if(len>0 && len<=destCapacity && array!=dest) { | ||||
885 | u_memcpyu_memcpy_71(dest, array, len); | ||||
886 | } | ||||
887 | return u_terminateUCharsu_terminateUChars_71(dest, destCapacity, len, &errorCode); | ||||
888 | } | ||||
889 | } | ||||
890 | |||||
891 | return len; | ||||
892 | } | ||||
893 | |||||
894 | int32_t | ||||
895 | UnicodeString::extract(int32_t start, | ||||
896 | int32_t length, | ||||
897 | char *target, | ||||
898 | int32_t targetCapacity, | ||||
899 | enum EInvariant) const | ||||
900 | { | ||||
901 | // if the arguments are illegal, then do nothing | ||||
902 | if(targetCapacity < 0 || (targetCapacity > 0 && target == NULL__null)) { | ||||
903 | return 0; | ||||
904 | } | ||||
905 | |||||
906 | // pin the indices to legal values | ||||
907 | pinIndices(start, length); | ||||
908 | |||||
909 | if(length <= targetCapacity) { | ||||
910 | u_UCharsToCharsu_UCharsToChars_71(getArrayStart() + start, target, length); | ||||
911 | } | ||||
912 | UErrorCode status = U_ZERO_ERROR; | ||||
913 | return u_terminateCharsu_terminateChars_71(target, targetCapacity, length, &status); | ||||
914 | } | ||||
915 | |||||
916 | UnicodeString | ||||
917 | UnicodeString::tempSubString(int32_t start, int32_t len) const { | ||||
918 | pinIndices(start, len); | ||||
919 | const UChar *array = getBuffer(); // not getArrayStart() to check kIsBogus & kOpenGetBuffer | ||||
920 | if(array==NULL__null) { | ||||
921 | array=fUnion.fStackFields.fBuffer; // anything not NULL because that would make an empty string | ||||
922 | len=-2; // bogus result string | ||||
923 | } | ||||
924 | return UnicodeString(FALSE0, array + start, len); | ||||
925 | } | ||||
926 | |||||
927 | int32_t | ||||
928 | UnicodeString::toUTF8(int32_t start, int32_t len, | ||||
929 | char *target, int32_t capacity) const { | ||||
930 | pinIndices(start, len); | ||||
931 | int32_t length8; | ||||
932 | UErrorCode errorCode = U_ZERO_ERROR; | ||||
933 | u_strToUTF8WithSubu_strToUTF8WithSub_71(target, capacity, &length8, | ||||
934 | getBuffer() + start, len, | ||||
935 | 0xFFFD, // Standard substitution character. | ||||
936 | NULL__null, // Don't care about number of substitutions. | ||||
937 | &errorCode); | ||||
938 | return length8; | ||||
939 | } | ||||
940 | |||||
941 | #if U_CHARSET_IS_UTF81 | ||||
942 | |||||
943 | int32_t | ||||
944 | UnicodeString::extract(int32_t start, int32_t len, | ||||
945 | char *target, uint32_t dstSize) const { | ||||
946 | // if the arguments are illegal, then do nothing | ||||
947 | if(/*dstSize < 0 || */(dstSize > 0 && target == 0)) { | ||||
948 | return 0; | ||||
949 | } | ||||
950 | return toUTF8(start, len, target, dstSize <= 0x7fffffff ? (int32_t)dstSize : 0x7fffffff); | ||||
951 | } | ||||
952 | |||||
953 | // else see unistr_cnv.cpp | ||||
954 | #endif | ||||
955 | |||||
956 | void | ||||
957 | UnicodeString::extractBetween(int32_t start, | ||||
958 | int32_t limit, | ||||
959 | UnicodeString& target) const { | ||||
960 | pinIndex(start); | ||||
961 | pinIndex(limit); | ||||
962 | doExtract(start, limit - start, target); | ||||
963 | } | ||||
964 | |||||
965 | // When converting from UTF-16 to UTF-8, the result will have at most 3 times | ||||
966 | // as many bytes as the source has UChars. | ||||
967 | // The "worst cases" are writing systems like Indic, Thai and CJK with | ||||
968 | // 3:1 bytes:UChars. | ||||
969 | void | ||||
970 | UnicodeString::toUTF8(ByteSink &sink) const { | ||||
971 | int32_t length16 = length(); | ||||
972 | if(length16 != 0) { | ||||
973 | char stackBuffer[1024]; | ||||
974 | int32_t capacity = (int32_t)sizeof(stackBuffer); | ||||
975 | UBool utf8IsOwned = FALSE0; | ||||
976 | char *utf8 = sink.GetAppendBuffer(length16 < capacity ? length16 : capacity, | ||||
977 | 3*length16, | ||||
978 | stackBuffer, capacity, | ||||
979 | &capacity); | ||||
980 | int32_t length8 = 0; | ||||
981 | UErrorCode errorCode = U_ZERO_ERROR; | ||||
982 | u_strToUTF8WithSubu_strToUTF8WithSub_71(utf8, capacity, &length8, | ||||
983 | getBuffer(), length16, | ||||
984 | 0xFFFD, // Standard substitution character. | ||||
985 | NULL__null, // Don't care about number of substitutions. | ||||
986 | &errorCode); | ||||
987 | if(errorCode == U_BUFFER_OVERFLOW_ERROR) { | ||||
988 | utf8 = (char *)uprv_mallocuprv_malloc_71(length8); | ||||
989 | if(utf8 != NULL__null) { | ||||
990 | utf8IsOwned = TRUE1; | ||||
991 | errorCode = U_ZERO_ERROR; | ||||
992 | u_strToUTF8WithSubu_strToUTF8WithSub_71(utf8, length8, &length8, | ||||
993 | getBuffer(), length16, | ||||
994 | 0xFFFD, // Standard substitution character. | ||||
995 | NULL__null, // Don't care about number of substitutions. | ||||
996 | &errorCode); | ||||
997 | } else { | ||||
998 | errorCode = U_MEMORY_ALLOCATION_ERROR; | ||||
999 | } | ||||
1000 | } | ||||
1001 | if(U_SUCCESS(errorCode)) { | ||||
1002 | sink.Append(utf8, length8); | ||||
1003 | sink.Flush(); | ||||
1004 | } | ||||
1005 | if(utf8IsOwned) { | ||||
1006 | uprv_freeuprv_free_71(utf8); | ||||
1007 | } | ||||
1008 | } | ||||
1009 | } | ||||
1010 | |||||
1011 | int32_t | ||||
1012 | UnicodeString::toUTF32(UChar32 *utf32, int32_t capacity, UErrorCode &errorCode) const { | ||||
1013 | int32_t length32=0; | ||||
1014 | if(U_SUCCESS(errorCode)) { | ||||
1015 | // getBuffer() and u_strToUTF32WithSub() check for illegal arguments. | ||||
1016 | u_strToUTF32WithSubu_strToUTF32WithSub_71(utf32, capacity, &length32, | ||||
1017 | getBuffer(), length(), | ||||
1018 | 0xfffd, // Substitution character. | ||||
1019 | NULL__null, // Don't care about number of substitutions. | ||||
1020 | &errorCode); | ||||
1021 | } | ||||
1022 | return length32; | ||||
1023 | } | ||||
1024 | |||||
1025 | int32_t | ||||
1026 | UnicodeString::indexOf(const UChar *srcChars, | ||||
1027 | int32_t srcStart, | ||||
1028 | int32_t srcLength, | ||||
1029 | int32_t start, | ||||
1030 | int32_t length) const | ||||
1031 | { | ||||
1032 | if(isBogus() || srcChars == 0 || srcStart < 0 || srcLength == 0) { | ||||
1033 | return -1; | ||||
1034 | } | ||||
1035 | |||||
1036 | // UnicodeString does not find empty substrings | ||||
1037 | if(srcLength < 0 && srcChars[srcStart] == 0) { | ||||
1038 | return -1; | ||||
1039 | } | ||||
1040 | |||||
1041 | // get the indices within bounds | ||||
1042 | pinIndices(start, length); | ||||
1043 | |||||
1044 | // find the first occurrence of the substring | ||||
1045 | const UChar *array = getArrayStart(); | ||||
1046 | const UChar *match = u_strFindFirstu_strFindFirst_71(array + start, length, srcChars + srcStart, srcLength); | ||||
1047 | if(match == NULL__null) { | ||||
1048 | return -1; | ||||
1049 | } else { | ||||
1050 | return (int32_t)(match - array); | ||||
1051 | } | ||||
1052 | } | ||||
1053 | |||||
1054 | int32_t | ||||
1055 | UnicodeString::doIndexOf(UChar c, | ||||
1056 | int32_t start, | ||||
1057 | int32_t length) const | ||||
1058 | { | ||||
1059 | // pin indices | ||||
1060 | pinIndices(start, length); | ||||
1061 | |||||
1062 | // find the first occurrence of c | ||||
1063 | const UChar *array = getArrayStart(); | ||||
1064 | const UChar *match = u_memchru_memchr_71(array + start, c, length); | ||||
1065 | if(match == NULL__null) { | ||||
1066 | return -1; | ||||
1067 | } else { | ||||
1068 | return (int32_t)(match - array); | ||||
1069 | } | ||||
1070 | } | ||||
1071 | |||||
1072 | int32_t | ||||
1073 | UnicodeString::doIndexOf(UChar32 c, | ||||
1074 | int32_t start, | ||||
1075 | int32_t length) const { | ||||
1076 | // pin indices | ||||
1077 | pinIndices(start, length); | ||||
1078 | |||||
1079 | // find the first occurrence of c | ||||
1080 | const UChar *array = getArrayStart(); | ||||
1081 | const UChar *match = u_memchr32u_memchr32_71(array + start, c, length); | ||||
1082 | if(match == NULL__null) { | ||||
1083 | return -1; | ||||
1084 | } else { | ||||
1085 | return (int32_t)(match - array); | ||||
1086 | } | ||||
1087 | } | ||||
1088 | |||||
1089 | int32_t | ||||
1090 | UnicodeString::lastIndexOf(const UChar *srcChars, | ||||
1091 | int32_t srcStart, | ||||
1092 | int32_t srcLength, | ||||
1093 | int32_t start, | ||||
1094 | int32_t length) const | ||||
1095 | { | ||||
1096 | if(isBogus() || srcChars == 0 || srcStart < 0 || srcLength == 0) { | ||||
1097 | return -1; | ||||
1098 | } | ||||
1099 | |||||
1100 | // UnicodeString does not find empty substrings | ||||
1101 | if(srcLength < 0 && srcChars[srcStart] == 0) { | ||||
1102 | return -1; | ||||
1103 | } | ||||
1104 | |||||
1105 | // get the indices within bounds | ||||
1106 | pinIndices(start, length); | ||||
1107 | |||||
1108 | // find the last occurrence of the substring | ||||
1109 | const UChar *array = getArrayStart(); | ||||
1110 | const UChar *match = u_strFindLastu_strFindLast_71(array + start, length, srcChars + srcStart, srcLength); | ||||
1111 | if(match == NULL__null) { | ||||
1112 | return -1; | ||||
1113 | } else { | ||||
1114 | return (int32_t)(match - array); | ||||
1115 | } | ||||
1116 | } | ||||
1117 | |||||
1118 | int32_t | ||||
1119 | UnicodeString::doLastIndexOf(UChar c, | ||||
1120 | int32_t start, | ||||
1121 | int32_t length) const | ||||
1122 | { | ||||
1123 | if(isBogus()) { | ||||
1124 | return -1; | ||||
1125 | } | ||||
1126 | |||||
1127 | // pin indices | ||||
1128 | pinIndices(start, length); | ||||
1129 | |||||
1130 | // find the last occurrence of c | ||||
1131 | const UChar *array = getArrayStart(); | ||||
1132 | const UChar *match = u_memrchru_memrchr_71(array + start, c, length); | ||||
1133 | if(match == NULL__null) { | ||||
1134 | return -1; | ||||
1135 | } else { | ||||
1136 | return (int32_t)(match - array); | ||||
1137 | } | ||||
1138 | } | ||||
1139 | |||||
1140 | int32_t | ||||
1141 | UnicodeString::doLastIndexOf(UChar32 c, | ||||
1142 | int32_t start, | ||||
1143 | int32_t length) const { | ||||
1144 | // pin indices | ||||
1145 | pinIndices(start, length); | ||||
1146 | |||||
1147 | // find the last occurrence of c | ||||
1148 | const UChar *array = getArrayStart(); | ||||
1149 | const UChar *match = u_memrchr32u_memrchr32_71(array + start, c, length); | ||||
1150 | if(match == NULL__null) { | ||||
1151 | return -1; | ||||
1152 | } else { | ||||
1153 | return (int32_t)(match - array); | ||||
1154 | } | ||||
1155 | } | ||||
1156 | |||||
1157 | //======================================== | ||||
1158 | // Write implementation | ||||
1159 | //======================================== | ||||
1160 | |||||
1161 | UnicodeString& | ||||
1162 | UnicodeString::findAndReplace(int32_t start, | ||||
1163 | int32_t length, | ||||
1164 | const UnicodeString& oldText, | ||||
1165 | int32_t oldStart, | ||||
1166 | int32_t oldLength, | ||||
1167 | const UnicodeString& newText, | ||||
1168 | int32_t newStart, | ||||
1169 | int32_t newLength) | ||||
1170 | { | ||||
1171 | if(isBogus() || oldText.isBogus() || newText.isBogus()) { | ||||
1172 | return *this; | ||||
1173 | } | ||||
1174 | |||||
1175 | pinIndices(start, length); | ||||
1176 | oldText.pinIndices(oldStart, oldLength); | ||||
1177 | newText.pinIndices(newStart, newLength); | ||||
1178 | |||||
1179 | if(oldLength == 0) { | ||||
1180 | return *this; | ||||
1181 | } | ||||
1182 | |||||
1183 | while(length > 0 && length >= oldLength) { | ||||
1184 | int32_t pos = indexOf(oldText, oldStart, oldLength, start, length); | ||||
1185 | if(pos < 0) { | ||||
1186 | // no more oldText's here: done | ||||
1187 | break; | ||||
1188 | } else { | ||||
1189 | // we found oldText, replace it by newText and go beyond it | ||||
1190 | replace(pos, oldLength, newText, newStart, newLength); | ||||
1191 | length -= pos + oldLength - start; | ||||
1192 | start = pos + newLength; | ||||
1193 | } | ||||
1194 | } | ||||
1195 | |||||
1196 | return *this; | ||||
1197 | } | ||||
1198 | |||||
1199 | |||||
1200 | void | ||||
1201 | UnicodeString::setToBogus() | ||||
1202 | { | ||||
1203 | releaseArray(); | ||||
1204 | |||||
1205 | fUnion.fFields.fLengthAndFlags = kIsBogus; | ||||
1206 | fUnion.fFields.fArray = 0; | ||||
1207 | fUnion.fFields.fCapacity = 0; | ||||
1208 | } | ||||
1209 | |||||
1210 | // turn a bogus string into an empty one | ||||
1211 | void | ||||
1212 | UnicodeString::unBogus() { | ||||
1213 | if(fUnion.fFields.fLengthAndFlags & kIsBogus) { | ||||
1214 | setToEmpty(); | ||||
1215 | } | ||||
1216 | } | ||||
1217 | |||||
1218 | const char16_t * | ||||
1219 | UnicodeString::getTerminatedBuffer() { | ||||
1220 | if(!isWritable()) { | ||||
1221 | return nullptr; | ||||
1222 | } | ||||
1223 | UChar *array = getArrayStart(); | ||||
1224 | int32_t len = length(); | ||||
1225 | if(len < getCapacity()) { | ||||
1226 | if(fUnion.fFields.fLengthAndFlags & kBufferIsReadonly) { | ||||
1227 | // If len<capacity on a read-only alias, then array[len] is | ||||
1228 | // either the original NUL (if constructed with (TRUE, s, length)) | ||||
1229 | // or one of the original string contents characters (if later truncated), | ||||
1230 | // therefore we can assume that array[len] is initialized memory. | ||||
1231 | if(array[len] == 0) { | ||||
1232 | return array; | ||||
1233 | } | ||||
1234 | } else if(((fUnion.fFields.fLengthAndFlags & kRefCounted) == 0 || refCount() == 1)) { | ||||
1235 | // kRefCounted: Do not write the NUL if the buffer is shared. | ||||
1236 | // That is mostly safe, except when the length of one copy was modified | ||||
1237 | // without copy-on-write, e.g., via truncate(newLength) or remove(void). | ||||
1238 | // Then the NUL would be written into the middle of another copy's string. | ||||
1239 | |||||
1240 | // Otherwise, the buffer is fully writable and it is anyway safe to write the NUL. | ||||
1241 | // Do not test if there is a NUL already because it might be uninitialized memory. | ||||
1242 | // (That would be safe, but tools like valgrind & Purify would complain.) | ||||
1243 | array[len] = 0; | ||||
1244 | return array; | ||||
1245 | } | ||||
1246 | } | ||||
1247 | if(len<INT32_MAX(2147483647) && cloneArrayIfNeeded(len+1)) { | ||||
1248 | array = getArrayStart(); | ||||
1249 | array[len] = 0; | ||||
1250 | return array; | ||||
1251 | } else { | ||||
1252 | return nullptr; | ||||
1253 | } | ||||
1254 | } | ||||
1255 | |||||
1256 | // setTo() analogous to the readonly-aliasing constructor with the same signature | ||||
1257 | UnicodeString & | ||||
1258 | UnicodeString::setTo(UBool isTerminated, | ||||
1259 | ConstChar16Ptr textPtr, | ||||
1260 | int32_t textLength) | ||||
1261 | { | ||||
1262 | if(fUnion.fFields.fLengthAndFlags & kOpenGetBuffer) { | ||||
1263 | // do not modify a string that has an "open" getBuffer(minCapacity) | ||||
1264 | return *this; | ||||
1265 | } | ||||
1266 | |||||
1267 | const UChar *text = textPtr; | ||||
1268 | if(text == NULL__null) { | ||||
1269 | // treat as an empty string, do not alias | ||||
1270 | releaseArray(); | ||||
1271 | setToEmpty(); | ||||
1272 | return *this; | ||||
1273 | } | ||||
1274 | |||||
1275 | if( textLength < -1 || | ||||
1276 | (textLength == -1 && !isTerminated) || | ||||
1277 | (textLength >= 0 && isTerminated && text[textLength] != 0) | ||||
1278 | ) { | ||||
1279 | setToBogus(); | ||||
1280 | return *this; | ||||
1281 | } | ||||
1282 | |||||
1283 | releaseArray(); | ||||
1284 | |||||
1285 | if(textLength == -1) { | ||||
1286 | // text is terminated, or else it would have failed the above test | ||||
1287 | textLength = u_strlenu_strlen_71(text); | ||||
1288 | } | ||||
1289 | fUnion.fFields.fLengthAndFlags = kReadonlyAlias; | ||||
1290 | setArray((UChar *)text, textLength, isTerminated ? textLength + 1 : textLength); | ||||
1291 | return *this; | ||||
1292 | } | ||||
1293 | |||||
1294 | // setTo() analogous to the writable-aliasing constructor with the same signature | ||||
1295 | UnicodeString & | ||||
1296 | UnicodeString::setTo(UChar *buffer, | ||||
1297 | int32_t buffLength, | ||||
1298 | int32_t buffCapacity) { | ||||
1299 | if(fUnion.fFields.fLengthAndFlags & kOpenGetBuffer) { | ||||
1300 | // do not modify a string that has an "open" getBuffer(minCapacity) | ||||
1301 | return *this; | ||||
1302 | } | ||||
1303 | |||||
1304 | if(buffer == NULL__null) { | ||||
1305 | // treat as an empty string, do not alias | ||||
1306 | releaseArray(); | ||||
1307 | setToEmpty(); | ||||
1308 | return *this; | ||||
1309 | } | ||||
1310 | |||||
1311 | if(buffLength < -1 || buffCapacity < 0 || buffLength > buffCapacity) { | ||||
1312 | setToBogus(); | ||||
1313 | return *this; | ||||
1314 | } else if(buffLength == -1) { | ||||
1315 | // buffLength = u_strlen(buff); but do not look beyond buffCapacity | ||||
1316 | const UChar *p = buffer, *limit = buffer + buffCapacity; | ||||
1317 | while(p != limit && *p != 0) { | ||||
1318 | ++p; | ||||
1319 | } | ||||
1320 | buffLength = (int32_t)(p - buffer); | ||||
1321 | } | ||||
1322 | |||||
1323 | releaseArray(); | ||||
1324 | |||||
1325 | fUnion.fFields.fLengthAndFlags = kWritableAlias; | ||||
1326 | setArray(buffer, buffLength, buffCapacity); | ||||
1327 | return *this; | ||||
1328 | } | ||||
1329 | |||||
1330 | UnicodeString &UnicodeString::setToUTF8(StringPiece utf8) { | ||||
1331 | unBogus(); | ||||
1332 | int32_t length = utf8.length(); | ||||
1333 | int32_t capacity; | ||||
1334 | // The UTF-16 string will be at most as long as the UTF-8 string. | ||||
1335 | if(length <= US_STACKBUF_SIZE) { | ||||
1336 | capacity = US_STACKBUF_SIZE; | ||||
1337 | } else { | ||||
1338 | capacity = length + 1; // +1 for the terminating NUL. | ||||
1339 | } | ||||
1340 | UChar *utf16 = getBuffer(capacity); | ||||
1341 | int32_t length16; | ||||
1342 | UErrorCode errorCode = U_ZERO_ERROR; | ||||
1343 | u_strFromUTF8WithSubu_strFromUTF8WithSub_71(utf16, getCapacity(), &length16, | ||||
1344 | utf8.data(), length, | ||||
1345 | 0xfffd, // Substitution character. | ||||
1346 | NULL__null, // Don't care about number of substitutions. | ||||
1347 | &errorCode); | ||||
1348 | releaseBuffer(length16); | ||||
1349 | if(U_FAILURE(errorCode)) { | ||||
1350 | setToBogus(); | ||||
1351 | } | ||||
1352 | return *this; | ||||
1353 | } | ||||
1354 | |||||
1355 | UnicodeString& | ||||
1356 | UnicodeString::setCharAt(int32_t offset, | ||||
1357 | UChar c) | ||||
1358 | { | ||||
1359 | int32_t len = length(); | ||||
1360 | if(cloneArrayIfNeeded() && len > 0) { | ||||
1361 | if(offset < 0) { | ||||
1362 | offset = 0; | ||||
1363 | } else if(offset >= len) { | ||||
1364 | offset = len - 1; | ||||
1365 | } | ||||
1366 | |||||
1367 | getArrayStart()[offset] = c; | ||||
1368 | } | ||||
1369 | return *this; | ||||
1370 | } | ||||
1371 | |||||
1372 | UnicodeString& | ||||
1373 | UnicodeString::replace(int32_t start, | ||||
1374 | int32_t _length, | ||||
1375 | UChar32 srcChar) { | ||||
1376 | UChar buffer[U16_MAX_LENGTH2]; | ||||
1377 | int32_t count = 0; | ||||
1378 | UBool isError = FALSE0; | ||||
1379 | U16_APPEND(buffer, count, U16_MAX_LENGTH, srcChar, isError)do { if((uint32_t)(srcChar)<=0xffff) { (buffer)[(count)++] =(uint16_t)(srcChar); } else if((uint32_t)(srcChar)<=0x10ffff && (count)+1<(2)) { (buffer)[(count)++]=(uint16_t )(((srcChar)>>10)+0xd7c0); (buffer)[(count)++]=(uint16_t )(((srcChar)&0x3ff)|0xdc00); } else { (isError)=true; } } while (false); | ||||
1380 | // We test isError so that the compiler does not complain that we don't. | ||||
1381 | // If isError (srcChar is not a valid code point) then count==0 which means | ||||
1382 | // we remove the source segment rather than replacing it with srcChar. | ||||
1383 | return doReplace(start, _length, buffer, 0, isError ? 0 : count); | ||||
1384 | } | ||||
1385 | |||||
1386 | UnicodeString& | ||||
1387 | UnicodeString::append(UChar32 srcChar) { | ||||
1388 | UChar buffer[U16_MAX_LENGTH2]; | ||||
1389 | int32_t _length = 0; | ||||
1390 | UBool isError = FALSE0; | ||||
1391 | U16_APPEND(buffer, _length, U16_MAX_LENGTH, srcChar, isError)do { if((uint32_t)(srcChar)<=0xffff) { (buffer)[(_length)++ ]=(uint16_t)(srcChar); } else if((uint32_t)(srcChar)<=0x10ffff && (_length)+1<(2)) { (buffer)[(_length)++]=(uint16_t )(((srcChar)>>10)+0xd7c0); (buffer)[(_length)++]=(uint16_t )(((srcChar)&0x3ff)|0xdc00); } else { (isError)=true; } } while (false); | ||||
1392 | // We test isError so that the compiler does not complain that we don't. | ||||
1393 | // If isError then _length==0 which turns the doAppend() into a no-op anyway. | ||||
1394 | return isError ? *this : doAppend(buffer, 0, _length); | ||||
1395 | } | ||||
1396 | |||||
1397 | UnicodeString& | ||||
1398 | UnicodeString::doReplace( int32_t start, | ||||
1399 | int32_t length, | ||||
1400 | const UnicodeString& src, | ||||
1401 | int32_t srcStart, | ||||
1402 | int32_t srcLength) | ||||
1403 | { | ||||
1404 | // pin the indices to legal values | ||||
1405 | src.pinIndices(srcStart, srcLength); | ||||
1406 | |||||
1407 | // get the characters from src | ||||
1408 | // and replace the range in ourselves with them | ||||
1409 | return doReplace(start, length, src.getArrayStart(), srcStart, srcLength); | ||||
1410 | } | ||||
1411 | |||||
1412 | UnicodeString& | ||||
1413 | UnicodeString::doReplace(int32_t start, | ||||
1414 | int32_t length, | ||||
1415 | const UChar *srcChars, | ||||
1416 | int32_t srcStart, | ||||
1417 | int32_t srcLength) | ||||
1418 | { | ||||
1419 | if(!isWritable()) { | ||||
1420 | return *this; | ||||
1421 | } | ||||
1422 | |||||
1423 | int32_t oldLength = this->length(); | ||||
1424 | |||||
1425 | // optimize (read-only alias).remove(0, start) and .remove(start, end) | ||||
1426 | if((fUnion.fFields.fLengthAndFlags&kBufferIsReadonly) && srcLength == 0) { | ||||
1427 | if(start == 0) { | ||||
1428 | // remove prefix by adjusting the array pointer | ||||
1429 | pinIndex(length); | ||||
1430 | fUnion.fFields.fArray += length; | ||||
1431 | fUnion.fFields.fCapacity -= length; | ||||
1432 | setLength(oldLength - length); | ||||
1433 | return *this; | ||||
1434 | } else { | ||||
1435 | pinIndex(start); | ||||
1436 | if(length >= (oldLength - start)) { | ||||
1437 | // remove suffix by reducing the length (like truncate()) | ||||
1438 | setLength(start); | ||||
1439 | fUnion.fFields.fCapacity = start; // not NUL-terminated any more | ||||
1440 | return *this; | ||||
1441 | } | ||||
1442 | } | ||||
1443 | } | ||||
1444 | |||||
1445 | if(start == oldLength) { | ||||
1446 | return doAppend(srcChars, srcStart, srcLength); | ||||
1447 | } | ||||
1448 | |||||
1449 | if(srcChars == 0) { | ||||
1450 | srcLength = 0; | ||||
1451 | } else { | ||||
1452 | // Perform all remaining operations relative to srcChars + srcStart. | ||||
1453 | // From this point forward, do not use srcStart. | ||||
1454 | srcChars += srcStart; | ||||
1455 | if (srcLength < 0) { | ||||
1456 | // get the srcLength if necessary | ||||
1457 | srcLength = u_strlenu_strlen_71(srcChars); | ||||
1458 | } | ||||
1459 | } | ||||
1460 | |||||
1461 | // pin the indices to legal values | ||||
1462 | pinIndices(start, length); | ||||
1463 | |||||
1464 | // Calculate the size of the string after the replace. | ||||
1465 | // Avoid int32_t overflow. | ||||
1466 | int32_t newLength = oldLength - length; | ||||
1467 | if(srcLength > (INT32_MAX(2147483647) - newLength)) { | ||||
1468 | setToBogus(); | ||||
1469 | return *this; | ||||
1470 | } | ||||
1471 | newLength += srcLength; | ||||
1472 | |||||
1473 | // Check for insertion into ourself | ||||
1474 | const UChar *oldArray = getArrayStart(); | ||||
1475 | if (isBufferWritable() && | ||||
1476 | oldArray < srcChars + srcLength && | ||||
1477 | srcChars < oldArray + oldLength) { | ||||
1478 | // Copy into a new UnicodeString and start over | ||||
1479 | UnicodeString copy(srcChars, srcLength); | ||||
1480 | if (copy.isBogus()) { | ||||
1481 | setToBogus(); | ||||
1482 | return *this; | ||||
1483 | } | ||||
1484 | return doReplace(start, length, copy.getArrayStart(), 0, srcLength); | ||||
1485 | } | ||||
1486 | |||||
1487 | // cloneArrayIfNeeded(doCopyArray=FALSE) may change fArray but will not copy the current contents; | ||||
1488 | // therefore we need to keep the current fArray | ||||
1489 | UChar oldStackBuffer[US_STACKBUF_SIZE]; | ||||
1490 | if((fUnion.fFields.fLengthAndFlags&kUsingStackBuffer) && (newLength > US_STACKBUF_SIZE)) { | ||||
1491 | // copy the stack buffer contents because it will be overwritten with | ||||
1492 | // fUnion.fFields values | ||||
1493 | u_memcpyu_memcpy_71(oldStackBuffer, oldArray, oldLength); | ||||
1494 | oldArray = oldStackBuffer; | ||||
1495 | } | ||||
1496 | |||||
1497 | // clone our array and allocate a bigger array if needed | ||||
1498 | int32_t *bufferToDelete = 0; | ||||
1499 | if(!cloneArrayIfNeeded(newLength, getGrowCapacity(newLength), | ||||
1500 | FALSE0, &bufferToDelete) | ||||
1501 | ) { | ||||
1502 | return *this; | ||||
1503 | } | ||||
1504 | |||||
1505 | // now do the replace | ||||
1506 | |||||
1507 | UChar *newArray = getArrayStart(); | ||||
1508 | if(newArray != oldArray) { | ||||
1509 | // if fArray changed, then we need to copy everything except what will change | ||||
1510 | us_arrayCopy(oldArray, 0, newArray, 0, start); | ||||
1511 | us_arrayCopy(oldArray, start + length, | ||||
1512 | newArray, start + srcLength, | ||||
1513 | oldLength - (start + length)); | ||||
1514 | } else if(length != srcLength) { | ||||
1515 | // fArray did not change; copy only the portion that isn't changing, leaving a hole | ||||
1516 | us_arrayCopy(oldArray, start + length, | ||||
1517 | newArray, start + srcLength, | ||||
1518 | oldLength - (start + length)); | ||||
1519 | } | ||||
1520 | |||||
1521 | // now fill in the hole with the new string | ||||
1522 | us_arrayCopy(srcChars, 0, newArray, start, srcLength); | ||||
1523 | |||||
1524 | setLength(newLength); | ||||
1525 | |||||
1526 | // delayed delete in case srcChars == fArray when we started, and | ||||
1527 | // to keep oldArray alive for the above operations | ||||
1528 | if (bufferToDelete) { | ||||
1529 | uprv_freeuprv_free_71(bufferToDelete); | ||||
1530 | } | ||||
1531 | |||||
1532 | return *this; | ||||
1533 | } | ||||
1534 | |||||
1535 | // Versions of doReplace() only for append() variants. | ||||
1536 | // doReplace() and doAppend() optimize for different cases. | ||||
1537 | |||||
1538 | UnicodeString& | ||||
1539 | UnicodeString::doAppend(const UnicodeString& src, int32_t srcStart, int32_t srcLength) { | ||||
1540 | if(srcLength == 0) { | ||||
1541 | return *this; | ||||
1542 | } | ||||
1543 | |||||
1544 | // pin the indices to legal values | ||||
1545 | src.pinIndices(srcStart, srcLength); | ||||
1546 | return doAppend(src.getArrayStart(), srcStart, srcLength); | ||||
1547 | } | ||||
1548 | |||||
1549 | UnicodeString& | ||||
1550 | UnicodeString::doAppend(const UChar *srcChars, int32_t srcStart, int32_t srcLength) { | ||||
1551 | if(!isWritable() || srcLength == 0 || srcChars == NULL__null) { | ||||
1552 | return *this; | ||||
1553 | } | ||||
1554 | |||||
1555 | // Perform all remaining operations relative to srcChars + srcStart. | ||||
1556 | // From this point forward, do not use srcStart. | ||||
1557 | srcChars += srcStart; | ||||
1558 | |||||
1559 | if(srcLength < 0) { | ||||
1560 | // get the srcLength if necessary | ||||
1561 | if((srcLength = u_strlenu_strlen_71(srcChars)) == 0) { | ||||
1562 | return *this; | ||||
1563 | } | ||||
1564 | } | ||||
1565 | |||||
1566 | int32_t oldLength = length(); | ||||
1567 | int32_t newLength; | ||||
1568 | if (uprv_add32_overflowuprv_add32_overflow_71(oldLength, srcLength, &newLength)) { | ||||
1569 | setToBogus(); | ||||
1570 | return *this; | ||||
1571 | } | ||||
1572 | |||||
1573 | // Check for append onto ourself | ||||
1574 | const UChar* oldArray = getArrayStart(); | ||||
1575 | if (isBufferWritable() && | ||||
1576 | oldArray < srcChars + srcLength && | ||||
1577 | srcChars < oldArray + oldLength) { | ||||
1578 | // Copy into a new UnicodeString and start over | ||||
1579 | UnicodeString copy(srcChars, srcLength); | ||||
1580 | if (copy.isBogus()) { | ||||
1581 | setToBogus(); | ||||
1582 | return *this; | ||||
1583 | } | ||||
1584 | return doAppend(copy.getArrayStart(), 0, srcLength); | ||||
1585 | } | ||||
1586 | |||||
1587 | // optimize append() onto a large-enough, owned string | ||||
1588 | if((newLength <= getCapacity() && isBufferWritable()) || | ||||
1589 | cloneArrayIfNeeded(newLength, getGrowCapacity(newLength))) { | ||||
1590 | UChar *newArray = getArrayStart(); | ||||
1591 | // Do not copy characters when | ||||
1592 | // UChar *buffer=str.getAppendBuffer(...); | ||||
1593 | // is followed by | ||||
1594 | // str.append(buffer, length); | ||||
1595 | // or | ||||
1596 | // str.appendString(buffer, length) | ||||
1597 | // or similar. | ||||
1598 | if(srcChars != newArray + oldLength) { | ||||
1599 | us_arrayCopy(srcChars, 0, newArray, oldLength, srcLength); | ||||
1600 | } | ||||
1601 | setLength(newLength); | ||||
1602 | } | ||||
1603 | return *this; | ||||
1604 | } | ||||
1605 | |||||
1606 | /** | ||||
1607 | * Replaceable API | ||||
1608 | */ | ||||
1609 | void | ||||
1610 | UnicodeString::handleReplaceBetween(int32_t start, | ||||
1611 | int32_t limit, | ||||
1612 | const UnicodeString& text) { | ||||
1613 | replaceBetween(start, limit, text); | ||||
1614 | } | ||||
1615 | |||||
1616 | /** | ||||
1617 | * Replaceable API | ||||
1618 | */ | ||||
1619 | void | ||||
1620 | UnicodeString::copy(int32_t start, int32_t limit, int32_t dest) { | ||||
1621 | if (limit <= start) { | ||||
1622 | return; // Nothing to do; avoid bogus malloc call | ||||
1623 | } | ||||
1624 | UChar* text = (UChar*) uprv_mallocuprv_malloc_71( sizeof(UChar) * (limit - start) ); | ||||
1625 | // Check to make sure text is not null. | ||||
1626 | if (text != NULL__null) { | ||||
1627 | extractBetween(start, limit, text, 0); | ||||
1628 | insert(dest, text, 0, limit - start); | ||||
1629 | uprv_freeuprv_free_71(text); | ||||
1630 | } | ||||
1631 | } | ||||
1632 | |||||
1633 | /** | ||||
1634 | * Replaceable API | ||||
1635 | * | ||||
1636 | * NOTE: This is for the Replaceable class. There is no rep.cpp, | ||||
1637 | * so we implement this function here. | ||||
1638 | */ | ||||
1639 | UBool Replaceable::hasMetaData() const { | ||||
1640 | return TRUE1; | ||||
1641 | } | ||||
1642 | |||||
1643 | /** | ||||
1644 | * Replaceable API | ||||
1645 | */ | ||||
1646 | UBool UnicodeString::hasMetaData() const { | ||||
1647 | return FALSE0; | ||||
1648 | } | ||||
1649 | |||||
1650 | UnicodeString& | ||||
1651 | UnicodeString::doReverse(int32_t start, int32_t length) { | ||||
1652 | if(length <= 1 || !cloneArrayIfNeeded()) { | ||||
1653 | return *this; | ||||
1654 | } | ||||
1655 | |||||
1656 | // pin the indices to legal values | ||||
1657 | pinIndices(start, length); | ||||
1658 | if(length <= 1) { // pinIndices() might have shrunk the length | ||||
1659 | return *this; | ||||
1660 | } | ||||
1661 | |||||
1662 | UChar *left = getArrayStart() + start; | ||||
1663 | UChar *right = left + length - 1; // -1 for inclusive boundary (length>=2) | ||||
1664 | UChar swap; | ||||
1665 | UBool hasSupplementary = FALSE0; | ||||
1666 | |||||
1667 | // Before the loop we know left<right because length>=2. | ||||
1668 | do { | ||||
1669 | hasSupplementary |= (UBool)U16_IS_LEAD(swap = *left)(((swap = *left)&0xfffffc00)==0xd800); | ||||
1670 | hasSupplementary |= (UBool)U16_IS_LEAD(*left++ = *right)(((*left++ = *right)&0xfffffc00)==0xd800); | ||||
1671 | *right-- = swap; | ||||
1672 | } while(left < right); | ||||
1673 | // Make sure to test the middle code unit of an odd-length string. | ||||
1674 | // Redundant if the length is even. | ||||
1675 | hasSupplementary |= (UBool)U16_IS_LEAD(*left)(((*left)&0xfffffc00)==0xd800); | ||||
1676 | |||||
1677 | /* if there are supplementary code points in the reversed range, then re-swap their surrogates */ | ||||
1678 | if(hasSupplementary) { | ||||
1679 | UChar swap2; | ||||
1680 | |||||
1681 | left = getArrayStart() + start; | ||||
1682 | right = left + length - 1; // -1 so that we can look at *(left+1) if left<right | ||||
1683 | while(left < right) { | ||||
1684 | if(U16_IS_TRAIL(swap = *left)(((swap = *left)&0xfffffc00)==0xdc00) && U16_IS_LEAD(swap2 = *(left + 1))(((swap2 = *(left + 1))&0xfffffc00)==0xd800)) { | ||||
1685 | *left++ = swap2; | ||||
1686 | *left++ = swap; | ||||
1687 | } else { | ||||
1688 | ++left; | ||||
1689 | } | ||||
1690 | } | ||||
1691 | } | ||||
1692 | |||||
1693 | return *this; | ||||
1694 | } | ||||
1695 | |||||
1696 | UBool | ||||
1697 | UnicodeString::padLeading(int32_t targetLength, | ||||
1698 | UChar padChar) | ||||
1699 | { | ||||
1700 | int32_t oldLength = length(); | ||||
1701 | if(oldLength >= targetLength || !cloneArrayIfNeeded(targetLength)) { | ||||
1702 | return FALSE0; | ||||
1703 | } else { | ||||
1704 | // move contents up by padding width | ||||
1705 | UChar *array = getArrayStart(); | ||||
1706 | int32_t start = targetLength - oldLength; | ||||
1707 | us_arrayCopy(array, 0, array, start, oldLength); | ||||
1708 | |||||
1709 | // fill in padding character | ||||
1710 | while(--start >= 0) { | ||||
1711 | array[start] = padChar; | ||||
1712 | } | ||||
1713 | setLength(targetLength); | ||||
1714 | return TRUE1; | ||||
1715 | } | ||||
1716 | } | ||||
1717 | |||||
1718 | UBool | ||||
1719 | UnicodeString::padTrailing(int32_t targetLength, | ||||
1720 | UChar padChar) | ||||
1721 | { | ||||
1722 | int32_t oldLength = length(); | ||||
1723 | if(oldLength >= targetLength || !cloneArrayIfNeeded(targetLength)) { | ||||
1724 | return FALSE0; | ||||
1725 | } else { | ||||
1726 | // fill in padding character | ||||
1727 | UChar *array = getArrayStart(); | ||||
1728 | int32_t length = targetLength; | ||||
1729 | while(--length >= oldLength) { | ||||
1730 | array[length] = padChar; | ||||
1731 | } | ||||
1732 | setLength(targetLength); | ||||
1733 | return TRUE1; | ||||
1734 | } | ||||
1735 | } | ||||
1736 | |||||
1737 | //======================================== | ||||
1738 | // Hashing | ||||
1739 | //======================================== | ||||
1740 | int32_t | ||||
1741 | UnicodeString::doHashCode() const | ||||
1742 | { | ||||
1743 | /* Delegate hash computation to uhash. This makes UnicodeString | ||||
1744 | * hashing consistent with UChar* hashing. */ | ||||
1745 | int32_t hashCode = ustr_hashUCharsNustr_hashUCharsN_71(getArrayStart(), length()); | ||||
1746 | if (hashCode == kInvalidHashCode) { | ||||
1747 | hashCode = kEmptyHashCode; | ||||
1748 | } | ||||
1749 | return hashCode; | ||||
1750 | } | ||||
1751 | |||||
1752 | //======================================== | ||||
1753 | // External Buffer | ||||
1754 | //======================================== | ||||
1755 | |||||
1756 | char16_t * | ||||
1757 | UnicodeString::getBuffer(int32_t minCapacity) { | ||||
1758 | if(minCapacity>=-1 && cloneArrayIfNeeded(minCapacity)) { | ||||
1759 | fUnion.fFields.fLengthAndFlags|=kOpenGetBuffer; | ||||
1760 | setZeroLength(); | ||||
1761 | return getArrayStart(); | ||||
1762 | } else { | ||||
1763 | return nullptr; | ||||
1764 | } | ||||
1765 | } | ||||
1766 | |||||
1767 | void | ||||
1768 | UnicodeString::releaseBuffer(int32_t newLength) { | ||||
1769 | if(fUnion.fFields.fLengthAndFlags&kOpenGetBuffer && newLength>=-1) { | ||||
1770 | // set the new fLength | ||||
1771 | int32_t capacity=getCapacity(); | ||||
1772 | if(newLength==-1) { | ||||
1773 | // the new length is the string length, capped by fCapacity | ||||
1774 | const UChar *array=getArrayStart(), *p=array, *limit=array+capacity; | ||||
1775 | while(p<limit && *p!=0) { | ||||
1776 | ++p; | ||||
1777 | } | ||||
1778 | newLength=(int32_t)(p-array); | ||||
1779 | } else if(newLength>capacity) { | ||||
1780 | newLength=capacity; | ||||
1781 | } | ||||
1782 | setLength(newLength); | ||||
1783 | fUnion.fFields.fLengthAndFlags&=~kOpenGetBuffer; | ||||
1784 | } | ||||
1785 | } | ||||
1786 | |||||
1787 | //======================================== | ||||
1788 | // Miscellaneous | ||||
1789 | //======================================== | ||||
1790 | UBool | ||||
1791 | UnicodeString::cloneArrayIfNeeded(int32_t newCapacity, | ||||
1792 | int32_t growCapacity, | ||||
1793 | UBool doCopyArray, | ||||
1794 | int32_t **pBufferToDelete, | ||||
1795 | UBool forceClone) { | ||||
1796 | // default parameters need to be static, therefore | ||||
1797 | // the defaults are -1 to have convenience defaults | ||||
1798 | if(newCapacity == -1) { | ||||
1799 | newCapacity = getCapacity(); | ||||
1800 | } | ||||
1801 | |||||
1802 | // while a getBuffer(minCapacity) is "open", | ||||
1803 | // prevent any modifications of the string by returning FALSE here | ||||
1804 | // if the string is bogus, then only an assignment or similar can revive it | ||||
1805 | if(!isWritable()) { | ||||
1806 | return FALSE0; | ||||
1807 | } | ||||
1808 | |||||
1809 | /* | ||||
1810 | * We need to make a copy of the array if | ||||
1811 | * the buffer is read-only, or | ||||
1812 | * the buffer is refCounted (shared), and refCount>1, or | ||||
1813 | * the buffer is too small. | ||||
1814 | * Return FALSE if memory could not be allocated. | ||||
1815 | */ | ||||
1816 | if(forceClone
| ||||
1817 | fUnion.fFields.fLengthAndFlags & kBufferIsReadonly || | ||||
1818 | (fUnion.fFields.fLengthAndFlags & kRefCounted && refCount() > 1) || | ||||
1819 | newCapacity > getCapacity() | ||||
1820 | ) { | ||||
1821 | // check growCapacity for default value and use of the stack buffer | ||||
1822 | if(growCapacity
| ||||
1823 | growCapacity = newCapacity; | ||||
1824 | } else if(newCapacity <= US_STACKBUF_SIZE && growCapacity > US_STACKBUF_SIZE) { | ||||
1825 | growCapacity = US_STACKBUF_SIZE; | ||||
1826 | } | ||||
1827 | |||||
1828 | // save old values | ||||
1829 | UChar oldStackBuffer[US_STACKBUF_SIZE]; | ||||
1830 | UChar *oldArray; | ||||
1831 | int32_t oldLength = length(); | ||||
1832 | int16_t flags = fUnion.fFields.fLengthAndFlags; | ||||
1833 | |||||
1834 | if(flags&kUsingStackBuffer) { | ||||
1835 | U_ASSERT(!(flags&kRefCounted))(void)0; /* kRefCounted and kUsingStackBuffer are mutally exclusive */ | ||||
1836 | if(doCopyArray
| ||||
1837 | // copy the stack buffer contents because it will be overwritten with | ||||
1838 | // fUnion.fFields values | ||||
1839 | us_arrayCopy(fUnion.fStackFields.fBuffer, 0, oldStackBuffer, 0, oldLength); | ||||
1840 | oldArray = oldStackBuffer; | ||||
1841 | } else { | ||||
1842 | oldArray = NULL__null; // no need to copy from the stack buffer to itself | ||||
1843 | } | ||||
1844 | } else { | ||||
1845 | oldArray = fUnion.fFields.fArray; | ||||
1846 | U_ASSERT(oldArray!=NULL)(void)0; /* when stack buffer is not used, oldArray must have a non-NULL reference */ | ||||
1847 | } | ||||
1848 | |||||
1849 | // allocate a new array | ||||
1850 | if(allocate(growCapacity) || | ||||
1851 | (newCapacity
| ||||
1852 | ) { | ||||
1853 | if(doCopyArray) { | ||||
1854 | // copy the contents | ||||
1855 | // do not copy more than what fits - it may be smaller than before | ||||
1856 | int32_t minLength = oldLength; | ||||
1857 | newCapacity = getCapacity(); | ||||
1858 | if(newCapacity < minLength) { | ||||
1859 | minLength = newCapacity; | ||||
1860 | } | ||||
1861 | if(oldArray != NULL__null) { | ||||
1862 | us_arrayCopy(oldArray, 0, getArrayStart(), 0, minLength); | ||||
1863 | } | ||||
1864 | setLength(minLength); | ||||
1865 | } else { | ||||
1866 | setZeroLength(); | ||||
1867 | } | ||||
1868 | |||||
1869 | // release the old array | ||||
1870 | if(flags & kRefCounted) { | ||||
1871 | // the array is refCounted; decrement and release if 0 | ||||
1872 | u_atomic_int32_t *pRefCount = ((u_atomic_int32_t *)oldArray - 1); | ||||
1873 | if(umtx_atomic_dec(pRefCount) == 0) { | ||||
1874 | if(pBufferToDelete == 0) { | ||||
1875 | // Note: cast to (void *) is needed with MSVC, where u_atomic_int32_t | ||||
1876 | // is defined as volatile. (Volatile has useful non-standard behavior | ||||
1877 | // with this compiler.) | ||||
1878 | uprv_freeuprv_free_71((void *)pRefCount); | ||||
1879 | } else { | ||||
1880 | // the caller requested to delete it himself | ||||
1881 | *pBufferToDelete = (int32_t *)pRefCount; | ||||
1882 | } | ||||
1883 | } | ||||
1884 | } | ||||
1885 | } else { | ||||
1886 | // not enough memory for growCapacity and not even for the smaller newCapacity | ||||
1887 | // reset the old values for setToBogus() to release the array | ||||
1888 | if(!(flags&kUsingStackBuffer)) { | ||||
1889 | fUnion.fFields.fArray = oldArray; | ||||
1890 | } | ||||
1891 | fUnion.fFields.fLengthAndFlags = flags; | ||||
1892 | setToBogus(); | ||||
1893 | return FALSE0; | ||||
1894 | } | ||||
1895 | } | ||||
1896 | return TRUE1; | ||||
1897 | } | ||||
1898 | |||||
1899 | // UnicodeStringAppendable ------------------------------------------------- *** | ||||
1900 | |||||
1901 | UnicodeStringAppendable::~UnicodeStringAppendable() {} | ||||
1902 | |||||
1903 | UBool | ||||
1904 | UnicodeStringAppendable::appendCodeUnit(UChar c) { | ||||
1905 | return str.doAppend(&c, 0, 1).isWritable(); | ||||
1906 | } | ||||
1907 | |||||
1908 | UBool | ||||
1909 | UnicodeStringAppendable::appendCodePoint(UChar32 c) { | ||||
1910 | UChar buffer[U16_MAX_LENGTH2]; | ||||
1911 | int32_t cLength = 0; | ||||
1912 | UBool isError = FALSE0; | ||||
1913 | U16_APPEND(buffer, cLength, U16_MAX_LENGTH, c, isError)do { if((uint32_t)(c)<=0xffff) { (buffer)[(cLength)++]=(uint16_t )(c); } else if((uint32_t)(c)<=0x10ffff && (cLength )+1<(2)) { (buffer)[(cLength)++]=(uint16_t)(((c)>>10 )+0xd7c0); (buffer)[(cLength)++]=(uint16_t)(((c)&0x3ff)|0xdc00 ); } else { (isError)=true; } } while (false); | ||||
1914 | return !isError && str.doAppend(buffer, 0, cLength).isWritable(); | ||||
1915 | } | ||||
1916 | |||||
1917 | UBool | ||||
1918 | UnicodeStringAppendable::appendString(const UChar *s, int32_t length) { | ||||
1919 | return str.doAppend(s, 0, length).isWritable(); | ||||
1920 | } | ||||
1921 | |||||
1922 | UBool | ||||
1923 | UnicodeStringAppendable::reserveAppendCapacity(int32_t appendCapacity) { | ||||
1924 | return str.cloneArrayIfNeeded(str.length() + appendCapacity); | ||||
| |||||
1925 | } | ||||
1926 | |||||
1927 | UChar * | ||||
1928 | UnicodeStringAppendable::getAppendBuffer(int32_t minCapacity, | ||||
1929 | int32_t desiredCapacityHint, | ||||
1930 | UChar *scratch, int32_t scratchCapacity, | ||||
1931 | int32_t *resultCapacity) { | ||||
1932 | if(minCapacity < 1 || scratchCapacity < minCapacity) { | ||||
1933 | *resultCapacity = 0; | ||||
1934 | return NULL__null; | ||||
1935 | } | ||||
1936 | int32_t oldLength = str.length(); | ||||
1937 | if(minCapacity <= (kMaxCapacity - oldLength) && | ||||
1938 | desiredCapacityHint <= (kMaxCapacity - oldLength) && | ||||
1939 | str.cloneArrayIfNeeded(oldLength + minCapacity, oldLength + desiredCapacityHint)) { | ||||
1940 | *resultCapacity = str.getCapacity() - oldLength; | ||||
1941 | return str.getArrayStart() + oldLength; | ||||
1942 | } | ||||
1943 | *resultCapacity = scratchCapacity; | ||||
1944 | return scratch; | ||||
1945 | } | ||||
1946 | |||||
1947 | U_NAMESPACE_END} | ||||
1948 | |||||
1949 | U_NAMESPACE_USEusing namespace icu_71; | ||||
1950 | |||||
1951 | U_CAPIextern "C" int32_t U_EXPORT2 | ||||
1952 | uhash_hashUnicodeStringuhash_hashUnicodeString_71(const UElement key) { | ||||
1953 | const UnicodeString *str = (const UnicodeString*) key.pointer; | ||||
1954 | return (str == NULL__null) ? 0 : str->hashCode(); | ||||
1955 | } | ||||
1956 | |||||
1957 | // Moved here from uhash_us.cpp so that using a UVector of UnicodeString* | ||||
1958 | // does not depend on hashtable code. | ||||
1959 | U_CAPIextern "C" UBool U_EXPORT2 | ||||
1960 | uhash_compareUnicodeStringuhash_compareUnicodeString_71(const UElement key1, const UElement key2) { | ||||
1961 | const UnicodeString *str1 = (const UnicodeString*) key1.pointer; | ||||
1962 | const UnicodeString *str2 = (const UnicodeString*) key2.pointer; | ||||
1963 | if (str1 == str2) { | ||||
1964 | return TRUE1; | ||||
1965 | } | ||||
1966 | if (str1 == NULL__null || str2 == NULL__null) { | ||||
1967 | return FALSE0; | ||||
1968 | } | ||||
1969 | return *str1 == *str2; | ||||
1970 | } | ||||
1971 | |||||
1972 | #ifdef U_STATIC_IMPLEMENTATION1 | ||||
1973 | /* | ||||
1974 | This should never be called. It is defined here to make sure that the | ||||
1975 | virtual vector deleting destructor is defined within unistr.cpp. | ||||
1976 | The vector deleting destructor is already a part of UObject, | ||||
1977 | but defining it here makes sure that it is included with this object file. | ||||
1978 | This makes sure that static library dependencies are kept to a minimum. | ||||
1979 | */ | ||||
1980 | #if defined(__clang__1) || U_GCC_MAJOR_MINOR(4 * 100 + 2) >= 1100 | ||||
1981 | #pragma GCC diagnostic push | ||||
1982 | #pragma GCC diagnostic ignored "-Wunused-function" | ||||
1983 | static void uprv_UnicodeStringDummy(void) { | ||||
1984 | delete [] (new UnicodeString[2]); | ||||
1985 | } | ||||
1986 | #pragma GCC diagnostic pop | ||||
1987 | #endif | ||||
1988 | #endif |
1 | // © 2016 and later: Unicode, Inc. and others. | |||
2 | // License & terms of use: http://www.unicode.org/copyright.html | |||
3 | /* | |||
4 | ********************************************************************** | |||
5 | * Copyright (C) 1997-2015, International Business Machines | |||
6 | * Corporation and others. All Rights Reserved. | |||
7 | ********************************************************************** | |||
8 | * | |||
9 | * File UMUTEX.H | |||
10 | * | |||
11 | * Modification History: | |||
12 | * | |||
13 | * Date Name Description | |||
14 | * 04/02/97 aliu Creation. | |||
15 | * 04/07/99 srl rewrite - C interface, multiple mutices | |||
16 | * 05/13/99 stephen Changed to umutex (from cmutex) | |||
17 | ****************************************************************************** | |||
18 | */ | |||
19 | ||||
20 | #ifndef UMUTEX_H | |||
21 | #define UMUTEX_H | |||
22 | ||||
23 | #include <atomic> | |||
24 | #include <condition_variable> | |||
25 | #include <mutex> | |||
26 | #include <type_traits> | |||
27 | ||||
28 | #include "unicode/utypes.h" | |||
29 | #include "unicode/uclean.h" | |||
30 | #include "unicode/uobject.h" | |||
31 | ||||
32 | #include "putilimp.h" | |||
33 | ||||
34 | #if defined(U_USER_ATOMICS_H) || defined(U_USER_MUTEX_H) | |||
35 | // Support for including an alternate implementation of atomic & mutex operations has been withdrawn. | |||
36 | // See issue ICU-20185. | |||
37 | #error U_USER_ATOMICS and U_USER_MUTEX_H are not supported | |||
38 | #endif | |||
39 | ||||
40 | // Export an explicit template instantiation of std::atomic<int32_t>. | |||
41 | // When building DLLs for Windows this is required as it is used as a data member of the exported SharedObject class. | |||
42 | // See digitlst.h, pluralaffix.h, datefmt.h, and others for similar examples. | |||
43 | // | |||
44 | // Similar story for std::atomic<std::mutex *>, and the exported UMutex class. | |||
45 | #if U_PF_WINDOWS1000 <= U_PLATFORM4000 && U_PLATFORM4000 <= U_PF_CYGWIN1900 && !defined(U_IN_DOXYGEN) | |||
46 | #if defined(__clang__1) || defined(_MSC_VER) | |||
47 | #if defined(__clang__1) | |||
48 | // Suppress the warning that the explicit instantiation after explicit specialization has no effect. | |||
49 | #pragma clang diagnostic push | |||
50 | #pragma clang diagnostic ignored "-Winstantiation-after-specialization" | |||
51 | #endif | |||
52 | template struct U_COMMON_API std::atomic<int32_t>; | |||
53 | template struct U_COMMON_API std::atomic<std::mutex *>; | |||
54 | #if defined(__clang__1) | |||
55 | #pragma clang diagnostic pop | |||
56 | #endif | |||
57 | #elif defined(__GNUC__4) | |||
58 | // For GCC this class is already exported/visible, so no need for U_COMMON_API. | |||
59 | template struct std::atomic<int32_t>; | |||
60 | template struct std::atomic<std::mutex *>; | |||
61 | #endif | |||
62 | #endif | |||
63 | ||||
64 | ||||
65 | U_NAMESPACE_BEGINnamespace icu_71 { | |||
66 | ||||
67 | /**************************************************************************** | |||
68 | * | |||
69 | * Low Level Atomic Operations, ICU wrappers for. | |||
70 | * | |||
71 | ****************************************************************************/ | |||
72 | ||||
73 | typedef std::atomic<int32_t> u_atomic_int32_t; | |||
74 | #define ATOMIC_INT32_T_INITIALIZER(val){ val } ATOMIC_VAR_INIT(val){ val } | |||
75 | ||||
76 | inline int32_t umtx_loadAcquire(u_atomic_int32_t &var) { | |||
77 | return var.load(std::memory_order_acquire); | |||
78 | } | |||
79 | ||||
80 | inline void umtx_storeRelease(u_atomic_int32_t &var, int32_t val) { | |||
81 | var.store(val, std::memory_order_release); | |||
82 | } | |||
83 | ||||
84 | inline int32_t umtx_atomic_inc(u_atomic_int32_t *var) { | |||
85 | return var->fetch_add(1) + 1; | |||
86 | } | |||
87 | ||||
88 | inline int32_t umtx_atomic_dec(u_atomic_int32_t *var) { | |||
89 | return var->fetch_sub(1) - 1; | |||
| ||||
90 | } | |||
91 | ||||
92 | ||||
93 | /************************************************************************************************* | |||
94 | * | |||
95 | * UInitOnce Definitions. | |||
96 | * | |||
97 | *************************************************************************************************/ | |||
98 | ||||
99 | struct UInitOnce { | |||
100 | u_atomic_int32_t fState; | |||
101 | UErrorCode fErrCode; | |||
102 | void reset() {fState = 0;} | |||
103 | UBool isReset() {return umtx_loadAcquire(fState) == 0;} | |||
104 | // Note: isReset() is used by service registration code. | |||
105 | // Thread safety of this usage needs review. | |||
106 | }; | |||
107 | ||||
108 | #define U_INITONCE_INITIALIZER{{ 0 }, U_ZERO_ERROR} {ATOMIC_INT32_T_INITIALIZER(0){ 0 }, U_ZERO_ERROR} | |||
109 | ||||
110 | ||||
111 | U_COMMON_API UBool U_EXPORT2 umtx_initImplPreInit(UInitOnce &); | |||
112 | U_COMMON_API void U_EXPORT2 umtx_initImplPostInit(UInitOnce &); | |||
113 | ||||
114 | template<class T> void umtx_initOnce(UInitOnce &uio, T *obj, void (U_CALLCONV T::*fp)()) { | |||
115 | if (umtx_loadAcquire(uio.fState) == 2) { | |||
116 | return; | |||
117 | } | |||
118 | if (umtx_initImplPreInit(uio)) { | |||
119 | (obj->*fp)(); | |||
120 | umtx_initImplPostInit(uio); | |||
121 | } | |||
122 | } | |||
123 | ||||
124 | ||||
125 | // umtx_initOnce variant for plain functions, or static class functions. | |||
126 | // No context parameter. | |||
127 | inline void umtx_initOnce(UInitOnce &uio, void (U_CALLCONV *fp)()) { | |||
128 | if (umtx_loadAcquire(uio.fState) == 2) { | |||
129 | return; | |||
130 | } | |||
131 | if (umtx_initImplPreInit(uio)) { | |||
132 | (*fp)(); | |||
133 | umtx_initImplPostInit(uio); | |||
134 | } | |||
135 | } | |||
136 | ||||
137 | // umtx_initOnce variant for plain functions, or static class functions. | |||
138 | // With ErrorCode, No context parameter. | |||
139 | inline void umtx_initOnce(UInitOnce &uio, void (U_CALLCONV *fp)(UErrorCode &), UErrorCode &errCode) { | |||
140 | if (U_FAILURE(errCode)) { | |||
141 | return; | |||
142 | } | |||
143 | if (umtx_loadAcquire(uio.fState) != 2 && umtx_initImplPreInit(uio)) { | |||
144 | // We run the initialization. | |||
145 | (*fp)(errCode); | |||
146 | uio.fErrCode = errCode; | |||
147 | umtx_initImplPostInit(uio); | |||
148 | } else { | |||
149 | // Someone else already ran the initialization. | |||
150 | if (U_FAILURE(uio.fErrCode)) { | |||
151 | errCode = uio.fErrCode; | |||
152 | } | |||
153 | } | |||
154 | } | |||
155 | ||||
156 | // umtx_initOnce variant for plain functions, or static class functions, | |||
157 | // with a context parameter. | |||
158 | template<class T> void umtx_initOnce(UInitOnce &uio, void (U_CALLCONV *fp)(T), T context) { | |||
159 | if (umtx_loadAcquire(uio.fState) == 2) { | |||
160 | return; | |||
161 | } | |||
162 | if (umtx_initImplPreInit(uio)) { | |||
163 | (*fp)(context); | |||
164 | umtx_initImplPostInit(uio); | |||
165 | } | |||
166 | } | |||
167 | ||||
168 | // umtx_initOnce variant for plain functions, or static class functions, | |||
169 | // with a context parameter and an error code. | |||
170 | template<class T> void umtx_initOnce(UInitOnce &uio, void (U_CALLCONV *fp)(T, UErrorCode &), T context, UErrorCode &errCode) { | |||
171 | if (U_FAILURE(errCode)) { | |||
172 | return; | |||
173 | } | |||
174 | if (umtx_loadAcquire(uio.fState) != 2 && umtx_initImplPreInit(uio)) { | |||
175 | // We run the initialization. | |||
176 | (*fp)(context, errCode); | |||
177 | uio.fErrCode = errCode; | |||
178 | umtx_initImplPostInit(uio); | |||
179 | } else { | |||
180 | // Someone else already ran the initialization. | |||
181 | if (U_FAILURE(uio.fErrCode)) { | |||
182 | errCode = uio.fErrCode; | |||
183 | } | |||
184 | } | |||
185 | } | |||
186 | ||||
187 | // UMutex should be constexpr-constructible, so that no initialization code | |||
188 | // is run during startup. | |||
189 | // This works on all C++ libraries except MS VS before VS2019. | |||
190 | #if (defined(_CPPLIB_VER) && !defined(_MSVC_STL_VERSION)) || \ | |||
191 | (defined(_MSVC_STL_VERSION) && _MSVC_STL_VERSION < 142) | |||
192 | // (VS std lib older than VS2017) || (VS std lib version < VS2019) | |||
193 | # define UMUTEX_CONSTEXPRconstexpr | |||
194 | #else | |||
195 | # define UMUTEX_CONSTEXPRconstexpr constexpr | |||
196 | #endif | |||
197 | ||||
198 | /** | |||
199 | * UMutex - ICU Mutex class. | |||
200 | * | |||
201 | * This is the preferred Mutex class for use within ICU implementation code. | |||
202 | * It is a thin wrapper over C++ std::mutex, with these additions: | |||
203 | * - Static instances are safe, not triggering static construction or destruction, | |||
204 | * and the associated order of construction or destruction issues. | |||
205 | * - Plumbed into u_cleanup() for destructing the underlying std::mutex, | |||
206 | * which frees any OS level resources they may be holding. | |||
207 | * | |||
208 | * Limitations: | |||
209 | * - Static or global instances only. Cannot be heap allocated. Cannot appear as a | |||
210 | * member of another class. | |||
211 | * - No condition variables or other advanced features. If needed, you will need to use | |||
212 | * std::mutex and std::condition_variable directly. For an example, see unifiedcache.cpp | |||
213 | * | |||
214 | * Typical Usage: | |||
215 | * static UMutex myMutex; | |||
216 | * | |||
217 | * { | |||
218 | * Mutex lock(myMutex); | |||
219 | * ... // Do stuff that is protected by myMutex; | |||
220 | * } // myMutex is released when lock goes out of scope. | |||
221 | */ | |||
222 | ||||
223 | class U_COMMON_API UMutex { | |||
224 | public: | |||
225 | UMUTEX_CONSTEXPRconstexpr UMutex() {} | |||
226 | ~UMutex() = default; | |||
227 | ||||
228 | UMutex(const UMutex &other) = delete; | |||
229 | UMutex &operator =(const UMutex &other) = delete; | |||
230 | void *operator new(size_t) = delete; | |||
231 | ||||
232 | // requirements for C++ BasicLockable, allows UMutex to work with std::lock_guard | |||
233 | void lock() { | |||
234 | std::mutex *m = fMutex.load(std::memory_order_acquire); | |||
235 | if (m == nullptr) { m = getMutex(); } | |||
236 | m->lock(); | |||
237 | } | |||
238 | void unlock() { fMutex.load(std::memory_order_relaxed)->unlock(); } | |||
239 | ||||
240 | static void cleanup(); | |||
241 | ||||
242 | private: | |||
243 | alignas(std::mutex) char fStorage[sizeof(std::mutex)] {}; | |||
244 | std::atomic<std::mutex *> fMutex { nullptr }; | |||
245 | ||||
246 | /** All initialized UMutexes are kept in a linked list, so that they can be found, | |||
247 | * and the underlying std::mutex destructed, by u_cleanup(). | |||
248 | */ | |||
249 | UMutex *fListLink { nullptr }; | |||
250 | static UMutex *gListHead; | |||
251 | ||||
252 | /** Out-of-line function to lazily initialize a UMutex on first use. | |||
253 | * Initial fast check is inline, in lock(). The returned value may never | |||
254 | * be nullptr. | |||
255 | */ | |||
256 | std::mutex *getMutex(); | |||
257 | }; | |||
258 | ||||
259 | ||||
260 | /* Lock a mutex. | |||
261 | * @param mutex The given mutex to be locked. Pass NULL to specify | |||
262 | * the global ICU mutex. Recursive locks are an error | |||
263 | * and may cause a deadlock on some platforms. | |||
264 | */ | |||
265 | U_CAPIextern "C" void U_EXPORT2 umtx_lockumtx_lock_71(UMutex* mutex); | |||
266 | ||||
267 | /* Unlock a mutex. | |||
268 | * @param mutex The given mutex to be unlocked. Pass NULL to specify | |||
269 | * the global ICU mutex. | |||
270 | */ | |||
271 | U_CAPIextern "C" void U_EXPORT2 umtx_unlockumtx_unlock_71 (UMutex* mutex); | |||
272 | ||||
273 | ||||
274 | U_NAMESPACE_END} | |||
275 | ||||
276 | #endif /* UMUTEX_H */ | |||
277 | /*eof*/ |