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