1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | ****************************************************************************** |
5 | * |
6 | * Copyright (C) 1998-2016, International Business Machines |
7 | * Corporation and others. All Rights Reserved. |
8 | * |
9 | ****************************************************************************** |
10 | * |
11 | * ucnv.c: |
12 | * Implements APIs for the ICU's codeset conversion library; |
13 | * mostly calls through internal functions; |
14 | * created by Bertrand A. Damiba |
15 | * |
16 | * Modification History: |
17 | * |
18 | * Date Name Description |
19 | * 04/04/99 helena Fixed internal header inclusion. |
20 | * 05/09/00 helena Added implementation to handle fallback mappings. |
21 | * 06/20/2000 helena OS/400 port changes; mostly typecast. |
22 | */ |
23 | |
24 | #include "unicode/utypes.h" |
25 | |
26 | #if !UCONFIG_NO_CONVERSION |
27 | |
28 | #include <memory> |
29 | |
30 | #include "unicode/ustring.h" |
31 | #include "unicode/ucnv.h" |
32 | #include "unicode/ucnv_err.h" |
33 | #include "unicode/uset.h" |
34 | #include "unicode/utf.h" |
35 | #include "unicode/utf16.h" |
36 | #include "putilimp.h" |
37 | #include "cmemory.h" |
38 | #include "cstring.h" |
39 | #include "uassert.h" |
40 | #include "utracimp.h" |
41 | #include "ustr_imp.h" |
42 | #include "ucnv_imp.h" |
43 | #include "ucnv_cnv.h" |
44 | #include "ucnv_bld.h" |
45 | |
46 | /* size of intermediate and preflighting buffers in ucnv_convert() */ |
47 | #define CHUNK_SIZE 1024 |
48 | |
49 | typedef struct UAmbiguousConverter { |
50 | const char *name; |
51 | const char16_t variant5c; |
52 | } UAmbiguousConverter; |
53 | |
54 | static const UAmbiguousConverter ambiguousConverters[]={ |
55 | { "ibm-897_P100-1995" , 0xa5 }, |
56 | { "ibm-942_P120-1999" , 0xa5 }, |
57 | { "ibm-943_P130-1999" , 0xa5 }, |
58 | { "ibm-946_P100-1995" , 0xa5 }, |
59 | { "ibm-33722_P120-1999" , 0xa5 }, |
60 | { "ibm-1041_P100-1995" , 0xa5 }, |
61 | /*{ "ibm-54191_P100-2006", 0xa5 },*/ |
62 | /*{ "ibm-62383_P100-2007", 0xa5 },*/ |
63 | /*{ "ibm-891_P100-1995", 0x20a9 },*/ |
64 | { "ibm-944_P100-1995" , 0x20a9 }, |
65 | { "ibm-949_P110-1999" , 0x20a9 }, |
66 | { "ibm-1363_P110-1997" , 0x20a9 }, |
67 | { "ISO_2022,locale=ko,version=0" , 0x20a9 }, |
68 | { "ibm-1088_P100-1995" , 0x20a9 } |
69 | }; |
70 | |
71 | /*Calls through createConverter */ |
72 | U_CAPI UConverter* U_EXPORT2 |
73 | ucnv_open (const char *name, |
74 | UErrorCode * err) |
75 | { |
76 | UConverter *r; |
77 | |
78 | if (err == nullptr || U_FAILURE (*err)) { |
79 | return nullptr; |
80 | } |
81 | |
82 | r = ucnv_createConverter(nullptr, name, err); |
83 | return r; |
84 | } |
85 | |
86 | U_CAPI UConverter* U_EXPORT2 |
87 | ucnv_openPackage (const char *packageName, const char *converterName, UErrorCode * err) |
88 | { |
89 | return ucnv_createConverterFromPackage(packageName, converterName, err); |
90 | } |
91 | |
92 | /*Extracts the char16_t* to a char* and calls through createConverter */ |
93 | U_CAPI UConverter* U_EXPORT2 |
94 | ucnv_openU (const char16_t * name, |
95 | UErrorCode * err) |
96 | { |
97 | char asciiName[UCNV_MAX_CONVERTER_NAME_LENGTH]; |
98 | |
99 | if (err == nullptr || U_FAILURE(*err)) |
100 | return nullptr; |
101 | if (name == nullptr) |
102 | return ucnv_open (nullptr, err); |
103 | if (u_strlen(name) >= UCNV_MAX_CONVERTER_NAME_LENGTH) |
104 | { |
105 | *err = U_ILLEGAL_ARGUMENT_ERROR; |
106 | return nullptr; |
107 | } |
108 | return ucnv_open(u_austrcpy(asciiName, name), err); |
109 | } |
110 | |
111 | /* Copy the string that is represented by the UConverterPlatform enum |
112 | * @param platformString An output buffer |
113 | * @param platform An enum representing a platform |
114 | * @return the length of the copied string. |
115 | */ |
116 | static int32_t |
117 | ucnv_copyPlatformString(char *platformString, UConverterPlatform pltfrm) |
118 | { |
119 | switch (pltfrm) |
120 | { |
121 | case UCNV_IBM: |
122 | uprv_strcpy(platformString, "ibm-" ); |
123 | return 4; |
124 | case UCNV_UNKNOWN: |
125 | break; |
126 | } |
127 | |
128 | /* default to empty string */ |
129 | *platformString = 0; |
130 | return 0; |
131 | } |
132 | |
133 | /*Assumes a $platform-#codepage.$CONVERTER_FILE_EXTENSION scheme and calls |
134 | *through createConverter*/ |
135 | U_CAPI UConverter* U_EXPORT2 |
136 | ucnv_openCCSID (int32_t codepage, |
137 | UConverterPlatform platform, |
138 | UErrorCode * err) |
139 | { |
140 | char myName[UCNV_MAX_CONVERTER_NAME_LENGTH]; |
141 | int32_t myNameLen; |
142 | |
143 | if (err == nullptr || U_FAILURE (*err)) |
144 | return nullptr; |
145 | |
146 | /* ucnv_copyPlatformString could return "ibm-" or "cp" */ |
147 | myNameLen = ucnv_copyPlatformString(myName, platform); |
148 | T_CString_integerToString(myName + myNameLen, codepage, 10); |
149 | |
150 | return ucnv_createConverter(nullptr, myName, err); |
151 | } |
152 | |
153 | /* Creating a temporary stack-based object that can be used in one thread, |
154 | and created from a converter that is shared across threads. |
155 | */ |
156 | |
157 | U_CAPI UConverter* U_EXPORT2 |
158 | ucnv_safeClone(const UConverter* cnv, void *stackBuffer, int32_t *pBufferSize, UErrorCode *status) |
159 | { |
160 | UConverter *localConverter, *allocatedConverter; |
161 | int32_t stackBufferSize; |
162 | int32_t bufferSizeNeeded; |
163 | UErrorCode cbErr; |
164 | UConverterToUnicodeArgs toUArgs = { |
165 | sizeof(UConverterToUnicodeArgs), |
166 | true, |
167 | nullptr, |
168 | nullptr, |
169 | nullptr, |
170 | nullptr, |
171 | nullptr, |
172 | nullptr |
173 | }; |
174 | UConverterFromUnicodeArgs fromUArgs = { |
175 | sizeof(UConverterFromUnicodeArgs), |
176 | true, |
177 | nullptr, |
178 | nullptr, |
179 | nullptr, |
180 | nullptr, |
181 | nullptr, |
182 | nullptr |
183 | }; |
184 | |
185 | UTRACE_ENTRY_OC(UTRACE_UCNV_CLONE); |
186 | |
187 | if (status == nullptr || U_FAILURE(*status)){ |
188 | UTRACE_EXIT_STATUS(status? *status: U_ILLEGAL_ARGUMENT_ERROR); |
189 | return nullptr; |
190 | } |
191 | |
192 | if (cnv == nullptr) { |
193 | *status = U_ILLEGAL_ARGUMENT_ERROR; |
194 | UTRACE_EXIT_STATUS(*status); |
195 | return nullptr; |
196 | } |
197 | |
198 | UTRACE_DATA3(UTRACE_OPEN_CLOSE, "clone converter %s at %p into stackBuffer %p" , |
199 | ucnv_getName(cnv, status), cnv, stackBuffer); |
200 | |
201 | if (cnv->sharedData->impl->safeClone != nullptr) { |
202 | /* call the custom safeClone function for sizing */ |
203 | bufferSizeNeeded = 0; |
204 | cnv->sharedData->impl->safeClone(cnv, nullptr, &bufferSizeNeeded, status); |
205 | if (U_FAILURE(*status)) { |
206 | UTRACE_EXIT_STATUS(*status); |
207 | return nullptr; |
208 | } |
209 | } |
210 | else |
211 | { |
212 | /* inherent sizing */ |
213 | bufferSizeNeeded = sizeof(UConverter); |
214 | } |
215 | |
216 | if (pBufferSize == nullptr) { |
217 | stackBufferSize = 1; |
218 | pBufferSize = &stackBufferSize; |
219 | } else { |
220 | stackBufferSize = *pBufferSize; |
221 | if (stackBufferSize <= 0){ /* 'preflighting' request - set needed size into *pBufferSize */ |
222 | *pBufferSize = bufferSizeNeeded; |
223 | UTRACE_EXIT_VALUE(bufferSizeNeeded); |
224 | return nullptr; |
225 | } |
226 | } |
227 | |
228 | /* Adjust (if necessary) the stackBuffer pointer to be aligned correctly for a UConverter. |
229 | * TODO(Jira ICU-20736) Redo this using std::align() once g++4.9 compatibility is no longer needed. |
230 | */ |
231 | if (stackBuffer) { |
232 | uintptr_t p = reinterpret_cast<uintptr_t>(stackBuffer); |
233 | uintptr_t aligned_p = (p + alignof(UConverter) - 1) & ~(alignof(UConverter) - 1); |
234 | ptrdiff_t pointerAdjustment = aligned_p - p; |
235 | if (bufferSizeNeeded + pointerAdjustment <= stackBufferSize) { |
236 | stackBuffer = reinterpret_cast<void *>(aligned_p); |
237 | stackBufferSize -= static_cast<int32_t>(pointerAdjustment); |
238 | } else { |
239 | /* prevent using the stack buffer but keep the size > 0 so that we do not just preflight */ |
240 | stackBufferSize = 1; |
241 | } |
242 | } |
243 | |
244 | /* Now, see if we must allocate any memory */ |
245 | if (stackBufferSize < bufferSizeNeeded || stackBuffer == nullptr) |
246 | { |
247 | /* allocate one here...*/ |
248 | localConverter = allocatedConverter = (UConverter *) uprv_malloc (bufferSizeNeeded); |
249 | |
250 | if(localConverter == nullptr) { |
251 | *status = U_MEMORY_ALLOCATION_ERROR; |
252 | UTRACE_EXIT_STATUS(*status); |
253 | return nullptr; |
254 | } |
255 | // If pBufferSize was nullptr as the input, pBufferSize is set to &stackBufferSize in this function. |
256 | if (pBufferSize != &stackBufferSize) { |
257 | *status = U_SAFECLONE_ALLOCATED_WARNING; |
258 | } |
259 | |
260 | /* record the fact that memory was allocated */ |
261 | *pBufferSize = bufferSizeNeeded; |
262 | } else { |
263 | /* just use the stack buffer */ |
264 | localConverter = (UConverter*) stackBuffer; |
265 | allocatedConverter = nullptr; |
266 | } |
267 | |
268 | uprv_memset(localConverter, 0, bufferSizeNeeded); |
269 | |
270 | /* Copy initial state */ |
271 | uprv_memcpy(localConverter, cnv, sizeof(UConverter)); |
272 | localConverter->isCopyLocal = localConverter->isExtraLocal = false; |
273 | |
274 | /* copy the substitution string */ |
275 | if (cnv->subChars == (uint8_t *)cnv->subUChars) { |
276 | localConverter->subChars = (uint8_t *)localConverter->subUChars; |
277 | } else { |
278 | localConverter->subChars = (uint8_t *)uprv_malloc(UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR); |
279 | if (localConverter->subChars == nullptr) { |
280 | uprv_free(allocatedConverter); |
281 | UTRACE_EXIT_STATUS(*status); |
282 | return nullptr; |
283 | } |
284 | uprv_memcpy(localConverter->subChars, cnv->subChars, UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR); |
285 | } |
286 | |
287 | /* now either call the safeclone fcn or not */ |
288 | if (cnv->sharedData->impl->safeClone != nullptr) { |
289 | /* call the custom safeClone function */ |
290 | localConverter = cnv->sharedData->impl->safeClone(cnv, localConverter, pBufferSize, status); |
291 | } |
292 | |
293 | if(localConverter==nullptr || U_FAILURE(*status)) { |
294 | if (allocatedConverter != nullptr && allocatedConverter->subChars != (uint8_t *)allocatedConverter->subUChars) { |
295 | uprv_free(allocatedConverter->subChars); |
296 | } |
297 | uprv_free(allocatedConverter); |
298 | UTRACE_EXIT_STATUS(*status); |
299 | return nullptr; |
300 | } |
301 | |
302 | /* increment refcount of shared data if needed */ |
303 | if (cnv->sharedData->isReferenceCounted) { |
304 | ucnv_incrementRefCount(cnv->sharedData); |
305 | } |
306 | |
307 | if(localConverter == (UConverter*)stackBuffer) { |
308 | /* we're using user provided data - set to not destroy */ |
309 | localConverter->isCopyLocal = true; |
310 | } |
311 | |
312 | /* allow callback functions to handle any memory allocation */ |
313 | toUArgs.converter = fromUArgs.converter = localConverter; |
314 | cbErr = U_ZERO_ERROR; |
315 | cnv->fromCharErrorBehaviour(cnv->toUContext, &toUArgs, nullptr, 0, UCNV_CLONE, &cbErr); |
316 | cbErr = U_ZERO_ERROR; |
317 | cnv->fromUCharErrorBehaviour(cnv->fromUContext, &fromUArgs, nullptr, 0, 0, UCNV_CLONE, &cbErr); |
318 | |
319 | UTRACE_EXIT_PTR_STATUS(localConverter, *status); |
320 | return localConverter; |
321 | } |
322 | |
323 | U_CAPI UConverter* U_EXPORT2 |
324 | ucnv_clone(const UConverter* cnv, UErrorCode *status) |
325 | { |
326 | return ucnv_safeClone(cnv, nullptr, nullptr, status); |
327 | } |
328 | |
329 | /*Decreases the reference counter in the shared immutable section of the object |
330 | *and frees the mutable part*/ |
331 | |
332 | U_CAPI void U_EXPORT2 |
333 | ucnv_close (UConverter * converter) |
334 | { |
335 | UErrorCode errorCode = U_ZERO_ERROR; |
336 | |
337 | UTRACE_ENTRY_OC(UTRACE_UCNV_CLOSE); |
338 | |
339 | if (converter == nullptr) |
340 | { |
341 | UTRACE_EXIT(); |
342 | return; |
343 | } |
344 | |
345 | UTRACE_DATA3(UTRACE_OPEN_CLOSE, "close converter %s at %p, isCopyLocal=%b" , |
346 | ucnv_getName(converter, &errorCode), converter, converter->isCopyLocal); |
347 | |
348 | /* In order to speed up the close, only call the callbacks when they have been changed. |
349 | This performance check will only work when the callbacks are set within a shared library |
350 | or from user code that statically links this code. */ |
351 | /* first, notify the callback functions that the converter is closed */ |
352 | if (converter->fromCharErrorBehaviour != UCNV_TO_U_DEFAULT_CALLBACK) { |
353 | UConverterToUnicodeArgs toUArgs = { |
354 | sizeof(UConverterToUnicodeArgs), |
355 | true, |
356 | nullptr, |
357 | nullptr, |
358 | nullptr, |
359 | nullptr, |
360 | nullptr, |
361 | nullptr |
362 | }; |
363 | |
364 | toUArgs.converter = converter; |
365 | errorCode = U_ZERO_ERROR; |
366 | converter->fromCharErrorBehaviour(converter->toUContext, &toUArgs, nullptr, 0, UCNV_CLOSE, &errorCode); |
367 | } |
368 | if (converter->fromUCharErrorBehaviour != UCNV_FROM_U_DEFAULT_CALLBACK) { |
369 | UConverterFromUnicodeArgs fromUArgs = { |
370 | sizeof(UConverterFromUnicodeArgs), |
371 | true, |
372 | nullptr, |
373 | nullptr, |
374 | nullptr, |
375 | nullptr, |
376 | nullptr, |
377 | nullptr |
378 | }; |
379 | fromUArgs.converter = converter; |
380 | errorCode = U_ZERO_ERROR; |
381 | converter->fromUCharErrorBehaviour(converter->fromUContext, &fromUArgs, nullptr, 0, 0, UCNV_CLOSE, &errorCode); |
382 | } |
383 | |
384 | if (converter->sharedData->impl->close != nullptr) { |
385 | converter->sharedData->impl->close(converter); |
386 | } |
387 | |
388 | if (converter->subChars != (uint8_t *)converter->subUChars) { |
389 | uprv_free(converter->subChars); |
390 | } |
391 | |
392 | if (converter->sharedData->isReferenceCounted) { |
393 | ucnv_unloadSharedDataIfReady(converter->sharedData); |
394 | } |
395 | |
396 | if(!converter->isCopyLocal){ |
397 | uprv_free(converter); |
398 | } |
399 | |
400 | UTRACE_EXIT(); |
401 | } |
402 | |
403 | /*returns a single Name from the list, will return nullptr if out of bounds |
404 | */ |
405 | U_CAPI const char* U_EXPORT2 |
406 | ucnv_getAvailableName (int32_t n) |
407 | { |
408 | if (0 <= n && n <= 0xffff) { |
409 | UErrorCode err = U_ZERO_ERROR; |
410 | const char *name = ucnv_bld_getAvailableConverter((uint16_t)n, &err); |
411 | if (U_SUCCESS(err)) { |
412 | return name; |
413 | } |
414 | } |
415 | return nullptr; |
416 | } |
417 | |
418 | U_CAPI int32_t U_EXPORT2 |
419 | ucnv_countAvailable () |
420 | { |
421 | UErrorCode err = U_ZERO_ERROR; |
422 | return ucnv_bld_countAvailableConverters(&err); |
423 | } |
424 | |
425 | U_CAPI void U_EXPORT2 |
426 | ucnv_getSubstChars (const UConverter * converter, |
427 | char *mySubChar, |
428 | int8_t * len, |
429 | UErrorCode * err) |
430 | { |
431 | if (U_FAILURE (*err)) |
432 | return; |
433 | |
434 | if (converter->subCharLen <= 0) { |
435 | /* Unicode string or empty string from ucnv_setSubstString(). */ |
436 | *len = 0; |
437 | return; |
438 | } |
439 | |
440 | if (*len < converter->subCharLen) /*not enough space in subChars */ |
441 | { |
442 | *err = U_INDEX_OUTOFBOUNDS_ERROR; |
443 | return; |
444 | } |
445 | |
446 | uprv_memcpy (mySubChar, converter->subChars, converter->subCharLen); /*fills in the subchars */ |
447 | *len = converter->subCharLen; /*store # of bytes copied to buffer */ |
448 | } |
449 | |
450 | U_CAPI void U_EXPORT2 |
451 | ucnv_setSubstChars (UConverter * converter, |
452 | const char *mySubChar, |
453 | int8_t len, |
454 | UErrorCode * err) |
455 | { |
456 | if (U_FAILURE (*err)) |
457 | return; |
458 | |
459 | /*Makes sure that the subChar is within the codepages char length boundaries */ |
460 | if ((len > converter->sharedData->staticData->maxBytesPerChar) |
461 | || (len < converter->sharedData->staticData->minBytesPerChar)) |
462 | { |
463 | *err = U_ILLEGAL_ARGUMENT_ERROR; |
464 | return; |
465 | } |
466 | |
467 | uprv_memcpy (converter->subChars, mySubChar, len); /*copies the subchars */ |
468 | converter->subCharLen = len; /*sets the new len */ |
469 | |
470 | /* |
471 | * There is currently (2001Feb) no separate API to set/get subChar1. |
472 | * In order to always have subChar written after it is explicitly set, |
473 | * we set subChar1 to 0. |
474 | */ |
475 | converter->subChar1 = 0; |
476 | |
477 | return; |
478 | } |
479 | |
480 | U_CAPI void U_EXPORT2 |
481 | ucnv_setSubstString(UConverter *cnv, |
482 | const char16_t *s, |
483 | int32_t length, |
484 | UErrorCode *err) { |
485 | alignas(UConverter) char cloneBuffer[U_CNV_SAFECLONE_BUFFERSIZE]; |
486 | char chars[UCNV_ERROR_BUFFER_LENGTH]; |
487 | |
488 | UConverter *clone; |
489 | uint8_t *subChars; |
490 | int32_t cloneSize, length8; |
491 | |
492 | /* Let the following functions check all arguments. */ |
493 | cloneSize = sizeof(cloneBuffer); |
494 | clone = ucnv_safeClone(cnv, cloneBuffer, &cloneSize, err); |
495 | ucnv_setFromUCallBack(clone, UCNV_FROM_U_CALLBACK_STOP, nullptr, nullptr, nullptr, err); |
496 | length8 = ucnv_fromUChars(clone, chars, (int32_t)sizeof(chars), s, length, err); |
497 | ucnv_close(clone); |
498 | if (U_FAILURE(*err)) { |
499 | return; |
500 | } |
501 | |
502 | if (cnv->sharedData->impl->writeSub == nullptr |
503 | #if !UCONFIG_NO_LEGACY_CONVERSION |
504 | || (cnv->sharedData->staticData->conversionType == UCNV_MBCS && |
505 | ucnv_MBCSGetType(cnv) != UCNV_EBCDIC_STATEFUL) |
506 | #endif |
507 | ) { |
508 | /* The converter is not stateful. Store the charset bytes as a fixed string. */ |
509 | subChars = (uint8_t *)chars; |
510 | } else { |
511 | /* |
512 | * The converter has a non-default writeSub() function, indicating |
513 | * that it is stateful. |
514 | * Store the Unicode string for on-the-fly conversion for correct |
515 | * state handling. |
516 | */ |
517 | if (length > UCNV_ERROR_BUFFER_LENGTH) { |
518 | /* |
519 | * Should not occur. The converter should output at least one byte |
520 | * per char16_t, which means that ucnv_fromUChars() should catch all |
521 | * overflows. |
522 | */ |
523 | *err = U_BUFFER_OVERFLOW_ERROR; |
524 | return; |
525 | } |
526 | subChars = (uint8_t *)s; |
527 | if (length < 0) { |
528 | length = u_strlen(s); |
529 | } |
530 | length8 = length * U_SIZEOF_UCHAR; |
531 | } |
532 | |
533 | /* |
534 | * For storing the substitution string, select either the small buffer inside |
535 | * UConverter or allocate a subChars buffer. |
536 | */ |
537 | if (length8 > UCNV_MAX_SUBCHAR_LEN) { |
538 | /* Use a separate buffer for the string. Outside UConverter to not make it too large. */ |
539 | if (cnv->subChars == (uint8_t *)cnv->subUChars) { |
540 | /* Allocate a new buffer for the string. */ |
541 | cnv->subChars = (uint8_t *)uprv_malloc(UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR); |
542 | if (cnv->subChars == nullptr) { |
543 | cnv->subChars = (uint8_t *)cnv->subUChars; |
544 | *err = U_MEMORY_ALLOCATION_ERROR; |
545 | return; |
546 | } |
547 | uprv_memset(cnv->subChars, 0, UCNV_ERROR_BUFFER_LENGTH * U_SIZEOF_UCHAR); |
548 | } |
549 | } |
550 | |
551 | /* Copy the substitution string into the UConverter or its subChars buffer. */ |
552 | if (length8 == 0) { |
553 | cnv->subCharLen = 0; |
554 | } else { |
555 | uprv_memcpy(cnv->subChars, subChars, length8); |
556 | if (subChars == (uint8_t *)chars) { |
557 | cnv->subCharLen = (int8_t)length8; |
558 | } else /* subChars == s */ { |
559 | cnv->subCharLen = (int8_t)-length; |
560 | } |
561 | } |
562 | |
563 | /* See comment in ucnv_setSubstChars(). */ |
564 | cnv->subChar1 = 0; |
565 | } |
566 | |
567 | /*resets the internal states of a converter |
568 | *goal : have the same behaviour than a freshly created converter |
569 | */ |
570 | static void _reset(UConverter *converter, UConverterResetChoice choice, |
571 | UBool callCallback) { |
572 | if(converter == nullptr) { |
573 | return; |
574 | } |
575 | |
576 | if(callCallback) { |
577 | /* first, notify the callback functions that the converter is reset */ |
578 | UErrorCode errorCode; |
579 | |
580 | if(choice<=UCNV_RESET_TO_UNICODE && converter->fromCharErrorBehaviour != UCNV_TO_U_DEFAULT_CALLBACK) { |
581 | UConverterToUnicodeArgs toUArgs = { |
582 | sizeof(UConverterToUnicodeArgs), |
583 | true, |
584 | nullptr, |
585 | nullptr, |
586 | nullptr, |
587 | nullptr, |
588 | nullptr, |
589 | nullptr |
590 | }; |
591 | toUArgs.converter = converter; |
592 | errorCode = U_ZERO_ERROR; |
593 | converter->fromCharErrorBehaviour(converter->toUContext, &toUArgs, nullptr, 0, UCNV_RESET, &errorCode); |
594 | } |
595 | if(choice!=UCNV_RESET_TO_UNICODE && converter->fromUCharErrorBehaviour != UCNV_FROM_U_DEFAULT_CALLBACK) { |
596 | UConverterFromUnicodeArgs fromUArgs = { |
597 | sizeof(UConverterFromUnicodeArgs), |
598 | true, |
599 | nullptr, |
600 | nullptr, |
601 | nullptr, |
602 | nullptr, |
603 | nullptr, |
604 | nullptr |
605 | }; |
606 | fromUArgs.converter = converter; |
607 | errorCode = U_ZERO_ERROR; |
608 | converter->fromUCharErrorBehaviour(converter->fromUContext, &fromUArgs, nullptr, 0, 0, UCNV_RESET, &errorCode); |
609 | } |
610 | } |
611 | |
612 | /* now reset the converter itself */ |
613 | if(choice<=UCNV_RESET_TO_UNICODE) { |
614 | converter->toUnicodeStatus = converter->sharedData->toUnicodeStatus; |
615 | converter->mode = 0; |
616 | converter->toULength = 0; |
617 | converter->invalidCharLength = converter->UCharErrorBufferLength = 0; |
618 | converter->preToULength = 0; |
619 | } |
620 | if(choice!=UCNV_RESET_TO_UNICODE) { |
621 | converter->fromUnicodeStatus = 0; |
622 | converter->fromUChar32 = 0; |
623 | converter->invalidUCharLength = converter->charErrorBufferLength = 0; |
624 | converter->preFromUFirstCP = U_SENTINEL; |
625 | converter->preFromULength = 0; |
626 | } |
627 | |
628 | if (converter->sharedData->impl->reset != nullptr) { |
629 | /* call the custom reset function */ |
630 | converter->sharedData->impl->reset(converter, choice); |
631 | } |
632 | } |
633 | |
634 | U_CAPI void U_EXPORT2 |
635 | ucnv_reset(UConverter *converter) |
636 | { |
637 | _reset(converter, UCNV_RESET_BOTH, true); |
638 | } |
639 | |
640 | U_CAPI void U_EXPORT2 |
641 | ucnv_resetToUnicode(UConverter *converter) |
642 | { |
643 | _reset(converter, UCNV_RESET_TO_UNICODE, true); |
644 | } |
645 | |
646 | U_CAPI void U_EXPORT2 |
647 | ucnv_resetFromUnicode(UConverter *converter) |
648 | { |
649 | _reset(converter, UCNV_RESET_FROM_UNICODE, true); |
650 | } |
651 | |
652 | U_CAPI int8_t U_EXPORT2 |
653 | ucnv_getMaxCharSize (const UConverter * converter) |
654 | { |
655 | return converter->maxBytesPerUChar; |
656 | } |
657 | |
658 | |
659 | U_CAPI int8_t U_EXPORT2 |
660 | ucnv_getMinCharSize (const UConverter * converter) |
661 | { |
662 | return converter->sharedData->staticData->minBytesPerChar; |
663 | } |
664 | |
665 | U_CAPI const char* U_EXPORT2 |
666 | ucnv_getName (const UConverter * converter, UErrorCode * err) |
667 | |
668 | { |
669 | if (U_FAILURE (*err)) |
670 | return nullptr; |
671 | if(converter->sharedData->impl->getName){ |
672 | const char* temp= converter->sharedData->impl->getName(converter); |
673 | if(temp) |
674 | return temp; |
675 | } |
676 | return converter->sharedData->staticData->name; |
677 | } |
678 | |
679 | U_CAPI int32_t U_EXPORT2 |
680 | ucnv_getCCSID(const UConverter * converter, |
681 | UErrorCode * err) |
682 | { |
683 | int32_t ccsid; |
684 | if (U_FAILURE (*err)) |
685 | return -1; |
686 | |
687 | ccsid = converter->sharedData->staticData->codepage; |
688 | if (ccsid == 0) { |
689 | /* Rare case. This is for cases like gb18030, |
690 | which doesn't have an IBM canonical name, but does have an IBM alias. */ |
691 | const char *standardName = ucnv_getStandardName(ucnv_getName(converter, err), "IBM" , err); |
692 | if (U_SUCCESS(*err) && standardName) { |
693 | const char *ccsidStr = uprv_strchr(standardName, '-'); |
694 | if (ccsidStr) { |
695 | ccsid = (int32_t)atol(ccsidStr+1); /* +1 to skip '-' */ |
696 | } |
697 | } |
698 | } |
699 | return ccsid; |
700 | } |
701 | |
702 | |
703 | U_CAPI UConverterPlatform U_EXPORT2 |
704 | ucnv_getPlatform (const UConverter * converter, |
705 | UErrorCode * err) |
706 | { |
707 | if (U_FAILURE (*err)) |
708 | return UCNV_UNKNOWN; |
709 | |
710 | return (UConverterPlatform)converter->sharedData->staticData->platform; |
711 | } |
712 | |
713 | U_CAPI void U_EXPORT2 |
714 | ucnv_getToUCallBack (const UConverter * converter, |
715 | UConverterToUCallback *action, |
716 | const void **context) |
717 | { |
718 | *action = converter->fromCharErrorBehaviour; |
719 | *context = converter->toUContext; |
720 | } |
721 | |
722 | U_CAPI void U_EXPORT2 |
723 | ucnv_getFromUCallBack (const UConverter * converter, |
724 | UConverterFromUCallback *action, |
725 | const void **context) |
726 | { |
727 | *action = converter->fromUCharErrorBehaviour; |
728 | *context = converter->fromUContext; |
729 | } |
730 | |
731 | U_CAPI void U_EXPORT2 |
732 | ucnv_setToUCallBack (UConverter * converter, |
733 | UConverterToUCallback newAction, |
734 | const void* newContext, |
735 | UConverterToUCallback *oldAction, |
736 | const void** oldContext, |
737 | UErrorCode * err) |
738 | { |
739 | if (U_FAILURE (*err)) |
740 | return; |
741 | if (oldAction) *oldAction = converter->fromCharErrorBehaviour; |
742 | converter->fromCharErrorBehaviour = newAction; |
743 | if (oldContext) *oldContext = converter->toUContext; |
744 | converter->toUContext = newContext; |
745 | } |
746 | |
747 | U_CAPI void U_EXPORT2 |
748 | ucnv_setFromUCallBack (UConverter * converter, |
749 | UConverterFromUCallback newAction, |
750 | const void* newContext, |
751 | UConverterFromUCallback *oldAction, |
752 | const void** oldContext, |
753 | UErrorCode * err) |
754 | { |
755 | if (U_FAILURE (*err)) |
756 | return; |
757 | if (oldAction) *oldAction = converter->fromUCharErrorBehaviour; |
758 | converter->fromUCharErrorBehaviour = newAction; |
759 | if (oldContext) *oldContext = converter->fromUContext; |
760 | converter->fromUContext = newContext; |
761 | } |
762 | |
763 | static void |
764 | _updateOffsets(int32_t *offsets, int32_t length, |
765 | int32_t sourceIndex, int32_t errorInputLength) { |
766 | int32_t *limit; |
767 | int32_t delta, offset; |
768 | |
769 | if(sourceIndex>=0) { |
770 | /* |
771 | * adjust each offset by adding the previous sourceIndex |
772 | * minus the length of the input sequence that caused an |
773 | * error, if any |
774 | */ |
775 | delta=sourceIndex-errorInputLength; |
776 | } else { |
777 | /* |
778 | * set each offset to -1 because this conversion function |
779 | * does not handle offsets |
780 | */ |
781 | delta=-1; |
782 | } |
783 | |
784 | limit=offsets+length; |
785 | if(delta==0) { |
786 | /* most common case, nothing to do */ |
787 | } else if(delta>0) { |
788 | /* add the delta to each offset (but not if the offset is <0) */ |
789 | while(offsets<limit) { |
790 | offset=*offsets; |
791 | if(offset>=0) { |
792 | *offsets=offset+delta; |
793 | } |
794 | ++offsets; |
795 | } |
796 | } else /* delta<0 */ { |
797 | /* |
798 | * set each offset to -1 because this conversion function |
799 | * does not handle offsets |
800 | * or the error input sequence started in a previous buffer |
801 | */ |
802 | while(offsets<limit) { |
803 | *offsets++=-1; |
804 | } |
805 | } |
806 | } |
807 | |
808 | /* ucnv_fromUnicode --------------------------------------------------------- */ |
809 | |
810 | /* |
811 | * Implementation note for m:n conversions |
812 | * |
813 | * While collecting source units to find the longest match for m:n conversion, |
814 | * some source units may need to be stored for a partial match. |
815 | * When a second buffer does not yield a match on all of the previously stored |
816 | * source units, then they must be "replayed", i.e., fed back into the converter. |
817 | * |
818 | * The code relies on the fact that replaying will not nest - |
819 | * converting a replay buffer will not result in a replay. |
820 | * This is because a replay is necessary only after the _continuation_ of a |
821 | * partial match failed, but a replay buffer is converted as a whole. |
822 | * It may result in some of its units being stored again for a partial match, |
823 | * but there will not be a continuation _during_ the replay which could fail. |
824 | * |
825 | * It is conceivable that a callback function could call the converter |
826 | * recursively in a way that causes another replay to be stored, but that |
827 | * would be an error in the callback function. |
828 | * Such violations will cause assertion failures in a debug build, |
829 | * and wrong output, but they will not cause a crash. |
830 | */ |
831 | |
832 | static void |
833 | _fromUnicodeWithCallback(UConverterFromUnicodeArgs *pArgs, UErrorCode *err) { |
834 | UConverterFromUnicode fromUnicode; |
835 | UConverter *cnv; |
836 | const char16_t *s; |
837 | char *t; |
838 | int32_t *offsets; |
839 | int32_t sourceIndex; |
840 | int32_t errorInputLength; |
841 | UBool converterSawEndOfInput, calledCallback; |
842 | |
843 | /* variables for m:n conversion */ |
844 | char16_t replay[UCNV_EXT_MAX_UCHARS]; |
845 | const char16_t *realSource, *realSourceLimit; |
846 | int32_t realSourceIndex; |
847 | UBool realFlush; |
848 | |
849 | cnv=pArgs->converter; |
850 | s=pArgs->source; |
851 | t=pArgs->target; |
852 | offsets=pArgs->offsets; |
853 | |
854 | /* get the converter implementation function */ |
855 | sourceIndex=0; |
856 | if(offsets==nullptr) { |
857 | fromUnicode=cnv->sharedData->impl->fromUnicode; |
858 | } else { |
859 | fromUnicode=cnv->sharedData->impl->fromUnicodeWithOffsets; |
860 | if(fromUnicode==nullptr) { |
861 | /* there is no WithOffsets implementation */ |
862 | fromUnicode=cnv->sharedData->impl->fromUnicode; |
863 | /* we will write -1 for each offset */ |
864 | sourceIndex=-1; |
865 | } |
866 | } |
867 | |
868 | if(cnv->preFromULength>=0) { |
869 | /* normal mode */ |
870 | realSource=nullptr; |
871 | |
872 | /* avoid compiler warnings - not otherwise necessary, and the values do not matter */ |
873 | realSourceLimit=nullptr; |
874 | realFlush=false; |
875 | realSourceIndex=0; |
876 | } else { |
877 | /* |
878 | * Previous m:n conversion stored source units from a partial match |
879 | * and failed to consume all of them. |
880 | * We need to "replay" them from a temporary buffer and convert them first. |
881 | */ |
882 | realSource=pArgs->source; |
883 | realSourceLimit=pArgs->sourceLimit; |
884 | realFlush=pArgs->flush; |
885 | realSourceIndex=sourceIndex; |
886 | |
887 | uprv_memcpy(replay, cnv->preFromU, -cnv->preFromULength*U_SIZEOF_UCHAR); |
888 | pArgs->source=replay; |
889 | pArgs->sourceLimit=replay-cnv->preFromULength; |
890 | pArgs->flush=false; |
891 | sourceIndex=-1; |
892 | |
893 | cnv->preFromULength=0; |
894 | } |
895 | |
896 | /* |
897 | * loop for conversion and error handling |
898 | * |
899 | * loop { |
900 | * convert |
901 | * loop { |
902 | * update offsets |
903 | * handle end of input |
904 | * handle errors/call callback |
905 | * } |
906 | * } |
907 | */ |
908 | for(;;) { |
909 | if(U_SUCCESS(*err)) { |
910 | /* convert */ |
911 | fromUnicode(pArgs, err); |
912 | |
913 | /* |
914 | * set a flag for whether the converter |
915 | * successfully processed the end of the input |
916 | * |
917 | * need not check cnv->preFromULength==0 because a replay (<0) will cause |
918 | * s<sourceLimit before converterSawEndOfInput is checked |
919 | */ |
920 | converterSawEndOfInput= |
921 | (UBool)(U_SUCCESS(*err) && |
922 | pArgs->flush && pArgs->source==pArgs->sourceLimit && |
923 | cnv->fromUChar32==0); |
924 | } else { |
925 | /* handle error from ucnv_convertEx() */ |
926 | converterSawEndOfInput=false; |
927 | } |
928 | |
929 | /* no callback called yet for this iteration */ |
930 | calledCallback=false; |
931 | |
932 | /* no sourceIndex adjustment for conversion, only for callback output */ |
933 | errorInputLength=0; |
934 | |
935 | /* |
936 | * loop for offsets and error handling |
937 | * |
938 | * iterates at most 3 times: |
939 | * 1. to clean up after the conversion function |
940 | * 2. after the callback |
941 | * 3. after the callback again if there was truncated input |
942 | */ |
943 | for(;;) { |
944 | /* update offsets if we write any */ |
945 | if(offsets!=nullptr) { |
946 | int32_t length=(int32_t)(pArgs->target-t); |
947 | if(length>0) { |
948 | _updateOffsets(offsets, length, sourceIndex, errorInputLength); |
949 | |
950 | /* |
951 | * if a converter handles offsets and updates the offsets |
952 | * pointer at the end, then pArgs->offset should not change |
953 | * here; |
954 | * however, some converters do not handle offsets at all |
955 | * (sourceIndex<0) or may not update the offsets pointer |
956 | */ |
957 | pArgs->offsets=offsets+=length; |
958 | } |
959 | |
960 | if(sourceIndex>=0) { |
961 | sourceIndex+=(int32_t)(pArgs->source-s); |
962 | } |
963 | } |
964 | |
965 | if(cnv->preFromULength<0) { |
966 | /* |
967 | * switch the source to new replay units (cannot occur while replaying) |
968 | * after offset handling and before end-of-input and callback handling |
969 | */ |
970 | if(realSource==nullptr) { |
971 | realSource=pArgs->source; |
972 | realSourceLimit=pArgs->sourceLimit; |
973 | realFlush=pArgs->flush; |
974 | realSourceIndex=sourceIndex; |
975 | |
976 | uprv_memcpy(replay, cnv->preFromU, -cnv->preFromULength*U_SIZEOF_UCHAR); |
977 | pArgs->source=replay; |
978 | pArgs->sourceLimit=replay-cnv->preFromULength; |
979 | pArgs->flush=false; |
980 | if((sourceIndex+=cnv->preFromULength)<0) { |
981 | sourceIndex=-1; |
982 | } |
983 | |
984 | cnv->preFromULength=0; |
985 | } else { |
986 | /* see implementation note before _fromUnicodeWithCallback() */ |
987 | U_ASSERT(realSource==nullptr); |
988 | *err=U_INTERNAL_PROGRAM_ERROR; |
989 | } |
990 | } |
991 | |
992 | /* update pointers */ |
993 | s=pArgs->source; |
994 | t=pArgs->target; |
995 | |
996 | if(U_SUCCESS(*err)) { |
997 | if(s<pArgs->sourceLimit) { |
998 | /* |
999 | * continue with the conversion loop while there is still input left |
1000 | * (continue converting by breaking out of only the inner loop) |
1001 | */ |
1002 | break; |
1003 | } else if(realSource!=nullptr) { |
1004 | /* switch back from replaying to the real source and continue */ |
1005 | pArgs->source=realSource; |
1006 | pArgs->sourceLimit=realSourceLimit; |
1007 | pArgs->flush=realFlush; |
1008 | sourceIndex=realSourceIndex; |
1009 | |
1010 | realSource=nullptr; |
1011 | break; |
1012 | } else if(pArgs->flush && cnv->fromUChar32!=0) { |
1013 | /* |
1014 | * the entire input stream is consumed |
1015 | * and there is a partial, truncated input sequence left |
1016 | */ |
1017 | |
1018 | /* inject an error and continue with callback handling */ |
1019 | *err=U_TRUNCATED_CHAR_FOUND; |
1020 | calledCallback=false; /* new error condition */ |
1021 | } else { |
1022 | /* input consumed */ |
1023 | if(pArgs->flush) { |
1024 | /* |
1025 | * return to the conversion loop once more if the flush |
1026 | * flag is set and the conversion function has not |
1027 | * successfully processed the end of the input yet |
1028 | * |
1029 | * (continue converting by breaking out of only the inner loop) |
1030 | */ |
1031 | if(!converterSawEndOfInput) { |
1032 | break; |
1033 | } |
1034 | |
1035 | /* reset the converter without calling the callback function */ |
1036 | _reset(cnv, UCNV_RESET_FROM_UNICODE, false); |
1037 | } |
1038 | |
1039 | /* done successfully */ |
1040 | return; |
1041 | } |
1042 | } |
1043 | |
1044 | /* U_FAILURE(*err) */ |
1045 | { |
1046 | UErrorCode e; |
1047 | |
1048 | if( calledCallback || |
1049 | (e=*err)==U_BUFFER_OVERFLOW_ERROR || |
1050 | (e!=U_INVALID_CHAR_FOUND && |
1051 | e!=U_ILLEGAL_CHAR_FOUND && |
1052 | e!=U_TRUNCATED_CHAR_FOUND) |
1053 | ) { |
1054 | /* |
1055 | * the callback did not or cannot resolve the error: |
1056 | * set output pointers and return |
1057 | * |
1058 | * the check for buffer overflow is redundant but it is |
1059 | * a high-runner case and hopefully documents the intent |
1060 | * well |
1061 | * |
1062 | * if we were replaying, then the replay buffer must be |
1063 | * copied back into the UConverter |
1064 | * and the real arguments must be restored |
1065 | */ |
1066 | if(realSource!=nullptr) { |
1067 | int32_t length; |
1068 | |
1069 | U_ASSERT(cnv->preFromULength==0); |
1070 | |
1071 | length=(int32_t)(pArgs->sourceLimit-pArgs->source); |
1072 | if(length>0) { |
1073 | u_memcpy(cnv->preFromU, pArgs->source, length); |
1074 | cnv->preFromULength=(int8_t)-length; |
1075 | } |
1076 | |
1077 | pArgs->source=realSource; |
1078 | pArgs->sourceLimit=realSourceLimit; |
1079 | pArgs->flush=realFlush; |
1080 | } |
1081 | |
1082 | return; |
1083 | } |
1084 | } |
1085 | |
1086 | /* callback handling */ |
1087 | { |
1088 | UChar32 codePoint; |
1089 | |
1090 | /* get and write the code point */ |
1091 | codePoint=cnv->fromUChar32; |
1092 | errorInputLength=0; |
1093 | U16_APPEND_UNSAFE(cnv->invalidUCharBuffer, errorInputLength, codePoint); |
1094 | cnv->invalidUCharLength=(int8_t)errorInputLength; |
1095 | |
1096 | /* set the converter state to deal with the next character */ |
1097 | cnv->fromUChar32=0; |
1098 | |
1099 | /* call the callback function */ |
1100 | cnv->fromUCharErrorBehaviour(cnv->fromUContext, pArgs, |
1101 | cnv->invalidUCharBuffer, errorInputLength, codePoint, |
1102 | *err==U_INVALID_CHAR_FOUND ? UCNV_UNASSIGNED : UCNV_ILLEGAL, |
1103 | err); |
1104 | } |
1105 | |
1106 | /* |
1107 | * loop back to the offset handling |
1108 | * |
1109 | * this flag will indicate after offset handling |
1110 | * that a callback was called; |
1111 | * if the callback did not resolve the error, then we return |
1112 | */ |
1113 | calledCallback=true; |
1114 | } |
1115 | } |
1116 | } |
1117 | |
1118 | /* |
1119 | * Output the fromUnicode overflow buffer. |
1120 | * Call this function if(cnv->charErrorBufferLength>0). |
1121 | * @return true if overflow |
1122 | */ |
1123 | static UBool |
1124 | ucnv_outputOverflowFromUnicode(UConverter *cnv, |
1125 | char **target, const char *targetLimit, |
1126 | int32_t **pOffsets, |
1127 | UErrorCode *err) { |
1128 | int32_t *offsets; |
1129 | char *overflow, *t; |
1130 | int32_t i, length; |
1131 | |
1132 | t=*target; |
1133 | if(pOffsets!=nullptr) { |
1134 | offsets=*pOffsets; |
1135 | } else { |
1136 | offsets=nullptr; |
1137 | } |
1138 | |
1139 | overflow=(char *)cnv->charErrorBuffer; |
1140 | length=cnv->charErrorBufferLength; |
1141 | i=0; |
1142 | while(i<length) { |
1143 | if(t==targetLimit) { |
1144 | /* the overflow buffer contains too much, keep the rest */ |
1145 | int32_t j=0; |
1146 | |
1147 | do { |
1148 | overflow[j++]=overflow[i++]; |
1149 | } while(i<length); |
1150 | |
1151 | cnv->charErrorBufferLength=(int8_t)j; |
1152 | *target=t; |
1153 | if(offsets!=nullptr) { |
1154 | *pOffsets=offsets; |
1155 | } |
1156 | *err=U_BUFFER_OVERFLOW_ERROR; |
1157 | return true; |
1158 | } |
1159 | |
1160 | /* copy the overflow contents to the target */ |
1161 | *t++=overflow[i++]; |
1162 | if(offsets!=nullptr) { |
1163 | *offsets++=-1; /* no source index available for old output */ |
1164 | } |
1165 | } |
1166 | |
1167 | /* the overflow buffer is completely copied to the target */ |
1168 | cnv->charErrorBufferLength=0; |
1169 | *target=t; |
1170 | if(offsets!=nullptr) { |
1171 | *pOffsets=offsets; |
1172 | } |
1173 | return false; |
1174 | } |
1175 | |
1176 | U_CAPI void U_EXPORT2 |
1177 | ucnv_fromUnicode(UConverter *cnv, |
1178 | char **target, const char *targetLimit, |
1179 | const char16_t **source, const char16_t *sourceLimit, |
1180 | int32_t *offsets, |
1181 | UBool flush, |
1182 | UErrorCode *err) { |
1183 | UConverterFromUnicodeArgs args; |
1184 | const char16_t *s; |
1185 | char *t; |
1186 | |
1187 | /* check parameters */ |
1188 | if(err==nullptr || U_FAILURE(*err)) { |
1189 | return; |
1190 | } |
1191 | |
1192 | if(cnv==nullptr || target==nullptr || source==nullptr) { |
1193 | *err=U_ILLEGAL_ARGUMENT_ERROR; |
1194 | return; |
1195 | } |
1196 | |
1197 | s=*source; |
1198 | t=*target; |
1199 | |
1200 | if ((const void *)U_MAX_PTR(sourceLimit) == (const void *)sourceLimit) { |
1201 | /* |
1202 | Prevent code from going into an infinite loop in case we do hit this |
1203 | limit. The limit pointer is expected to be on a char16_t * boundary. |
1204 | This also prevents the next argument check from failing. |
1205 | */ |
1206 | sourceLimit = (const char16_t *)(((const char *)sourceLimit) - 1); |
1207 | } |
1208 | |
1209 | /* |
1210 | * All these conditions should never happen. |
1211 | * |
1212 | * 1) Make sure that the limits are >= to the address source or target |
1213 | * |
1214 | * 2) Make sure that the buffer sizes do not exceed the number range for |
1215 | * int32_t because some functions use the size (in units or bytes) |
1216 | * rather than comparing pointers, and because offsets are int32_t values. |
1217 | * |
1218 | * size_t is guaranteed to be unsigned and large enough for the job. |
1219 | * |
1220 | * Return with an error instead of adjusting the limits because we would |
1221 | * not be able to maintain the semantics that either the source must be |
1222 | * consumed or the target filled (unless an error occurs). |
1223 | * An adjustment would be targetLimit=t+0x7fffffff; for example. |
1224 | * |
1225 | * 3) Make sure that the user didn't incorrectly cast a char16_t * pointer |
1226 | * to a char * pointer and provide an incomplete char16_t code unit. |
1227 | */ |
1228 | if (sourceLimit<s || targetLimit<t || |
1229 | ((size_t)(sourceLimit-s)>(size_t)0x3fffffff && sourceLimit>s) || |
1230 | ((size_t)(targetLimit-t)>(size_t)0x7fffffff && targetLimit>t) || |
1231 | (((const char *)sourceLimit-(const char *)s) & 1) != 0) |
1232 | { |
1233 | *err=U_ILLEGAL_ARGUMENT_ERROR; |
1234 | return; |
1235 | } |
1236 | |
1237 | /* output the target overflow buffer */ |
1238 | if( cnv->charErrorBufferLength>0 && |
1239 | ucnv_outputOverflowFromUnicode(cnv, target, targetLimit, &offsets, err) |
1240 | ) { |
1241 | /* U_BUFFER_OVERFLOW_ERROR */ |
1242 | return; |
1243 | } |
1244 | /* *target may have moved, therefore stop using t */ |
1245 | |
1246 | if(!flush && s==sourceLimit && cnv->preFromULength>=0) { |
1247 | /* the overflow buffer is emptied and there is no new input: we are done */ |
1248 | return; |
1249 | } |
1250 | |
1251 | /* |
1252 | * Do not simply return with a buffer overflow error if |
1253 | * !flush && t==targetLimit |
1254 | * because it is possible that the source will not generate any output. |
1255 | * For example, the skip callback may be called; |
1256 | * it does not output anything. |
1257 | */ |
1258 | |
1259 | /* prepare the converter arguments */ |
1260 | args.converter=cnv; |
1261 | args.flush=flush; |
1262 | args.offsets=offsets; |
1263 | args.source=s; |
1264 | args.sourceLimit=sourceLimit; |
1265 | args.target=*target; |
1266 | args.targetLimit=targetLimit; |
1267 | args.size=sizeof(args); |
1268 | |
1269 | _fromUnicodeWithCallback(&args, err); |
1270 | |
1271 | *source=args.source; |
1272 | *target=args.target; |
1273 | } |
1274 | |
1275 | /* ucnv_toUnicode() --------------------------------------------------------- */ |
1276 | |
1277 | static void |
1278 | _toUnicodeWithCallback(UConverterToUnicodeArgs *pArgs, UErrorCode *err) { |
1279 | UConverterToUnicode toUnicode; |
1280 | UConverter *cnv; |
1281 | const char *s; |
1282 | char16_t *t; |
1283 | int32_t *offsets; |
1284 | int32_t sourceIndex; |
1285 | int32_t errorInputLength; |
1286 | UBool converterSawEndOfInput, calledCallback; |
1287 | |
1288 | /* variables for m:n conversion */ |
1289 | char replay[UCNV_EXT_MAX_BYTES]; |
1290 | const char *realSource, *realSourceLimit; |
1291 | int32_t realSourceIndex; |
1292 | UBool realFlush; |
1293 | |
1294 | cnv=pArgs->converter; |
1295 | s=pArgs->source; |
1296 | t=pArgs->target; |
1297 | offsets=pArgs->offsets; |
1298 | |
1299 | /* get the converter implementation function */ |
1300 | sourceIndex=0; |
1301 | if(offsets==nullptr) { |
1302 | toUnicode=cnv->sharedData->impl->toUnicode; |
1303 | } else { |
1304 | toUnicode=cnv->sharedData->impl->toUnicodeWithOffsets; |
1305 | if(toUnicode==nullptr) { |
1306 | /* there is no WithOffsets implementation */ |
1307 | toUnicode=cnv->sharedData->impl->toUnicode; |
1308 | /* we will write -1 for each offset */ |
1309 | sourceIndex=-1; |
1310 | } |
1311 | } |
1312 | |
1313 | if(cnv->preToULength>=0) { |
1314 | /* normal mode */ |
1315 | realSource=nullptr; |
1316 | |
1317 | /* avoid compiler warnings - not otherwise necessary, and the values do not matter */ |
1318 | realSourceLimit=nullptr; |
1319 | realFlush=false; |
1320 | realSourceIndex=0; |
1321 | } else { |
1322 | /* |
1323 | * Previous m:n conversion stored source units from a partial match |
1324 | * and failed to consume all of them. |
1325 | * We need to "replay" them from a temporary buffer and convert them first. |
1326 | */ |
1327 | realSource=pArgs->source; |
1328 | realSourceLimit=pArgs->sourceLimit; |
1329 | realFlush=pArgs->flush; |
1330 | realSourceIndex=sourceIndex; |
1331 | |
1332 | uprv_memcpy(replay, cnv->preToU, -cnv->preToULength); |
1333 | pArgs->source=replay; |
1334 | pArgs->sourceLimit=replay-cnv->preToULength; |
1335 | pArgs->flush=false; |
1336 | sourceIndex=-1; |
1337 | |
1338 | cnv->preToULength=0; |
1339 | } |
1340 | |
1341 | /* |
1342 | * loop for conversion and error handling |
1343 | * |
1344 | * loop { |
1345 | * convert |
1346 | * loop { |
1347 | * update offsets |
1348 | * handle end of input |
1349 | * handle errors/call callback |
1350 | * } |
1351 | * } |
1352 | */ |
1353 | for(;;) { |
1354 | if(U_SUCCESS(*err)) { |
1355 | /* convert */ |
1356 | toUnicode(pArgs, err); |
1357 | |
1358 | /* |
1359 | * set a flag for whether the converter |
1360 | * successfully processed the end of the input |
1361 | * |
1362 | * need not check cnv->preToULength==0 because a replay (<0) will cause |
1363 | * s<sourceLimit before converterSawEndOfInput is checked |
1364 | */ |
1365 | converterSawEndOfInput= |
1366 | (UBool)(U_SUCCESS(*err) && |
1367 | pArgs->flush && pArgs->source==pArgs->sourceLimit && |
1368 | cnv->toULength==0); |
1369 | } else { |
1370 | /* handle error from getNextUChar() or ucnv_convertEx() */ |
1371 | converterSawEndOfInput=false; |
1372 | } |
1373 | |
1374 | /* no callback called yet for this iteration */ |
1375 | calledCallback=false; |
1376 | |
1377 | /* no sourceIndex adjustment for conversion, only for callback output */ |
1378 | errorInputLength=0; |
1379 | |
1380 | /* |
1381 | * loop for offsets and error handling |
1382 | * |
1383 | * iterates at most 3 times: |
1384 | * 1. to clean up after the conversion function |
1385 | * 2. after the callback |
1386 | * 3. after the callback again if there was truncated input |
1387 | */ |
1388 | for(;;) { |
1389 | /* update offsets if we write any */ |
1390 | if(offsets!=nullptr) { |
1391 | int32_t length=(int32_t)(pArgs->target-t); |
1392 | if(length>0) { |
1393 | _updateOffsets(offsets, length, sourceIndex, errorInputLength); |
1394 | |
1395 | /* |
1396 | * if a converter handles offsets and updates the offsets |
1397 | * pointer at the end, then pArgs->offset should not change |
1398 | * here; |
1399 | * however, some converters do not handle offsets at all |
1400 | * (sourceIndex<0) or may not update the offsets pointer |
1401 | */ |
1402 | pArgs->offsets=offsets+=length; |
1403 | } |
1404 | |
1405 | if(sourceIndex>=0) { |
1406 | sourceIndex+=(int32_t)(pArgs->source-s); |
1407 | } |
1408 | } |
1409 | |
1410 | if(cnv->preToULength<0) { |
1411 | /* |
1412 | * switch the source to new replay units (cannot occur while replaying) |
1413 | * after offset handling and before end-of-input and callback handling |
1414 | */ |
1415 | if(realSource==nullptr) { |
1416 | realSource=pArgs->source; |
1417 | realSourceLimit=pArgs->sourceLimit; |
1418 | realFlush=pArgs->flush; |
1419 | realSourceIndex=sourceIndex; |
1420 | |
1421 | uprv_memcpy(replay, cnv->preToU, -cnv->preToULength); |
1422 | pArgs->source=replay; |
1423 | pArgs->sourceLimit=replay-cnv->preToULength; |
1424 | pArgs->flush=false; |
1425 | if((sourceIndex+=cnv->preToULength)<0) { |
1426 | sourceIndex=-1; |
1427 | } |
1428 | |
1429 | cnv->preToULength=0; |
1430 | } else { |
1431 | /* see implementation note before _fromUnicodeWithCallback() */ |
1432 | U_ASSERT(realSource==nullptr); |
1433 | *err=U_INTERNAL_PROGRAM_ERROR; |
1434 | } |
1435 | } |
1436 | |
1437 | /* update pointers */ |
1438 | s=pArgs->source; |
1439 | t=pArgs->target; |
1440 | |
1441 | if(U_SUCCESS(*err)) { |
1442 | if(s<pArgs->sourceLimit) { |
1443 | /* |
1444 | * continue with the conversion loop while there is still input left |
1445 | * (continue converting by breaking out of only the inner loop) |
1446 | */ |
1447 | break; |
1448 | } else if(realSource!=nullptr) { |
1449 | /* switch back from replaying to the real source and continue */ |
1450 | pArgs->source=realSource; |
1451 | pArgs->sourceLimit=realSourceLimit; |
1452 | pArgs->flush=realFlush; |
1453 | sourceIndex=realSourceIndex; |
1454 | |
1455 | realSource=nullptr; |
1456 | break; |
1457 | } else if(pArgs->flush && cnv->toULength>0) { |
1458 | /* |
1459 | * the entire input stream is consumed |
1460 | * and there is a partial, truncated input sequence left |
1461 | */ |
1462 | |
1463 | /* inject an error and continue with callback handling */ |
1464 | *err=U_TRUNCATED_CHAR_FOUND; |
1465 | calledCallback=false; /* new error condition */ |
1466 | } else { |
1467 | /* input consumed */ |
1468 | if(pArgs->flush) { |
1469 | /* |
1470 | * return to the conversion loop once more if the flush |
1471 | * flag is set and the conversion function has not |
1472 | * successfully processed the end of the input yet |
1473 | * |
1474 | * (continue converting by breaking out of only the inner loop) |
1475 | */ |
1476 | if(!converterSawEndOfInput) { |
1477 | break; |
1478 | } |
1479 | |
1480 | /* reset the converter without calling the callback function */ |
1481 | _reset(cnv, UCNV_RESET_TO_UNICODE, false); |
1482 | } |
1483 | |
1484 | /* done successfully */ |
1485 | return; |
1486 | } |
1487 | } |
1488 | |
1489 | /* U_FAILURE(*err) */ |
1490 | { |
1491 | UErrorCode e; |
1492 | |
1493 | if( calledCallback || |
1494 | (e=*err)==U_BUFFER_OVERFLOW_ERROR || |
1495 | (e!=U_INVALID_CHAR_FOUND && |
1496 | e!=U_ILLEGAL_CHAR_FOUND && |
1497 | e!=U_TRUNCATED_CHAR_FOUND && |
1498 | e!=U_ILLEGAL_ESCAPE_SEQUENCE && |
1499 | e!=U_UNSUPPORTED_ESCAPE_SEQUENCE) |
1500 | ) { |
1501 | /* |
1502 | * the callback did not or cannot resolve the error: |
1503 | * set output pointers and return |
1504 | * |
1505 | * the check for buffer overflow is redundant but it is |
1506 | * a high-runner case and hopefully documents the intent |
1507 | * well |
1508 | * |
1509 | * if we were replaying, then the replay buffer must be |
1510 | * copied back into the UConverter |
1511 | * and the real arguments must be restored |
1512 | */ |
1513 | if(realSource!=nullptr) { |
1514 | int32_t length; |
1515 | |
1516 | U_ASSERT(cnv->preToULength==0); |
1517 | |
1518 | length=(int32_t)(pArgs->sourceLimit-pArgs->source); |
1519 | if(length>0) { |
1520 | uprv_memcpy(cnv->preToU, pArgs->source, length); |
1521 | cnv->preToULength=(int8_t)-length; |
1522 | } |
1523 | |
1524 | pArgs->source=realSource; |
1525 | pArgs->sourceLimit=realSourceLimit; |
1526 | pArgs->flush=realFlush; |
1527 | } |
1528 | |
1529 | return; |
1530 | } |
1531 | } |
1532 | |
1533 | /* copy toUBytes[] to invalidCharBuffer[] */ |
1534 | errorInputLength=cnv->invalidCharLength=cnv->toULength; |
1535 | if(errorInputLength>0) { |
1536 | uprv_memcpy(cnv->invalidCharBuffer, cnv->toUBytes, errorInputLength); |
1537 | } |
1538 | |
1539 | /* set the converter state to deal with the next character */ |
1540 | cnv->toULength=0; |
1541 | |
1542 | /* call the callback function */ |
1543 | if(cnv->toUCallbackReason==UCNV_ILLEGAL && *err==U_INVALID_CHAR_FOUND) { |
1544 | cnv->toUCallbackReason = UCNV_UNASSIGNED; |
1545 | } |
1546 | cnv->fromCharErrorBehaviour(cnv->toUContext, pArgs, |
1547 | cnv->invalidCharBuffer, errorInputLength, |
1548 | cnv->toUCallbackReason, |
1549 | err); |
1550 | cnv->toUCallbackReason = UCNV_ILLEGAL; /* reset to default value */ |
1551 | |
1552 | /* |
1553 | * loop back to the offset handling |
1554 | * |
1555 | * this flag will indicate after offset handling |
1556 | * that a callback was called; |
1557 | * if the callback did not resolve the error, then we return |
1558 | */ |
1559 | calledCallback=true; |
1560 | } |
1561 | } |
1562 | } |
1563 | |
1564 | /* |
1565 | * Output the toUnicode overflow buffer. |
1566 | * Call this function if(cnv->UCharErrorBufferLength>0). |
1567 | * @return true if overflow |
1568 | */ |
1569 | static UBool |
1570 | ucnv_outputOverflowToUnicode(UConverter *cnv, |
1571 | char16_t **target, const char16_t *targetLimit, |
1572 | int32_t **pOffsets, |
1573 | UErrorCode *err) { |
1574 | int32_t *offsets; |
1575 | char16_t *overflow, *t; |
1576 | int32_t i, length; |
1577 | |
1578 | t=*target; |
1579 | if(pOffsets!=nullptr) { |
1580 | offsets=*pOffsets; |
1581 | } else { |
1582 | offsets=nullptr; |
1583 | } |
1584 | |
1585 | overflow=cnv->UCharErrorBuffer; |
1586 | length=cnv->UCharErrorBufferLength; |
1587 | i=0; |
1588 | while(i<length) { |
1589 | if(t==targetLimit) { |
1590 | /* the overflow buffer contains too much, keep the rest */ |
1591 | int32_t j=0; |
1592 | |
1593 | do { |
1594 | overflow[j++]=overflow[i++]; |
1595 | } while(i<length); |
1596 | |
1597 | cnv->UCharErrorBufferLength=(int8_t)j; |
1598 | *target=t; |
1599 | if(offsets!=nullptr) { |
1600 | *pOffsets=offsets; |
1601 | } |
1602 | *err=U_BUFFER_OVERFLOW_ERROR; |
1603 | return true; |
1604 | } |
1605 | |
1606 | /* copy the overflow contents to the target */ |
1607 | *t++=overflow[i++]; |
1608 | if(offsets!=nullptr) { |
1609 | *offsets++=-1; /* no source index available for old output */ |
1610 | } |
1611 | } |
1612 | |
1613 | /* the overflow buffer is completely copied to the target */ |
1614 | cnv->UCharErrorBufferLength=0; |
1615 | *target=t; |
1616 | if(offsets!=nullptr) { |
1617 | *pOffsets=offsets; |
1618 | } |
1619 | return false; |
1620 | } |
1621 | |
1622 | U_CAPI void U_EXPORT2 |
1623 | ucnv_toUnicode(UConverter *cnv, |
1624 | char16_t **target, const char16_t *targetLimit, |
1625 | const char **source, const char *sourceLimit, |
1626 | int32_t *offsets, |
1627 | UBool flush, |
1628 | UErrorCode *err) { |
1629 | UConverterToUnicodeArgs args; |
1630 | const char *s; |
1631 | char16_t *t; |
1632 | |
1633 | /* check parameters */ |
1634 | if(err==nullptr || U_FAILURE(*err)) { |
1635 | return; |
1636 | } |
1637 | |
1638 | if(cnv==nullptr || target==nullptr || source==nullptr) { |
1639 | *err=U_ILLEGAL_ARGUMENT_ERROR; |
1640 | return; |
1641 | } |
1642 | |
1643 | s=*source; |
1644 | t=*target; |
1645 | |
1646 | if ((const void *)U_MAX_PTR(targetLimit) == (const void *)targetLimit) { |
1647 | /* |
1648 | Prevent code from going into an infinite loop in case we do hit this |
1649 | limit. The limit pointer is expected to be on a char16_t * boundary. |
1650 | This also prevents the next argument check from failing. |
1651 | */ |
1652 | targetLimit = (const char16_t *)(((const char *)targetLimit) - 1); |
1653 | } |
1654 | |
1655 | /* |
1656 | * All these conditions should never happen. |
1657 | * |
1658 | * 1) Make sure that the limits are >= to the address source or target |
1659 | * |
1660 | * 2) Make sure that the buffer sizes do not exceed the number range for |
1661 | * int32_t because some functions use the size (in units or bytes) |
1662 | * rather than comparing pointers, and because offsets are int32_t values. |
1663 | * |
1664 | * size_t is guaranteed to be unsigned and large enough for the job. |
1665 | * |
1666 | * Return with an error instead of adjusting the limits because we would |
1667 | * not be able to maintain the semantics that either the source must be |
1668 | * consumed or the target filled (unless an error occurs). |
1669 | * An adjustment would be sourceLimit=t+0x7fffffff; for example. |
1670 | * |
1671 | * 3) Make sure that the user didn't incorrectly cast a char16_t * pointer |
1672 | * to a char * pointer and provide an incomplete char16_t code unit. |
1673 | */ |
1674 | if (sourceLimit<s || targetLimit<t || |
1675 | ((size_t)(sourceLimit-s)>(size_t)0x7fffffff && sourceLimit>s) || |
1676 | ((size_t)(targetLimit-t)>(size_t)0x3fffffff && targetLimit>t) || |
1677 | (((const char *)targetLimit-(const char *)t) & 1) != 0 |
1678 | ) { |
1679 | *err=U_ILLEGAL_ARGUMENT_ERROR; |
1680 | return; |
1681 | } |
1682 | |
1683 | /* output the target overflow buffer */ |
1684 | if( cnv->UCharErrorBufferLength>0 && |
1685 | ucnv_outputOverflowToUnicode(cnv, target, targetLimit, &offsets, err) |
1686 | ) { |
1687 | /* U_BUFFER_OVERFLOW_ERROR */ |
1688 | return; |
1689 | } |
1690 | /* *target may have moved, therefore stop using t */ |
1691 | |
1692 | if(!flush && s==sourceLimit && cnv->preToULength>=0) { |
1693 | /* the overflow buffer is emptied and there is no new input: we are done */ |
1694 | return; |
1695 | } |
1696 | |
1697 | /* |
1698 | * Do not simply return with a buffer overflow error if |
1699 | * !flush && t==targetLimit |
1700 | * because it is possible that the source will not generate any output. |
1701 | * For example, the skip callback may be called; |
1702 | * it does not output anything. |
1703 | */ |
1704 | |
1705 | /* prepare the converter arguments */ |
1706 | args.converter=cnv; |
1707 | args.flush=flush; |
1708 | args.offsets=offsets; |
1709 | args.source=s; |
1710 | args.sourceLimit=sourceLimit; |
1711 | args.target=*target; |
1712 | args.targetLimit=targetLimit; |
1713 | args.size=sizeof(args); |
1714 | |
1715 | _toUnicodeWithCallback(&args, err); |
1716 | |
1717 | *source=args.source; |
1718 | *target=args.target; |
1719 | } |
1720 | |
1721 | /* ucnv_to/fromUChars() ----------------------------------------------------- */ |
1722 | |
1723 | U_CAPI int32_t U_EXPORT2 |
1724 | ucnv_fromUChars(UConverter *cnv, |
1725 | char *dest, int32_t destCapacity, |
1726 | const char16_t *src, int32_t srcLength, |
1727 | UErrorCode *pErrorCode) { |
1728 | const char16_t *srcLimit; |
1729 | char *originalDest, *destLimit; |
1730 | int32_t destLength; |
1731 | |
1732 | /* check arguments */ |
1733 | if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { |
1734 | return 0; |
1735 | } |
1736 | |
1737 | if( cnv==nullptr || |
1738 | destCapacity<0 || (destCapacity>0 && dest==nullptr) || |
1739 | srcLength<-1 || (srcLength!=0 && src==nullptr) |
1740 | ) { |
1741 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
1742 | return 0; |
1743 | } |
1744 | |
1745 | /* initialize */ |
1746 | ucnv_resetFromUnicode(cnv); |
1747 | originalDest=dest; |
1748 | if(srcLength==-1) { |
1749 | srcLength=u_strlen(src); |
1750 | } |
1751 | if(srcLength>0) { |
1752 | srcLimit=src+srcLength; |
1753 | destCapacity=pinCapacity(dest, destCapacity); |
1754 | destLimit=dest+destCapacity; |
1755 | |
1756 | /* perform the conversion */ |
1757 | ucnv_fromUnicode(cnv, &dest, destLimit, &src, srcLimit, 0, true, pErrorCode); |
1758 | destLength=(int32_t)(dest-originalDest); |
1759 | |
1760 | /* if an overflow occurs, then get the preflighting length */ |
1761 | if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) { |
1762 | char buffer[1024]; |
1763 | |
1764 | destLimit=buffer+sizeof(buffer); |
1765 | do { |
1766 | dest=buffer; |
1767 | *pErrorCode=U_ZERO_ERROR; |
1768 | ucnv_fromUnicode(cnv, &dest, destLimit, &src, srcLimit, 0, true, pErrorCode); |
1769 | destLength+=(int32_t)(dest-buffer); |
1770 | } while(*pErrorCode==U_BUFFER_OVERFLOW_ERROR); |
1771 | } |
1772 | } else { |
1773 | destLength=0; |
1774 | } |
1775 | |
1776 | return u_terminateChars(originalDest, destCapacity, destLength, pErrorCode); |
1777 | } |
1778 | |
1779 | U_CAPI int32_t U_EXPORT2 |
1780 | ucnv_toUChars(UConverter *cnv, |
1781 | char16_t *dest, int32_t destCapacity, |
1782 | const char *src, int32_t srcLength, |
1783 | UErrorCode *pErrorCode) { |
1784 | const char *srcLimit; |
1785 | char16_t *originalDest, *destLimit; |
1786 | int32_t destLength; |
1787 | |
1788 | /* check arguments */ |
1789 | if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { |
1790 | return 0; |
1791 | } |
1792 | |
1793 | if( cnv==nullptr || |
1794 | destCapacity<0 || (destCapacity>0 && dest==nullptr) || |
1795 | srcLength<-1 || (srcLength!=0 && src==nullptr)) |
1796 | { |
1797 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
1798 | return 0; |
1799 | } |
1800 | |
1801 | /* initialize */ |
1802 | ucnv_resetToUnicode(cnv); |
1803 | originalDest=dest; |
1804 | if(srcLength==-1) { |
1805 | srcLength=(int32_t)uprv_strlen(src); |
1806 | } |
1807 | if(srcLength>0) { |
1808 | srcLimit=src+srcLength; |
1809 | destCapacity=pinCapacity(dest, destCapacity); |
1810 | destLimit=dest+destCapacity; |
1811 | |
1812 | /* perform the conversion */ |
1813 | ucnv_toUnicode(cnv, &dest, destLimit, &src, srcLimit, 0, true, pErrorCode); |
1814 | destLength=(int32_t)(dest-originalDest); |
1815 | |
1816 | /* if an overflow occurs, then get the preflighting length */ |
1817 | if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) |
1818 | { |
1819 | char16_t buffer[1024]; |
1820 | |
1821 | destLimit=buffer+UPRV_LENGTHOF(buffer); |
1822 | do { |
1823 | dest=buffer; |
1824 | *pErrorCode=U_ZERO_ERROR; |
1825 | ucnv_toUnicode(cnv, &dest, destLimit, &src, srcLimit, 0, true, pErrorCode); |
1826 | destLength+=(int32_t)(dest-buffer); |
1827 | } |
1828 | while(*pErrorCode==U_BUFFER_OVERFLOW_ERROR); |
1829 | } |
1830 | } else { |
1831 | destLength=0; |
1832 | } |
1833 | |
1834 | return u_terminateUChars(originalDest, destCapacity, destLength, pErrorCode); |
1835 | } |
1836 | |
1837 | /* ucnv_getNextUChar() ------------------------------------------------------ */ |
1838 | |
1839 | U_CAPI UChar32 U_EXPORT2 |
1840 | ucnv_getNextUChar(UConverter *cnv, |
1841 | const char **source, const char *sourceLimit, |
1842 | UErrorCode *err) { |
1843 | UConverterToUnicodeArgs args; |
1844 | char16_t buffer[U16_MAX_LENGTH]; |
1845 | const char *s; |
1846 | UChar32 c; |
1847 | int32_t i, length; |
1848 | |
1849 | /* check parameters */ |
1850 | if(err==nullptr || U_FAILURE(*err)) { |
1851 | return 0xffff; |
1852 | } |
1853 | |
1854 | if(cnv==nullptr || source==nullptr) { |
1855 | *err=U_ILLEGAL_ARGUMENT_ERROR; |
1856 | return 0xffff; |
1857 | } |
1858 | |
1859 | s=*source; |
1860 | if(sourceLimit<s) { |
1861 | *err=U_ILLEGAL_ARGUMENT_ERROR; |
1862 | return 0xffff; |
1863 | } |
1864 | |
1865 | /* |
1866 | * Make sure that the buffer sizes do not exceed the number range for |
1867 | * int32_t because some functions use the size (in units or bytes) |
1868 | * rather than comparing pointers, and because offsets are int32_t values. |
1869 | * |
1870 | * size_t is guaranteed to be unsigned and large enough for the job. |
1871 | * |
1872 | * Return with an error instead of adjusting the limits because we would |
1873 | * not be able to maintain the semantics that either the source must be |
1874 | * consumed or the target filled (unless an error occurs). |
1875 | * An adjustment would be sourceLimit=t+0x7fffffff; for example. |
1876 | */ |
1877 | if(((size_t)(sourceLimit-s)>(size_t)0x7fffffff && sourceLimit>s)) { |
1878 | *err=U_ILLEGAL_ARGUMENT_ERROR; |
1879 | return 0xffff; |
1880 | } |
1881 | |
1882 | c=U_SENTINEL; |
1883 | |
1884 | /* flush the target overflow buffer */ |
1885 | if(cnv->UCharErrorBufferLength>0) { |
1886 | char16_t *overflow; |
1887 | |
1888 | overflow=cnv->UCharErrorBuffer; |
1889 | i=0; |
1890 | length=cnv->UCharErrorBufferLength; |
1891 | U16_NEXT(overflow, i, length, c); |
1892 | |
1893 | /* move the remaining overflow contents up to the beginning */ |
1894 | if((cnv->UCharErrorBufferLength=(int8_t)(length-i))>0) { |
1895 | uprv_memmove(cnv->UCharErrorBuffer, cnv->UCharErrorBuffer+i, |
1896 | cnv->UCharErrorBufferLength*U_SIZEOF_UCHAR); |
1897 | } |
1898 | |
1899 | if(!U16_IS_LEAD(c) || i<length) { |
1900 | return c; |
1901 | } |
1902 | /* |
1903 | * Continue if the overflow buffer contained only a lead surrogate, |
1904 | * in case the converter outputs single surrogates from complete |
1905 | * input sequences. |
1906 | */ |
1907 | } |
1908 | |
1909 | /* |
1910 | * flush==true is implied for ucnv_getNextUChar() |
1911 | * |
1912 | * do not simply return even if s==sourceLimit because the converter may |
1913 | * not have seen flush==true before |
1914 | */ |
1915 | |
1916 | /* prepare the converter arguments */ |
1917 | args.converter=cnv; |
1918 | args.flush=true; |
1919 | args.offsets=nullptr; |
1920 | args.source=s; |
1921 | args.sourceLimit=sourceLimit; |
1922 | args.target=buffer; |
1923 | args.targetLimit=buffer+1; |
1924 | args.size=sizeof(args); |
1925 | |
1926 | if(c<0) { |
1927 | /* |
1928 | * call the native getNextUChar() implementation if we are |
1929 | * at a character boundary (toULength==0) |
1930 | * |
1931 | * unlike with _toUnicode(), getNextUChar() implementations must set |
1932 | * U_TRUNCATED_CHAR_FOUND for truncated input, |
1933 | * in addition to setting toULength/toUBytes[] |
1934 | */ |
1935 | if(cnv->toULength==0 && cnv->sharedData->impl->getNextUChar!=nullptr) { |
1936 | c=cnv->sharedData->impl->getNextUChar(&args, err); |
1937 | *source=s=args.source; |
1938 | if(*err==U_INDEX_OUTOFBOUNDS_ERROR) { |
1939 | /* reset the converter without calling the callback function */ |
1940 | _reset(cnv, UCNV_RESET_TO_UNICODE, false); |
1941 | return 0xffff; /* no output */ |
1942 | } else if(U_SUCCESS(*err) && c>=0) { |
1943 | return c; |
1944 | /* |
1945 | * else fall through to use _toUnicode() because |
1946 | * UCNV_GET_NEXT_UCHAR_USE_TO_U: the native function did not want to handle it after all |
1947 | * U_FAILURE: call _toUnicode() for callback handling (do not output c) |
1948 | */ |
1949 | } |
1950 | } |
1951 | |
1952 | /* convert to one char16_t in buffer[0], or handle getNextUChar() errors */ |
1953 | _toUnicodeWithCallback(&args, err); |
1954 | |
1955 | if(*err==U_BUFFER_OVERFLOW_ERROR) { |
1956 | *err=U_ZERO_ERROR; |
1957 | } |
1958 | |
1959 | i=0; |
1960 | length=(int32_t)(args.target-buffer); |
1961 | } else { |
1962 | /* write the lead surrogate from the overflow buffer */ |
1963 | buffer[0]=(char16_t)c; |
1964 | args.target=buffer+1; |
1965 | i=0; |
1966 | length=1; |
1967 | } |
1968 | |
1969 | /* buffer contents starts at i and ends before length */ |
1970 | |
1971 | if(U_FAILURE(*err)) { |
1972 | c=0xffff; /* no output */ |
1973 | } else if(length==0) { |
1974 | /* no input or only state changes */ |
1975 | *err=U_INDEX_OUTOFBOUNDS_ERROR; |
1976 | /* no need to reset explicitly because _toUnicodeWithCallback() did it */ |
1977 | c=0xffff; /* no output */ |
1978 | } else { |
1979 | c=buffer[0]; |
1980 | i=1; |
1981 | if(!U16_IS_LEAD(c)) { |
1982 | /* consume c=buffer[0], done */ |
1983 | } else { |
1984 | /* got a lead surrogate, see if a trail surrogate follows */ |
1985 | char16_t c2; |
1986 | |
1987 | if(cnv->UCharErrorBufferLength>0) { |
1988 | /* got overflow output from the conversion */ |
1989 | if(U16_IS_TRAIL(c2=cnv->UCharErrorBuffer[0])) { |
1990 | /* got a trail surrogate, too */ |
1991 | c=U16_GET_SUPPLEMENTARY(c, c2); |
1992 | |
1993 | /* move the remaining overflow contents up to the beginning */ |
1994 | if((--cnv->UCharErrorBufferLength)>0) { |
1995 | uprv_memmove(cnv->UCharErrorBuffer, cnv->UCharErrorBuffer+1, |
1996 | cnv->UCharErrorBufferLength*U_SIZEOF_UCHAR); |
1997 | } |
1998 | } else { |
1999 | /* c is an unpaired lead surrogate, just return it */ |
2000 | } |
2001 | } else if(args.source<sourceLimit) { |
2002 | /* convert once more, to buffer[1] */ |
2003 | args.targetLimit=buffer+2; |
2004 | _toUnicodeWithCallback(&args, err); |
2005 | if(*err==U_BUFFER_OVERFLOW_ERROR) { |
2006 | *err=U_ZERO_ERROR; |
2007 | } |
2008 | |
2009 | length=(int32_t)(args.target-buffer); |
2010 | if(U_SUCCESS(*err) && length==2 && U16_IS_TRAIL(c2=buffer[1])) { |
2011 | /* got a trail surrogate, too */ |
2012 | c=U16_GET_SUPPLEMENTARY(c, c2); |
2013 | i=2; |
2014 | } |
2015 | } |
2016 | } |
2017 | } |
2018 | |
2019 | /* |
2020 | * move leftover output from buffer[i..length[ |
2021 | * into the beginning of the overflow buffer |
2022 | */ |
2023 | if(i<length) { |
2024 | /* move further overflow back */ |
2025 | int32_t delta=length-i; |
2026 | if((length=cnv->UCharErrorBufferLength)>0) { |
2027 | uprv_memmove(cnv->UCharErrorBuffer+delta, cnv->UCharErrorBuffer, |
2028 | length*U_SIZEOF_UCHAR); |
2029 | } |
2030 | cnv->UCharErrorBufferLength=(int8_t)(length+delta); |
2031 | |
2032 | cnv->UCharErrorBuffer[0]=buffer[i++]; |
2033 | if(delta>1) { |
2034 | cnv->UCharErrorBuffer[1]=buffer[i]; |
2035 | } |
2036 | } |
2037 | |
2038 | *source=args.source; |
2039 | return c; |
2040 | } |
2041 | |
2042 | /* ucnv_convert() and siblings ---------------------------------------------- */ |
2043 | |
2044 | U_CAPI void U_EXPORT2 |
2045 | ucnv_convertEx(UConverter *targetCnv, UConverter *sourceCnv, |
2046 | char **target, const char *targetLimit, |
2047 | const char **source, const char *sourceLimit, |
2048 | char16_t *pivotStart, char16_t **pivotSource, |
2049 | char16_t **pivotTarget, const char16_t *pivotLimit, |
2050 | UBool reset, UBool flush, |
2051 | UErrorCode *pErrorCode) { |
2052 | char16_t pivotBuffer[CHUNK_SIZE]; |
2053 | const char16_t *myPivotSource; |
2054 | char16_t *myPivotTarget; |
2055 | const char *s; |
2056 | char *t; |
2057 | |
2058 | UConverterToUnicodeArgs toUArgs; |
2059 | UConverterFromUnicodeArgs fromUArgs; |
2060 | UConverterConvert convert; |
2061 | |
2062 | /* error checking */ |
2063 | if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { |
2064 | return; |
2065 | } |
2066 | |
2067 | if( targetCnv==nullptr || sourceCnv==nullptr || |
2068 | source==nullptr || *source==nullptr || |
2069 | target==nullptr || *target==nullptr || targetLimit==nullptr |
2070 | ) { |
2071 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
2072 | return; |
2073 | } |
2074 | |
2075 | s=*source; |
2076 | t=*target; |
2077 | if((sourceLimit!=nullptr && sourceLimit<s) || targetLimit<t) { |
2078 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
2079 | return; |
2080 | } |
2081 | |
2082 | /* |
2083 | * Make sure that the buffer sizes do not exceed the number range for |
2084 | * int32_t. See ucnv_toUnicode() for a more detailed comment. |
2085 | */ |
2086 | if( |
2087 | (sourceLimit!=nullptr && ((size_t)(sourceLimit-s)>(size_t)0x7fffffff && sourceLimit>s)) || |
2088 | ((size_t)(targetLimit-t)>(size_t)0x7fffffff && targetLimit>t) |
2089 | ) { |
2090 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
2091 | return; |
2092 | } |
2093 | |
2094 | if(pivotStart==nullptr) { |
2095 | if(!flush) { |
2096 | /* streaming conversion requires an explicit pivot buffer */ |
2097 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
2098 | return; |
2099 | } |
2100 | |
2101 | /* use the stack pivot buffer */ |
2102 | myPivotSource=myPivotTarget=pivotStart=pivotBuffer; |
2103 | pivotSource=(char16_t **)&myPivotSource; |
2104 | pivotTarget=&myPivotTarget; |
2105 | pivotLimit=pivotBuffer+CHUNK_SIZE; |
2106 | } else if( pivotStart>=pivotLimit || |
2107 | pivotSource==nullptr || *pivotSource==nullptr || |
2108 | pivotTarget==nullptr || *pivotTarget==nullptr || |
2109 | pivotLimit==nullptr |
2110 | ) { |
2111 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
2112 | return; |
2113 | } |
2114 | |
2115 | if(sourceLimit==nullptr) { |
2116 | /* get limit of single-byte-NUL-terminated source string */ |
2117 | sourceLimit=uprv_strchr(*source, 0); |
2118 | } |
2119 | |
2120 | if(reset) { |
2121 | ucnv_resetToUnicode(sourceCnv); |
2122 | ucnv_resetFromUnicode(targetCnv); |
2123 | *pivotSource=*pivotTarget=pivotStart; |
2124 | } else if(targetCnv->charErrorBufferLength>0) { |
2125 | /* output the targetCnv overflow buffer */ |
2126 | if(ucnv_outputOverflowFromUnicode(targetCnv, target, targetLimit, nullptr, pErrorCode)) { |
2127 | /* U_BUFFER_OVERFLOW_ERROR */ |
2128 | return; |
2129 | } |
2130 | /* *target has moved, therefore stop using t */ |
2131 | |
2132 | if( !flush && |
2133 | targetCnv->preFromULength>=0 && *pivotSource==*pivotTarget && |
2134 | sourceCnv->UCharErrorBufferLength==0 && sourceCnv->preToULength>=0 && s==sourceLimit |
2135 | ) { |
2136 | /* the fromUnicode overflow buffer is emptied and there is no new input: we are done */ |
2137 | return; |
2138 | } |
2139 | } |
2140 | |
2141 | /* Is direct-UTF-8 conversion available? */ |
2142 | if( sourceCnv->sharedData->staticData->conversionType==UCNV_UTF8 && |
2143 | targetCnv->sharedData->impl->fromUTF8!=nullptr |
2144 | ) { |
2145 | convert=targetCnv->sharedData->impl->fromUTF8; |
2146 | } else if( targetCnv->sharedData->staticData->conversionType==UCNV_UTF8 && |
2147 | sourceCnv->sharedData->impl->toUTF8!=nullptr |
2148 | ) { |
2149 | convert=sourceCnv->sharedData->impl->toUTF8; |
2150 | } else { |
2151 | convert=nullptr; |
2152 | } |
2153 | |
2154 | /* |
2155 | * If direct-UTF-8 conversion is available, then we use a smaller |
2156 | * pivot buffer for error handling and partial matches |
2157 | * so that we quickly return to direct conversion. |
2158 | * |
2159 | * 32 is large enough for UCNV_EXT_MAX_UCHARS and UCNV_ERROR_BUFFER_LENGTH. |
2160 | * |
2161 | * We could reduce the pivot buffer size further, at the cost of |
2162 | * buffer overflows from callbacks. |
2163 | * The pivot buffer should not be smaller than the maximum number of |
2164 | * fromUnicode extension table input UChars |
2165 | * (for m:n conversion, see |
2166 | * targetCnv->sharedData->mbcs.extIndexes[UCNV_EXT_COUNT_UCHARS]) |
2167 | * or 2 for surrogate pairs. |
2168 | * |
2169 | * Too small a buffer can cause thrashing between pivoting and direct |
2170 | * conversion, with function call overhead outweighing the benefits |
2171 | * of direct conversion. |
2172 | */ |
2173 | if(convert!=nullptr && (pivotLimit-pivotStart)>32) { |
2174 | pivotLimit=pivotStart+32; |
2175 | } |
2176 | |
2177 | /* prepare the converter arguments */ |
2178 | fromUArgs.converter=targetCnv; |
2179 | fromUArgs.flush=false; |
2180 | fromUArgs.offsets=nullptr; |
2181 | fromUArgs.target=*target; |
2182 | fromUArgs.targetLimit=targetLimit; |
2183 | fromUArgs.size=sizeof(fromUArgs); |
2184 | |
2185 | toUArgs.converter=sourceCnv; |
2186 | toUArgs.flush=flush; |
2187 | toUArgs.offsets=nullptr; |
2188 | toUArgs.source=s; |
2189 | toUArgs.sourceLimit=sourceLimit; |
2190 | toUArgs.targetLimit=pivotLimit; |
2191 | toUArgs.size=sizeof(toUArgs); |
2192 | |
2193 | /* |
2194 | * TODO: Consider separating this function into two functions, |
2195 | * extracting exactly the conversion loop, |
2196 | * for readability and to reduce the set of visible variables. |
2197 | * |
2198 | * Otherwise stop using s and t from here on. |
2199 | */ |
2200 | s=t=nullptr; |
2201 | |
2202 | /* |
2203 | * conversion loop |
2204 | * |
2205 | * The sequence of steps in the loop may appear backward, |
2206 | * but the principle is simple: |
2207 | * In the chain of |
2208 | * source - sourceCnv overflow - pivot - targetCnv overflow - target |
2209 | * empty out later buffers before refilling them from earlier ones. |
2210 | * |
2211 | * The targetCnv overflow buffer is flushed out only once before the loop. |
2212 | */ |
2213 | for(;;) { |
2214 | /* |
2215 | * if(pivot not empty or error or replay or flush fromUnicode) { |
2216 | * fromUnicode(pivot -> target); |
2217 | * } |
2218 | * |
2219 | * For pivoting conversion; and for direct conversion for |
2220 | * error callback handling and flushing the replay buffer. |
2221 | */ |
2222 | if( *pivotSource<*pivotTarget || |
2223 | U_FAILURE(*pErrorCode) || |
2224 | targetCnv->preFromULength<0 || |
2225 | fromUArgs.flush |
2226 | ) { |
2227 | fromUArgs.source=*pivotSource; |
2228 | fromUArgs.sourceLimit=*pivotTarget; |
2229 | _fromUnicodeWithCallback(&fromUArgs, pErrorCode); |
2230 | if(U_FAILURE(*pErrorCode)) { |
2231 | /* target overflow, or conversion error */ |
2232 | *pivotSource=(char16_t *)fromUArgs.source; |
2233 | break; |
2234 | } |
2235 | |
2236 | /* |
2237 | * _fromUnicodeWithCallback() must have consumed the pivot contents |
2238 | * (*pivotSource==*pivotTarget) since it returned with U_SUCCESS() |
2239 | */ |
2240 | } |
2241 | |
2242 | /* The pivot buffer is empty; reset it so we start at pivotStart. */ |
2243 | *pivotSource=*pivotTarget=pivotStart; |
2244 | |
2245 | /* |
2246 | * if(sourceCnv overflow buffer not empty) { |
2247 | * move(sourceCnv overflow buffer -> pivot); |
2248 | * continue; |
2249 | * } |
2250 | */ |
2251 | /* output the sourceCnv overflow buffer */ |
2252 | if(sourceCnv->UCharErrorBufferLength>0) { |
2253 | if(ucnv_outputOverflowToUnicode(sourceCnv, pivotTarget, pivotLimit, nullptr, pErrorCode)) { |
2254 | /* U_BUFFER_OVERFLOW_ERROR */ |
2255 | *pErrorCode=U_ZERO_ERROR; |
2256 | } |
2257 | continue; |
2258 | } |
2259 | |
2260 | /* |
2261 | * check for end of input and break if done |
2262 | * |
2263 | * Checking both flush and fromUArgs.flush ensures that the converters |
2264 | * have been called with the flush flag set if the ucnv_convertEx() |
2265 | * caller set it. |
2266 | */ |
2267 | if( toUArgs.source==sourceLimit && |
2268 | sourceCnv->preToULength>=0 && sourceCnv->toULength==0 && |
2269 | (!flush || fromUArgs.flush) |
2270 | ) { |
2271 | /* done successfully */ |
2272 | break; |
2273 | } |
2274 | |
2275 | /* |
2276 | * use direct conversion if available |
2277 | * but not if continuing a partial match |
2278 | * or flushing the toUnicode replay buffer |
2279 | */ |
2280 | if(convert!=nullptr && targetCnv->preFromUFirstCP<0 && sourceCnv->preToULength==0) { |
2281 | if(*pErrorCode==U_USING_DEFAULT_WARNING) { |
2282 | /* remove a warning that may be set by this function */ |
2283 | *pErrorCode=U_ZERO_ERROR; |
2284 | } |
2285 | convert(&fromUArgs, &toUArgs, pErrorCode); |
2286 | if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) { |
2287 | break; |
2288 | } else if(U_FAILURE(*pErrorCode)) { |
2289 | if(sourceCnv->toULength>0) { |
2290 | /* |
2291 | * Fall through to calling _toUnicodeWithCallback() |
2292 | * for callback handling. |
2293 | * |
2294 | * The pivot buffer will be reset with |
2295 | * *pivotSource=*pivotTarget=pivotStart; |
2296 | * which indicates a toUnicode error to the caller |
2297 | * (*pivotSource==pivotStart shows no pivot UChars consumed). |
2298 | */ |
2299 | } else { |
2300 | /* |
2301 | * Indicate a fromUnicode error to the caller |
2302 | * (*pivotSource>pivotStart shows some pivot UChars consumed). |
2303 | */ |
2304 | *pivotSource=*pivotTarget=pivotStart+1; |
2305 | /* |
2306 | * Loop around to calling _fromUnicodeWithCallbacks() |
2307 | * for callback handling. |
2308 | */ |
2309 | continue; |
2310 | } |
2311 | } else if(*pErrorCode==U_USING_DEFAULT_WARNING) { |
2312 | /* |
2313 | * No error, but the implementation requested to temporarily |
2314 | * fall back to pivoting. |
2315 | */ |
2316 | *pErrorCode=U_ZERO_ERROR; |
2317 | /* |
2318 | * The following else branches are almost identical to the end-of-input |
2319 | * handling in _toUnicodeWithCallback(). |
2320 | * Avoid calling it just for the end of input. |
2321 | */ |
2322 | } else if(flush && sourceCnv->toULength>0) { /* flush==toUArgs.flush */ |
2323 | /* |
2324 | * the entire input stream is consumed |
2325 | * and there is a partial, truncated input sequence left |
2326 | */ |
2327 | |
2328 | /* inject an error and continue with callback handling */ |
2329 | *pErrorCode=U_TRUNCATED_CHAR_FOUND; |
2330 | } else { |
2331 | /* input consumed */ |
2332 | if(flush) { |
2333 | /* reset the converters without calling the callback functions */ |
2334 | _reset(sourceCnv, UCNV_RESET_TO_UNICODE, false); |
2335 | _reset(targetCnv, UCNV_RESET_FROM_UNICODE, false); |
2336 | } |
2337 | |
2338 | /* done successfully */ |
2339 | break; |
2340 | } |
2341 | } |
2342 | |
2343 | /* |
2344 | * toUnicode(source -> pivot); |
2345 | * |
2346 | * For pivoting conversion; and for direct conversion for |
2347 | * error callback handling, continuing partial matches |
2348 | * and flushing the replay buffer. |
2349 | * |
2350 | * The pivot buffer is empty and reset. |
2351 | */ |
2352 | toUArgs.target=pivotStart; /* ==*pivotTarget */ |
2353 | /* toUArgs.targetLimit=pivotLimit; already set before the loop */ |
2354 | _toUnicodeWithCallback(&toUArgs, pErrorCode); |
2355 | *pivotTarget=toUArgs.target; |
2356 | if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) { |
2357 | /* pivot overflow: continue with the conversion loop */ |
2358 | *pErrorCode=U_ZERO_ERROR; |
2359 | } else if(U_FAILURE(*pErrorCode) || (!flush && *pivotTarget==pivotStart)) { |
2360 | /* conversion error, or there was nothing left to convert */ |
2361 | break; |
2362 | } |
2363 | /* |
2364 | * else: |
2365 | * _toUnicodeWithCallback() wrote into the pivot buffer, |
2366 | * continue with fromUnicode conversion. |
2367 | * |
2368 | * Set the fromUnicode flush flag if we flush and if toUnicode has |
2369 | * processed the end of the input. |
2370 | */ |
2371 | if( flush && toUArgs.source==sourceLimit && |
2372 | sourceCnv->preToULength>=0 && |
2373 | sourceCnv->UCharErrorBufferLength==0 |
2374 | ) { |
2375 | fromUArgs.flush=true; |
2376 | } |
2377 | } |
2378 | |
2379 | /* |
2380 | * The conversion loop is exited when one of the following is true: |
2381 | * - the entire source text has been converted successfully to the target buffer |
2382 | * - a target buffer overflow occurred |
2383 | * - a conversion error occurred |
2384 | */ |
2385 | |
2386 | *source=toUArgs.source; |
2387 | *target=fromUArgs.target; |
2388 | |
2389 | /* terminate the target buffer if possible */ |
2390 | if(flush && U_SUCCESS(*pErrorCode)) { |
2391 | if(*target!=targetLimit) { |
2392 | **target=0; |
2393 | if(*pErrorCode==U_STRING_NOT_TERMINATED_WARNING) { |
2394 | *pErrorCode=U_ZERO_ERROR; |
2395 | } |
2396 | } else { |
2397 | *pErrorCode=U_STRING_NOT_TERMINATED_WARNING; |
2398 | } |
2399 | } |
2400 | } |
2401 | |
2402 | /* internal implementation of ucnv_convert() etc. with preflighting */ |
2403 | static int32_t |
2404 | ucnv_internalConvert(UConverter *outConverter, UConverter *inConverter, |
2405 | char *target, int32_t targetCapacity, |
2406 | const char *source, int32_t sourceLength, |
2407 | UErrorCode *pErrorCode) { |
2408 | char16_t pivotBuffer[CHUNK_SIZE]; |
2409 | char16_t *pivot, *pivot2; |
2410 | |
2411 | char *myTarget; |
2412 | const char *sourceLimit; |
2413 | const char *targetLimit; |
2414 | int32_t targetLength=0; |
2415 | |
2416 | /* set up */ |
2417 | if(sourceLength<0) { |
2418 | sourceLimit=uprv_strchr(source, 0); |
2419 | } else { |
2420 | sourceLimit=source+sourceLength; |
2421 | } |
2422 | |
2423 | /* if there is no input data, we're done */ |
2424 | if(source==sourceLimit) { |
2425 | return u_terminateChars(target, targetCapacity, 0, pErrorCode); |
2426 | } |
2427 | |
2428 | pivot=pivot2=pivotBuffer; |
2429 | myTarget=target; |
2430 | targetLength=0; |
2431 | |
2432 | if(targetCapacity>0) { |
2433 | /* perform real conversion */ |
2434 | targetLimit=target+targetCapacity; |
2435 | ucnv_convertEx(outConverter, inConverter, |
2436 | &myTarget, targetLimit, |
2437 | &source, sourceLimit, |
2438 | pivotBuffer, &pivot, &pivot2, pivotBuffer+CHUNK_SIZE, |
2439 | false, |
2440 | true, |
2441 | pErrorCode); |
2442 | targetLength=(int32_t)(myTarget-target); |
2443 | } |
2444 | |
2445 | /* |
2446 | * If the output buffer is exhausted (or we are only "preflighting"), we need to stop writing |
2447 | * to it but continue the conversion in order to store in targetCapacity |
2448 | * the number of bytes that was required. |
2449 | */ |
2450 | if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR || targetCapacity==0) |
2451 | { |
2452 | char targetBuffer[CHUNK_SIZE]; |
2453 | |
2454 | targetLimit=targetBuffer+CHUNK_SIZE; |
2455 | do { |
2456 | *pErrorCode=U_ZERO_ERROR; |
2457 | myTarget=targetBuffer; |
2458 | ucnv_convertEx(outConverter, inConverter, |
2459 | &myTarget, targetLimit, |
2460 | &source, sourceLimit, |
2461 | pivotBuffer, &pivot, &pivot2, pivotBuffer+CHUNK_SIZE, |
2462 | false, |
2463 | true, |
2464 | pErrorCode); |
2465 | targetLength+=(int32_t)(myTarget-targetBuffer); |
2466 | } while(*pErrorCode==U_BUFFER_OVERFLOW_ERROR); |
2467 | |
2468 | /* done with preflighting, set warnings and errors as appropriate */ |
2469 | return u_terminateChars(target, targetCapacity, targetLength, pErrorCode); |
2470 | } |
2471 | |
2472 | /* no need to call u_terminateChars() because ucnv_convertEx() took care of that */ |
2473 | return targetLength; |
2474 | } |
2475 | |
2476 | U_CAPI int32_t U_EXPORT2 |
2477 | ucnv_convert(const char *toConverterName, const char *fromConverterName, |
2478 | char *target, int32_t targetCapacity, |
2479 | const char *source, int32_t sourceLength, |
2480 | UErrorCode *pErrorCode) { |
2481 | UConverter in, out; /* stack-allocated */ |
2482 | UConverter *inConverter, *outConverter; |
2483 | int32_t targetLength; |
2484 | |
2485 | if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { |
2486 | return 0; |
2487 | } |
2488 | |
2489 | if( source==nullptr || sourceLength<-1 || |
2490 | targetCapacity<0 || (targetCapacity>0 && target==nullptr) |
2491 | ) { |
2492 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
2493 | return 0; |
2494 | } |
2495 | |
2496 | /* if there is no input data, we're done */ |
2497 | if(sourceLength==0 || (sourceLength<0 && *source==0)) { |
2498 | return u_terminateChars(target, targetCapacity, 0, pErrorCode); |
2499 | } |
2500 | |
2501 | /* create the converters */ |
2502 | inConverter=ucnv_createConverter(&in, fromConverterName, pErrorCode); |
2503 | if(U_FAILURE(*pErrorCode)) { |
2504 | return 0; |
2505 | } |
2506 | |
2507 | outConverter=ucnv_createConverter(&out, toConverterName, pErrorCode); |
2508 | if(U_FAILURE(*pErrorCode)) { |
2509 | ucnv_close(inConverter); |
2510 | return 0; |
2511 | } |
2512 | |
2513 | targetLength=ucnv_internalConvert(outConverter, inConverter, |
2514 | target, targetCapacity, |
2515 | source, sourceLength, |
2516 | pErrorCode); |
2517 | |
2518 | ucnv_close(inConverter); |
2519 | ucnv_close(outConverter); |
2520 | |
2521 | return targetLength; |
2522 | } |
2523 | |
2524 | /* @internal */ |
2525 | static int32_t |
2526 | ucnv_convertAlgorithmic(UBool convertToAlgorithmic, |
2527 | UConverterType algorithmicType, |
2528 | UConverter *cnv, |
2529 | char *target, int32_t targetCapacity, |
2530 | const char *source, int32_t sourceLength, |
2531 | UErrorCode *pErrorCode) { |
2532 | UConverter algoConverterStatic; /* stack-allocated */ |
2533 | UConverter *algoConverter, *to, *from; |
2534 | int32_t targetLength; |
2535 | |
2536 | if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { |
2537 | return 0; |
2538 | } |
2539 | |
2540 | if( cnv==nullptr || source==nullptr || sourceLength<-1 || |
2541 | targetCapacity<0 || (targetCapacity>0 && target==nullptr) |
2542 | ) { |
2543 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
2544 | return 0; |
2545 | } |
2546 | |
2547 | /* if there is no input data, we're done */ |
2548 | if(sourceLength==0 || (sourceLength<0 && *source==0)) { |
2549 | return u_terminateChars(target, targetCapacity, 0, pErrorCode); |
2550 | } |
2551 | |
2552 | /* create the algorithmic converter */ |
2553 | algoConverter=ucnv_createAlgorithmicConverter(&algoConverterStatic, algorithmicType, |
2554 | "" , 0, pErrorCode); |
2555 | if(U_FAILURE(*pErrorCode)) { |
2556 | return 0; |
2557 | } |
2558 | |
2559 | /* reset the other converter */ |
2560 | if(convertToAlgorithmic) { |
2561 | /* cnv->Unicode->algo */ |
2562 | ucnv_resetToUnicode(cnv); |
2563 | to=algoConverter; |
2564 | from=cnv; |
2565 | } else { |
2566 | /* algo->Unicode->cnv */ |
2567 | ucnv_resetFromUnicode(cnv); |
2568 | from=algoConverter; |
2569 | to=cnv; |
2570 | } |
2571 | |
2572 | targetLength=ucnv_internalConvert(to, from, |
2573 | target, targetCapacity, |
2574 | source, sourceLength, |
2575 | pErrorCode); |
2576 | |
2577 | ucnv_close(algoConverter); |
2578 | |
2579 | return targetLength; |
2580 | } |
2581 | |
2582 | U_CAPI int32_t U_EXPORT2 |
2583 | ucnv_toAlgorithmic(UConverterType algorithmicType, |
2584 | UConverter *cnv, |
2585 | char *target, int32_t targetCapacity, |
2586 | const char *source, int32_t sourceLength, |
2587 | UErrorCode *pErrorCode) { |
2588 | return ucnv_convertAlgorithmic(true, algorithmicType, cnv, |
2589 | target, targetCapacity, |
2590 | source, sourceLength, |
2591 | pErrorCode); |
2592 | } |
2593 | |
2594 | U_CAPI int32_t U_EXPORT2 |
2595 | ucnv_fromAlgorithmic(UConverter *cnv, |
2596 | UConverterType algorithmicType, |
2597 | char *target, int32_t targetCapacity, |
2598 | const char *source, int32_t sourceLength, |
2599 | UErrorCode *pErrorCode) UPRV_NO_SANITIZE_UNDEFINED { |
2600 | |
2601 | if(algorithmicType<0 || UCNV_NUMBER_OF_SUPPORTED_CONVERTER_TYPES<=algorithmicType) { |
2602 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
2603 | return 0; |
2604 | } |
2605 | return ucnv_convertAlgorithmic(false, algorithmicType, cnv, |
2606 | target, targetCapacity, |
2607 | source, sourceLength, |
2608 | pErrorCode); |
2609 | } |
2610 | |
2611 | U_CAPI UConverterType U_EXPORT2 |
2612 | ucnv_getType(const UConverter* converter) |
2613 | { |
2614 | int8_t type = converter->sharedData->staticData->conversionType; |
2615 | #if !UCONFIG_NO_LEGACY_CONVERSION |
2616 | if(type == UCNV_MBCS) { |
2617 | return ucnv_MBCSGetType(converter); |
2618 | } |
2619 | #endif |
2620 | return (UConverterType)type; |
2621 | } |
2622 | |
2623 | U_CAPI void U_EXPORT2 |
2624 | ucnv_getStarters(const UConverter* converter, |
2625 | UBool starters[256], |
2626 | UErrorCode* err) |
2627 | { |
2628 | if (err == nullptr || U_FAILURE(*err)) { |
2629 | return; |
2630 | } |
2631 | |
2632 | if(converter->sharedData->impl->getStarters != nullptr) { |
2633 | converter->sharedData->impl->getStarters(converter, starters, err); |
2634 | } else { |
2635 | *err = U_ILLEGAL_ARGUMENT_ERROR; |
2636 | } |
2637 | } |
2638 | |
2639 | static const UAmbiguousConverter *ucnv_getAmbiguous(const UConverter *cnv) |
2640 | { |
2641 | UErrorCode errorCode; |
2642 | const char *name; |
2643 | int32_t i; |
2644 | |
2645 | if(cnv==nullptr) { |
2646 | return nullptr; |
2647 | } |
2648 | |
2649 | errorCode=U_ZERO_ERROR; |
2650 | name=ucnv_getName(cnv, &errorCode); |
2651 | if(U_FAILURE(errorCode)) { |
2652 | return nullptr; |
2653 | } |
2654 | |
2655 | for(i=0; i<UPRV_LENGTHOF(ambiguousConverters); ++i) |
2656 | { |
2657 | if(0==uprv_strcmp(name, ambiguousConverters[i].name)) |
2658 | { |
2659 | return ambiguousConverters+i; |
2660 | } |
2661 | } |
2662 | |
2663 | return nullptr; |
2664 | } |
2665 | |
2666 | U_CAPI void U_EXPORT2 |
2667 | ucnv_fixFileSeparator(const UConverter *cnv, |
2668 | char16_t* source, |
2669 | int32_t sourceLength) { |
2670 | const UAmbiguousConverter *a; |
2671 | int32_t i; |
2672 | char16_t variant5c; |
2673 | |
2674 | if(cnv==nullptr || source==nullptr || sourceLength<=0 || (a=ucnv_getAmbiguous(cnv))==nullptr) |
2675 | { |
2676 | return; |
2677 | } |
2678 | |
2679 | variant5c=a->variant5c; |
2680 | for(i=0; i<sourceLength; ++i) { |
2681 | if(source[i]==variant5c) { |
2682 | source[i]=0x5c; |
2683 | } |
2684 | } |
2685 | } |
2686 | |
2687 | U_CAPI UBool U_EXPORT2 |
2688 | ucnv_isAmbiguous(const UConverter *cnv) { |
2689 | return (UBool)(ucnv_getAmbiguous(cnv)!=nullptr); |
2690 | } |
2691 | |
2692 | U_CAPI void U_EXPORT2 |
2693 | ucnv_setFallback(UConverter *cnv, UBool usesFallback) |
2694 | { |
2695 | cnv->useFallback = usesFallback; |
2696 | } |
2697 | |
2698 | U_CAPI UBool U_EXPORT2 |
2699 | ucnv_usesFallback(const UConverter *cnv) |
2700 | { |
2701 | return cnv->useFallback; |
2702 | } |
2703 | |
2704 | U_CAPI void U_EXPORT2 |
2705 | ucnv_getInvalidChars (const UConverter * converter, |
2706 | char *errBytes, |
2707 | int8_t * len, |
2708 | UErrorCode * err) |
2709 | { |
2710 | if (err == nullptr || U_FAILURE(*err)) |
2711 | { |
2712 | return; |
2713 | } |
2714 | if (len == nullptr || errBytes == nullptr || converter == nullptr) |
2715 | { |
2716 | *err = U_ILLEGAL_ARGUMENT_ERROR; |
2717 | return; |
2718 | } |
2719 | if (*len < converter->invalidCharLength) |
2720 | { |
2721 | *err = U_INDEX_OUTOFBOUNDS_ERROR; |
2722 | return; |
2723 | } |
2724 | if ((*len = converter->invalidCharLength) > 0) |
2725 | { |
2726 | uprv_memcpy (errBytes, converter->invalidCharBuffer, *len); |
2727 | } |
2728 | } |
2729 | |
2730 | U_CAPI void U_EXPORT2 |
2731 | ucnv_getInvalidUChars (const UConverter * converter, |
2732 | char16_t *errChars, |
2733 | int8_t * len, |
2734 | UErrorCode * err) |
2735 | { |
2736 | if (err == nullptr || U_FAILURE(*err)) |
2737 | { |
2738 | return; |
2739 | } |
2740 | if (len == nullptr || errChars == nullptr || converter == nullptr) |
2741 | { |
2742 | *err = U_ILLEGAL_ARGUMENT_ERROR; |
2743 | return; |
2744 | } |
2745 | if (*len < converter->invalidUCharLength) |
2746 | { |
2747 | *err = U_INDEX_OUTOFBOUNDS_ERROR; |
2748 | return; |
2749 | } |
2750 | if ((*len = converter->invalidUCharLength) > 0) |
2751 | { |
2752 | u_memcpy (errChars, converter->invalidUCharBuffer, *len); |
2753 | } |
2754 | } |
2755 | |
2756 | #define SIG_MAX_LEN 5 |
2757 | |
2758 | U_CAPI const char* U_EXPORT2 |
2759 | ucnv_detectUnicodeSignature( const char* source, |
2760 | int32_t sourceLength, |
2761 | int32_t* signatureLength, |
2762 | UErrorCode* pErrorCode) { |
2763 | int32_t dummy; |
2764 | |
2765 | /* initial 0xa5 bytes: make sure that if we read <SIG_MAX_LEN |
2766 | * bytes we don't misdetect something |
2767 | */ |
2768 | char start[SIG_MAX_LEN]={ '\xa5', '\xa5', '\xa5', '\xa5', '\xa5' }; |
2769 | int i = 0; |
2770 | |
2771 | if((pErrorCode==nullptr) || U_FAILURE(*pErrorCode)){ |
2772 | return nullptr; |
2773 | } |
2774 | |
2775 | if(source == nullptr || sourceLength < -1){ |
2776 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
2777 | return nullptr; |
2778 | } |
2779 | |
2780 | if(signatureLength == nullptr) { |
2781 | signatureLength = &dummy; |
2782 | } |
2783 | |
2784 | if(sourceLength==-1){ |
2785 | sourceLength=(int32_t)uprv_strlen(source); |
2786 | } |
2787 | |
2788 | |
2789 | while(i<sourceLength&& i<SIG_MAX_LEN){ |
2790 | start[i]=source[i]; |
2791 | i++; |
2792 | } |
2793 | |
2794 | if(start[0] == '\xFE' && start[1] == '\xFF') { |
2795 | *signatureLength=2; |
2796 | return "UTF-16BE" ; |
2797 | } else if(start[0] == '\xFF' && start[1] == '\xFE') { |
2798 | if(start[2] == '\x00' && start[3] =='\x00') { |
2799 | *signatureLength=4; |
2800 | return "UTF-32LE" ; |
2801 | } else { |
2802 | *signatureLength=2; |
2803 | return "UTF-16LE" ; |
2804 | } |
2805 | } else if(start[0] == '\xEF' && start[1] == '\xBB' && start[2] == '\xBF') { |
2806 | *signatureLength=3; |
2807 | return "UTF-8" ; |
2808 | } else if(start[0] == '\x00' && start[1] == '\x00' && |
2809 | start[2] == '\xFE' && start[3]=='\xFF') { |
2810 | *signatureLength=4; |
2811 | return "UTF-32BE" ; |
2812 | } else if(start[0] == '\x0E' && start[1] == '\xFE' && start[2] == '\xFF') { |
2813 | *signatureLength=3; |
2814 | return "SCSU" ; |
2815 | } else if(start[0] == '\xFB' && start[1] == '\xEE' && start[2] == '\x28') { |
2816 | *signatureLength=3; |
2817 | return "BOCU-1" ; |
2818 | } else if(start[0] == '\x2B' && start[1] == '\x2F' && start[2] == '\x76') { |
2819 | /* |
2820 | * UTF-7: Initial U+FEFF is encoded as +/v8 or +/v9 or +/v+ or +/v/ |
2821 | * depending on the second UTF-16 code unit. |
2822 | * Detect the entire, closed Unicode mode sequence +/v8- for only U+FEFF |
2823 | * if it occurs. |
2824 | * |
2825 | * So far we have +/v |
2826 | */ |
2827 | if(start[3] == '\x38' && start[4] == '\x2D') { |
2828 | /* 5 bytes +/v8- */ |
2829 | *signatureLength=5; |
2830 | return "UTF-7" ; |
2831 | } else if(start[3] == '\x38' || start[3] == '\x39' || start[3] == '\x2B' || start[3] == '\x2F') { |
2832 | /* 4 bytes +/v8 or +/v9 or +/v+ or +/v/ */ |
2833 | *signatureLength=4; |
2834 | return "UTF-7" ; |
2835 | } |
2836 | }else if(start[0]=='\xDD' && start[1]== '\x73'&& start[2]=='\x66' && start[3]=='\x73'){ |
2837 | *signatureLength=4; |
2838 | return "UTF-EBCDIC" ; |
2839 | } |
2840 | |
2841 | |
2842 | /* no known Unicode signature byte sequence recognized */ |
2843 | *signatureLength=0; |
2844 | return nullptr; |
2845 | } |
2846 | |
2847 | U_CAPI int32_t U_EXPORT2 |
2848 | ucnv_fromUCountPending(const UConverter* cnv, UErrorCode* status) |
2849 | { |
2850 | if(status == nullptr || U_FAILURE(*status)){ |
2851 | return -1; |
2852 | } |
2853 | if(cnv == nullptr){ |
2854 | *status = U_ILLEGAL_ARGUMENT_ERROR; |
2855 | return -1; |
2856 | } |
2857 | |
2858 | if(cnv->preFromUFirstCP >= 0){ |
2859 | return U16_LENGTH(cnv->preFromUFirstCP)+cnv->preFromULength ; |
2860 | }else if(cnv->preFromULength < 0){ |
2861 | return -cnv->preFromULength ; |
2862 | }else if(cnv->fromUChar32 > 0){ |
2863 | return 1; |
2864 | } |
2865 | return 0; |
2866 | |
2867 | } |
2868 | |
2869 | U_CAPI int32_t U_EXPORT2 |
2870 | ucnv_toUCountPending(const UConverter* cnv, UErrorCode* status){ |
2871 | |
2872 | if(status == nullptr || U_FAILURE(*status)){ |
2873 | return -1; |
2874 | } |
2875 | if(cnv == nullptr){ |
2876 | *status = U_ILLEGAL_ARGUMENT_ERROR; |
2877 | return -1; |
2878 | } |
2879 | |
2880 | if(cnv->preToULength > 0){ |
2881 | return cnv->preToULength ; |
2882 | }else if(cnv->preToULength < 0){ |
2883 | return -cnv->preToULength; |
2884 | }else if(cnv->toULength > 0){ |
2885 | return cnv->toULength; |
2886 | } |
2887 | return 0; |
2888 | } |
2889 | |
2890 | U_CAPI UBool U_EXPORT2 |
2891 | ucnv_isFixedWidth(UConverter *cnv, UErrorCode *status){ |
2892 | if (U_FAILURE(*status)) { |
2893 | return false; |
2894 | } |
2895 | |
2896 | if (cnv == nullptr) { |
2897 | *status = U_ILLEGAL_ARGUMENT_ERROR; |
2898 | return false; |
2899 | } |
2900 | |
2901 | switch (ucnv_getType(cnv)) { |
2902 | case UCNV_SBCS: |
2903 | case UCNV_DBCS: |
2904 | case UCNV_UTF32_BigEndian: |
2905 | case UCNV_UTF32_LittleEndian: |
2906 | case UCNV_UTF32: |
2907 | case UCNV_US_ASCII: |
2908 | return true; |
2909 | default: |
2910 | return false; |
2911 | } |
2912 | } |
2913 | #endif |
2914 | |
2915 | /* |
2916 | * Hey, Emacs, please set the following: |
2917 | * |
2918 | * Local Variables: |
2919 | * indent-tabs-mode: nil |
2920 | * End: |
2921 | * |
2922 | */ |
2923 | |