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) 2001-2016, International Business Machines |
7 | * Corporation and others. All Rights Reserved. |
8 | * |
9 | ****************************************************************************** |
10 | * |
11 | * File ustrtrns.cpp |
12 | * |
13 | * Modification History: |
14 | * |
15 | * Date Name Description |
16 | * 9/10/2001 Ram Creation. |
17 | ****************************************************************************** |
18 | */ |
19 | |
20 | /******************************************************************************* |
21 | * |
22 | * u_strTo* and u_strFrom* APIs |
23 | * WCS functions moved to ustr_wcs.c for better modularization |
24 | * |
25 | ******************************************************************************* |
26 | */ |
27 | |
28 | |
29 | #include "unicode/putil.h" |
30 | #include "unicode/ustring.h" |
31 | #include "unicode/utf.h" |
32 | #include "unicode/utf8.h" |
33 | #include "unicode/utf16.h" |
34 | #include "cstring.h" |
35 | #include "cmemory.h" |
36 | #include "ustr_imp.h" |
37 | #include "uassert.h" |
38 | |
39 | U_CAPI char16_t* U_EXPORT2 |
40 | u_strFromUTF32WithSub(char16_t *dest, |
41 | int32_t destCapacity, |
42 | int32_t *pDestLength, |
43 | const UChar32 *src, |
44 | int32_t srcLength, |
45 | UChar32 subchar, int32_t *pNumSubstitutions, |
46 | UErrorCode *pErrorCode) { |
47 | const UChar32 *srcLimit; |
48 | UChar32 ch; |
49 | char16_t *destLimit; |
50 | char16_t *pDest; |
51 | int32_t reqLength; |
52 | int32_t numSubstitutions; |
53 | |
54 | /* args check */ |
55 | if(U_FAILURE(*pErrorCode)){ |
56 | return nullptr; |
57 | } |
58 | if( (src==nullptr && srcLength!=0) || srcLength < -1 || |
59 | (destCapacity<0) || (dest == nullptr && destCapacity > 0) || |
60 | subchar > 0x10ffff || U_IS_SURROGATE(subchar) |
61 | ) { |
62 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
63 | return nullptr; |
64 | } |
65 | |
66 | if(pNumSubstitutions != nullptr) { |
67 | *pNumSubstitutions = 0; |
68 | } |
69 | |
70 | pDest = dest; |
71 | destLimit = (dest!=nullptr)?(dest + destCapacity):nullptr; |
72 | reqLength = 0; |
73 | numSubstitutions = 0; |
74 | |
75 | if(srcLength < 0) { |
76 | /* simple loop for conversion of a NUL-terminated BMP string */ |
77 | while((ch=*src) != 0 && |
78 | ((uint32_t)ch < 0xd800 || (0xe000 <= ch && ch <= 0xffff))) { |
79 | ++src; |
80 | if(pDest < destLimit) { |
81 | *pDest++ = (char16_t)ch; |
82 | } else { |
83 | ++reqLength; |
84 | } |
85 | } |
86 | srcLimit = src; |
87 | if(ch != 0) { |
88 | /* "complicated" case, find the end of the remaining string */ |
89 | while(*++srcLimit != 0) {} |
90 | } |
91 | } else { |
92 | srcLimit = (src!=nullptr)?(src + srcLength):nullptr; |
93 | } |
94 | |
95 | /* convert with length */ |
96 | while(src < srcLimit) { |
97 | ch = *src++; |
98 | do { |
99 | /* usually "loops" once; twice only for writing subchar */ |
100 | if((uint32_t)ch < 0xd800 || (0xe000 <= ch && ch <= 0xffff)) { |
101 | if(pDest < destLimit) { |
102 | *pDest++ = (char16_t)ch; |
103 | } else { |
104 | ++reqLength; |
105 | } |
106 | break; |
107 | } else if(0x10000 <= ch && ch <= 0x10ffff) { |
108 | if(pDest!=nullptr && ((pDest + 2) <= destLimit)) { |
109 | *pDest++ = U16_LEAD(ch); |
110 | *pDest++ = U16_TRAIL(ch); |
111 | } else { |
112 | reqLength += 2; |
113 | } |
114 | break; |
115 | } else if((ch = subchar) < 0) { |
116 | /* surrogate code point, or not a Unicode code point at all */ |
117 | *pErrorCode = U_INVALID_CHAR_FOUND; |
118 | return nullptr; |
119 | } else { |
120 | ++numSubstitutions; |
121 | } |
122 | } while(true); |
123 | } |
124 | |
125 | reqLength += (int32_t)(pDest - dest); |
126 | if(pDestLength) { |
127 | *pDestLength = reqLength; |
128 | } |
129 | if(pNumSubstitutions != nullptr) { |
130 | *pNumSubstitutions = numSubstitutions; |
131 | } |
132 | |
133 | /* Terminate the buffer */ |
134 | u_terminateUChars(dest, destCapacity, reqLength, pErrorCode); |
135 | |
136 | return dest; |
137 | } |
138 | |
139 | U_CAPI char16_t* U_EXPORT2 |
140 | u_strFromUTF32(char16_t *dest, |
141 | int32_t destCapacity, |
142 | int32_t *pDestLength, |
143 | const UChar32 *src, |
144 | int32_t srcLength, |
145 | UErrorCode *pErrorCode) { |
146 | return u_strFromUTF32WithSub( |
147 | dest, destCapacity, pDestLength, |
148 | src, srcLength, |
149 | U_SENTINEL, nullptr, |
150 | pErrorCode); |
151 | } |
152 | |
153 | U_CAPI UChar32* U_EXPORT2 |
154 | u_strToUTF32WithSub(UChar32 *dest, |
155 | int32_t destCapacity, |
156 | int32_t *pDestLength, |
157 | const char16_t *src, |
158 | int32_t srcLength, |
159 | UChar32 subchar, int32_t *pNumSubstitutions, |
160 | UErrorCode *pErrorCode) { |
161 | const char16_t *srcLimit; |
162 | UChar32 ch; |
163 | char16_t ch2; |
164 | UChar32 *destLimit; |
165 | UChar32 *pDest; |
166 | int32_t reqLength; |
167 | int32_t numSubstitutions; |
168 | |
169 | /* args check */ |
170 | if(U_FAILURE(*pErrorCode)){ |
171 | return nullptr; |
172 | } |
173 | if( (src==nullptr && srcLength!=0) || srcLength < -1 || |
174 | (destCapacity<0) || (dest == nullptr && destCapacity > 0) || |
175 | subchar > 0x10ffff || U_IS_SURROGATE(subchar) |
176 | ) { |
177 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
178 | return nullptr; |
179 | } |
180 | |
181 | if(pNumSubstitutions != nullptr) { |
182 | *pNumSubstitutions = 0; |
183 | } |
184 | |
185 | pDest = dest; |
186 | destLimit = (dest!=nullptr)?(dest + destCapacity):nullptr; |
187 | reqLength = 0; |
188 | numSubstitutions = 0; |
189 | |
190 | if(srcLength < 0) { |
191 | /* simple loop for conversion of a NUL-terminated BMP string */ |
192 | while((ch=*src) != 0 && !U16_IS_SURROGATE(ch)) { |
193 | ++src; |
194 | if(pDest < destLimit) { |
195 | *pDest++ = ch; |
196 | } else { |
197 | ++reqLength; |
198 | } |
199 | } |
200 | srcLimit = src; |
201 | if(ch != 0) { |
202 | /* "complicated" case, find the end of the remaining string */ |
203 | while(*++srcLimit != 0) {} |
204 | } |
205 | } else { |
206 | srcLimit = (src!=nullptr)?(src + srcLength):nullptr; |
207 | } |
208 | |
209 | /* convert with length */ |
210 | while(src < srcLimit) { |
211 | ch = *src++; |
212 | if(!U16_IS_SURROGATE(ch)) { |
213 | /* write or count ch below */ |
214 | } else if(U16_IS_SURROGATE_LEAD(ch) && src < srcLimit && U16_IS_TRAIL(ch2 = *src)) { |
215 | ++src; |
216 | ch = U16_GET_SUPPLEMENTARY(ch, ch2); |
217 | } else if((ch = subchar) < 0) { |
218 | /* unpaired surrogate */ |
219 | *pErrorCode = U_INVALID_CHAR_FOUND; |
220 | return nullptr; |
221 | } else { |
222 | ++numSubstitutions; |
223 | } |
224 | if(pDest < destLimit) { |
225 | *pDest++ = ch; |
226 | } else { |
227 | ++reqLength; |
228 | } |
229 | } |
230 | |
231 | reqLength += (int32_t)(pDest - dest); |
232 | if(pDestLength) { |
233 | *pDestLength = reqLength; |
234 | } |
235 | if(pNumSubstitutions != nullptr) { |
236 | *pNumSubstitutions = numSubstitutions; |
237 | } |
238 | |
239 | /* Terminate the buffer */ |
240 | u_terminateUChar32s(dest, destCapacity, reqLength, pErrorCode); |
241 | |
242 | return dest; |
243 | } |
244 | |
245 | U_CAPI UChar32* U_EXPORT2 |
246 | u_strToUTF32(UChar32 *dest, |
247 | int32_t destCapacity, |
248 | int32_t *pDestLength, |
249 | const char16_t *src, |
250 | int32_t srcLength, |
251 | UErrorCode *pErrorCode) { |
252 | return u_strToUTF32WithSub( |
253 | dest, destCapacity, pDestLength, |
254 | src, srcLength, |
255 | U_SENTINEL, nullptr, |
256 | pErrorCode); |
257 | } |
258 | |
259 | U_CAPI char16_t* U_EXPORT2 |
260 | u_strFromUTF8WithSub(char16_t *dest, |
261 | int32_t destCapacity, |
262 | int32_t *pDestLength, |
263 | const char* src, |
264 | int32_t srcLength, |
265 | UChar32 subchar, int32_t *pNumSubstitutions, |
266 | UErrorCode *pErrorCode){ |
267 | /* args check */ |
268 | if(U_FAILURE(*pErrorCode)) { |
269 | return nullptr; |
270 | } |
271 | if( (src==nullptr && srcLength!=0) || srcLength < -1 || |
272 | (destCapacity<0) || (dest == nullptr && destCapacity > 0) || |
273 | subchar > 0x10ffff || U_IS_SURROGATE(subchar) |
274 | ) { |
275 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
276 | return nullptr; |
277 | } |
278 | |
279 | if(pNumSubstitutions!=nullptr) { |
280 | *pNumSubstitutions=0; |
281 | } |
282 | char16_t *pDest = dest; |
283 | char16_t *pDestLimit = dest+destCapacity; |
284 | int32_t reqLength = 0; |
285 | int32_t numSubstitutions=0; |
286 | |
287 | /* |
288 | * Inline processing of UTF-8 byte sequences: |
289 | * |
290 | * Byte sequences for the most common characters are handled inline in |
291 | * the conversion loops. In order to reduce the path lengths for those |
292 | * characters, the tests are arranged in a kind of binary search. |
293 | * ASCII (<=0x7f) is checked first, followed by the dividing point |
294 | * between 2- and 3-byte sequences (0xe0). |
295 | * The 3-byte branch is tested first to speed up CJK text. |
296 | * The compiler should combine the subtractions for the two tests for 0xe0. |
297 | * Each branch then tests for the other end of its range. |
298 | */ |
299 | |
300 | if(srcLength < 0){ |
301 | /* |
302 | * Transform a NUL-terminated string. |
303 | * The code explicitly checks for NULs only in the lead byte position. |
304 | * A NUL byte in the trail byte position fails the trail byte range check anyway. |
305 | */ |
306 | int32_t i; |
307 | UChar32 c; |
308 | for(i = 0; (c = (uint8_t)src[i]) != 0 && (pDest < pDestLimit);) { |
309 | // modified copy of U8_NEXT() |
310 | ++i; |
311 | if(U8_IS_SINGLE(c)) { |
312 | *pDest++=(char16_t)c; |
313 | } else { |
314 | uint8_t __t1, __t2; |
315 | if( /* handle U+0800..U+FFFF inline */ |
316 | (0xe0<=(c) && (c)<0xf0) && |
317 | U8_IS_VALID_LEAD3_AND_T1((c), src[i]) && |
318 | (__t2=src[(i)+1]-0x80)<=0x3f) { |
319 | *pDest++ = (((c)&0xf)<<12)|((src[i]&0x3f)<<6)|__t2; |
320 | i+=2; |
321 | } else if( /* handle U+0080..U+07FF inline */ |
322 | ((c)<0xe0 && (c)>=0xc2) && |
323 | (__t1=src[i]-0x80)<=0x3f) { |
324 | *pDest++ = (((c)&0x1f)<<6)|__t1; |
325 | ++(i); |
326 | } else { |
327 | /* function call for "complicated" and error cases */ |
328 | (c)=utf8_nextCharSafeBody((const uint8_t *)src, &(i), -1, c, -1); |
329 | if(c<0 && (++numSubstitutions, c = subchar) < 0) { |
330 | *pErrorCode = U_INVALID_CHAR_FOUND; |
331 | return nullptr; |
332 | } else if(c<=0xFFFF) { |
333 | *(pDest++)=(char16_t)c; |
334 | } else { |
335 | *(pDest++)=U16_LEAD(c); |
336 | if(pDest<pDestLimit) { |
337 | *(pDest++)=U16_TRAIL(c); |
338 | } else { |
339 | reqLength++; |
340 | break; |
341 | } |
342 | } |
343 | } |
344 | } |
345 | } |
346 | |
347 | /* Pre-flight the rest of the string. */ |
348 | while((c = (uint8_t)src[i]) != 0) { |
349 | // modified copy of U8_NEXT() |
350 | ++i; |
351 | if(U8_IS_SINGLE(c)) { |
352 | ++reqLength; |
353 | } else { |
354 | uint8_t __t1, __t2; |
355 | if( /* handle U+0800..U+FFFF inline */ |
356 | (0xe0<=(c) && (c)<0xf0) && |
357 | U8_IS_VALID_LEAD3_AND_T1((c), src[i]) && |
358 | (__t2=src[(i)+1]-0x80)<=0x3f) { |
359 | ++reqLength; |
360 | i+=2; |
361 | } else if( /* handle U+0080..U+07FF inline */ |
362 | ((c)<0xe0 && (c)>=0xc2) && |
363 | (__t1=src[i]-0x80)<=0x3f) { |
364 | ++reqLength; |
365 | ++(i); |
366 | } else { |
367 | /* function call for "complicated" and error cases */ |
368 | (c)=utf8_nextCharSafeBody((const uint8_t *)src, &(i), -1, c, -1); |
369 | if(c<0 && (++numSubstitutions, c = subchar) < 0) { |
370 | *pErrorCode = U_INVALID_CHAR_FOUND; |
371 | return nullptr; |
372 | } |
373 | reqLength += U16_LENGTH(c); |
374 | } |
375 | } |
376 | } |
377 | } else /* srcLength >= 0 */ { |
378 | /* Faster loop without ongoing checking for srcLength and pDestLimit. */ |
379 | int32_t i = 0; |
380 | UChar32 c; |
381 | for(;;) { |
382 | /* |
383 | * Each iteration of the inner loop progresses by at most 3 UTF-8 |
384 | * bytes and one char16_t, for most characters. |
385 | * For supplementary code points (4 & 2), which are rare, |
386 | * there is an additional adjustment. |
387 | */ |
388 | int32_t count = (int32_t)(pDestLimit - pDest); |
389 | int32_t count2 = (srcLength - i) / 3; |
390 | if(count > count2) { |
391 | count = count2; /* min(remaining dest, remaining src/3) */ |
392 | } |
393 | if(count < 3) { |
394 | /* |
395 | * Too much overhead if we get near the end of the string, |
396 | * continue with the next loop. |
397 | */ |
398 | break; |
399 | } |
400 | |
401 | do { |
402 | // modified copy of U8_NEXT() |
403 | c = (uint8_t)src[i++]; |
404 | if(U8_IS_SINGLE(c)) { |
405 | *pDest++=(char16_t)c; |
406 | } else { |
407 | uint8_t __t1, __t2; |
408 | if( /* handle U+0800..U+FFFF inline */ |
409 | (0xe0<=(c) && (c)<0xf0) && |
410 | ((i)+1)<srcLength && |
411 | U8_IS_VALID_LEAD3_AND_T1((c), src[i]) && |
412 | (__t2=src[(i)+1]-0x80)<=0x3f) { |
413 | *pDest++ = (((c)&0xf)<<12)|((src[i]&0x3f)<<6)|__t2; |
414 | i+=2; |
415 | } else if( /* handle U+0080..U+07FF inline */ |
416 | ((c)<0xe0 && (c)>=0xc2) && |
417 | ((i)!=srcLength) && |
418 | (__t1=src[i]-0x80)<=0x3f) { |
419 | *pDest++ = (((c)&0x1f)<<6)|__t1; |
420 | ++(i); |
421 | } else { |
422 | if(c >= 0xf0 || subchar > 0xffff) { |
423 | // We may read up to four bytes and write up to two UChars, |
424 | // which we didn't account for with computing count, |
425 | // so we adjust it here. |
426 | if(--count == 0) { |
427 | --i; // back out byte c |
428 | break; |
429 | } |
430 | } |
431 | |
432 | /* function call for "complicated" and error cases */ |
433 | (c)=utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, c, -1); |
434 | if(c<0 && (++numSubstitutions, c = subchar) < 0) { |
435 | *pErrorCode = U_INVALID_CHAR_FOUND; |
436 | return nullptr; |
437 | } else if(c<=0xFFFF) { |
438 | *(pDest++)=(char16_t)c; |
439 | } else { |
440 | *(pDest++)=U16_LEAD(c); |
441 | *(pDest++)=U16_TRAIL(c); |
442 | } |
443 | } |
444 | } |
445 | } while(--count > 0); |
446 | } |
447 | |
448 | while(i < srcLength && (pDest < pDestLimit)) { |
449 | // modified copy of U8_NEXT() |
450 | c = (uint8_t)src[i++]; |
451 | if(U8_IS_SINGLE(c)) { |
452 | *pDest++=(char16_t)c; |
453 | } else { |
454 | uint8_t __t1, __t2; |
455 | if( /* handle U+0800..U+FFFF inline */ |
456 | (0xe0<=(c) && (c)<0xf0) && |
457 | ((i)+1)<srcLength && |
458 | U8_IS_VALID_LEAD3_AND_T1((c), src[i]) && |
459 | (__t2=src[(i)+1]-0x80)<=0x3f) { |
460 | *pDest++ = (((c)&0xf)<<12)|((src[i]&0x3f)<<6)|__t2; |
461 | i+=2; |
462 | } else if( /* handle U+0080..U+07FF inline */ |
463 | ((c)<0xe0 && (c)>=0xc2) && |
464 | ((i)!=srcLength) && |
465 | (__t1=src[i]-0x80)<=0x3f) { |
466 | *pDest++ = (((c)&0x1f)<<6)|__t1; |
467 | ++(i); |
468 | } else { |
469 | /* function call for "complicated" and error cases */ |
470 | (c)=utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, c, -1); |
471 | if(c<0 && (++numSubstitutions, c = subchar) < 0) { |
472 | *pErrorCode = U_INVALID_CHAR_FOUND; |
473 | return nullptr; |
474 | } else if(c<=0xFFFF) { |
475 | *(pDest++)=(char16_t)c; |
476 | } else { |
477 | *(pDest++)=U16_LEAD(c); |
478 | if(pDest<pDestLimit) { |
479 | *(pDest++)=U16_TRAIL(c); |
480 | } else { |
481 | reqLength++; |
482 | break; |
483 | } |
484 | } |
485 | } |
486 | } |
487 | } |
488 | |
489 | /* Pre-flight the rest of the string. */ |
490 | while(i < srcLength) { |
491 | // modified copy of U8_NEXT() |
492 | c = (uint8_t)src[i++]; |
493 | if(U8_IS_SINGLE(c)) { |
494 | ++reqLength; |
495 | } else { |
496 | uint8_t __t1, __t2; |
497 | if( /* handle U+0800..U+FFFF inline */ |
498 | (0xe0<=(c) && (c)<0xf0) && |
499 | ((i)+1)<srcLength && |
500 | U8_IS_VALID_LEAD3_AND_T1((c), src[i]) && |
501 | (__t2=src[(i)+1]-0x80)<=0x3f) { |
502 | ++reqLength; |
503 | i+=2; |
504 | } else if( /* handle U+0080..U+07FF inline */ |
505 | ((c)<0xe0 && (c)>=0xc2) && |
506 | ((i)!=srcLength) && |
507 | (__t1=src[i]-0x80)<=0x3f) { |
508 | ++reqLength; |
509 | ++(i); |
510 | } else { |
511 | /* function call for "complicated" and error cases */ |
512 | (c)=utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, c, -1); |
513 | if(c<0 && (++numSubstitutions, c = subchar) < 0) { |
514 | *pErrorCode = U_INVALID_CHAR_FOUND; |
515 | return nullptr; |
516 | } |
517 | reqLength += U16_LENGTH(c); |
518 | } |
519 | } |
520 | } |
521 | } |
522 | |
523 | reqLength+=(int32_t)(pDest - dest); |
524 | |
525 | if(pNumSubstitutions!=nullptr) { |
526 | *pNumSubstitutions=numSubstitutions; |
527 | } |
528 | |
529 | if(pDestLength){ |
530 | *pDestLength = reqLength; |
531 | } |
532 | |
533 | /* Terminate the buffer */ |
534 | u_terminateUChars(dest,destCapacity,reqLength,pErrorCode); |
535 | |
536 | return dest; |
537 | } |
538 | |
539 | U_CAPI char16_t* U_EXPORT2 |
540 | u_strFromUTF8(char16_t *dest, |
541 | int32_t destCapacity, |
542 | int32_t *pDestLength, |
543 | const char* src, |
544 | int32_t srcLength, |
545 | UErrorCode *pErrorCode){ |
546 | return u_strFromUTF8WithSub( |
547 | dest, destCapacity, pDestLength, |
548 | src, srcLength, |
549 | U_SENTINEL, nullptr, |
550 | pErrorCode); |
551 | } |
552 | |
553 | U_CAPI char16_t * U_EXPORT2 |
554 | u_strFromUTF8Lenient(char16_t *dest, |
555 | int32_t destCapacity, |
556 | int32_t *pDestLength, |
557 | const char *src, |
558 | int32_t srcLength, |
559 | UErrorCode *pErrorCode) { |
560 | char16_t *pDest = dest; |
561 | UChar32 ch; |
562 | int32_t reqLength = 0; |
563 | uint8_t* pSrc = (uint8_t*) src; |
564 | |
565 | /* args check */ |
566 | if(U_FAILURE(*pErrorCode)){ |
567 | return nullptr; |
568 | } |
569 | |
570 | if( (src==nullptr && srcLength!=0) || srcLength < -1 || |
571 | (destCapacity<0) || (dest == nullptr && destCapacity > 0) |
572 | ) { |
573 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
574 | return nullptr; |
575 | } |
576 | |
577 | if(srcLength < 0) { |
578 | /* Transform a NUL-terminated string. */ |
579 | char16_t *pDestLimit = (dest!=nullptr)?(dest+destCapacity):nullptr; |
580 | uint8_t t1, t2, t3; /* trail bytes */ |
581 | |
582 | while(((ch = *pSrc) != 0) && (pDest < pDestLimit)) { |
583 | if(ch < 0xc0) { |
584 | /* |
585 | * ASCII, or a trail byte in lead position which is treated like |
586 | * a single-byte sequence for better character boundary |
587 | * resynchronization after illegal sequences. |
588 | */ |
589 | *pDest++=(char16_t)ch; |
590 | ++pSrc; |
591 | continue; |
592 | } else if(ch < 0xe0) { /* U+0080..U+07FF */ |
593 | if((t1 = pSrc[1]) != 0) { |
594 | /* 0x3080 = (0xc0 << 6) + 0x80 */ |
595 | *pDest++ = (char16_t)((ch << 6) + t1 - 0x3080); |
596 | pSrc += 2; |
597 | continue; |
598 | } |
599 | } else if(ch < 0xf0) { /* U+0800..U+FFFF */ |
600 | if((t1 = pSrc[1]) != 0 && (t2 = pSrc[2]) != 0) { |
601 | /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) */ |
602 | /* 0x2080 = (0x80 << 6) + 0x80 */ |
603 | *pDest++ = (char16_t)((ch << 12) + (t1 << 6) + t2 - 0x2080); |
604 | pSrc += 3; |
605 | continue; |
606 | } |
607 | } else /* f0..f4 */ { /* U+10000..U+10FFFF */ |
608 | if((t1 = pSrc[1]) != 0 && (t2 = pSrc[2]) != 0 && (t3 = pSrc[3]) != 0) { |
609 | pSrc += 4; |
610 | /* 0x3c82080 = (0xf0 << 18) + (0x80 << 12) + (0x80 << 6) + 0x80 */ |
611 | ch = (ch << 18) + (t1 << 12) + (t2 << 6) + t3 - 0x3c82080; |
612 | *(pDest++) = U16_LEAD(ch); |
613 | if(pDest < pDestLimit) { |
614 | *(pDest++) = U16_TRAIL(ch); |
615 | } else { |
616 | reqLength = 1; |
617 | break; |
618 | } |
619 | continue; |
620 | } |
621 | } |
622 | |
623 | /* truncated character at the end */ |
624 | *pDest++ = 0xfffd; |
625 | while(*++pSrc != 0) {} |
626 | break; |
627 | } |
628 | |
629 | /* Pre-flight the rest of the string. */ |
630 | while((ch = *pSrc) != 0) { |
631 | if(ch < 0xc0) { |
632 | /* |
633 | * ASCII, or a trail byte in lead position which is treated like |
634 | * a single-byte sequence for better character boundary |
635 | * resynchronization after illegal sequences. |
636 | */ |
637 | ++reqLength; |
638 | ++pSrc; |
639 | continue; |
640 | } else if(ch < 0xe0) { /* U+0080..U+07FF */ |
641 | if(pSrc[1] != 0) { |
642 | ++reqLength; |
643 | pSrc += 2; |
644 | continue; |
645 | } |
646 | } else if(ch < 0xf0) { /* U+0800..U+FFFF */ |
647 | if(pSrc[1] != 0 && pSrc[2] != 0) { |
648 | ++reqLength; |
649 | pSrc += 3; |
650 | continue; |
651 | } |
652 | } else /* f0..f4 */ { /* U+10000..U+10FFFF */ |
653 | if(pSrc[1] != 0 && pSrc[2] != 0 && pSrc[3] != 0) { |
654 | reqLength += 2; |
655 | pSrc += 4; |
656 | continue; |
657 | } |
658 | } |
659 | |
660 | /* truncated character at the end */ |
661 | ++reqLength; |
662 | break; |
663 | } |
664 | } else /* srcLength >= 0 */ { |
665 | const uint8_t *pSrcLimit = (pSrc!=nullptr)?(pSrc + srcLength):nullptr; |
666 | |
667 | /* |
668 | * This function requires that if srcLength is given, then it must be |
669 | * destCapatity >= srcLength so that we need not check for |
670 | * destination buffer overflow in the loop. |
671 | */ |
672 | if(destCapacity < srcLength) { |
673 | if(pDestLength != nullptr) { |
674 | *pDestLength = srcLength; /* this likely overestimates the true destLength! */ |
675 | } |
676 | *pErrorCode = U_BUFFER_OVERFLOW_ERROR; |
677 | return nullptr; |
678 | } |
679 | |
680 | if((pSrcLimit - pSrc) >= 4) { |
681 | pSrcLimit -= 3; /* temporarily reduce pSrcLimit */ |
682 | |
683 | /* in this loop, we can always access at least 4 bytes, up to pSrc+3 */ |
684 | do { |
685 | ch = *pSrc++; |
686 | if(ch < 0xc0) { |
687 | /* |
688 | * ASCII, or a trail byte in lead position which is treated like |
689 | * a single-byte sequence for better character boundary |
690 | * resynchronization after illegal sequences. |
691 | */ |
692 | *pDest++=(char16_t)ch; |
693 | } else if(ch < 0xe0) { /* U+0080..U+07FF */ |
694 | /* 0x3080 = (0xc0 << 6) + 0x80 */ |
695 | *pDest++ = (char16_t)((ch << 6) + *pSrc++ - 0x3080); |
696 | } else if(ch < 0xf0) { /* U+0800..U+FFFF */ |
697 | /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) */ |
698 | /* 0x2080 = (0x80 << 6) + 0x80 */ |
699 | ch = (ch << 12) + (*pSrc++ << 6); |
700 | *pDest++ = (char16_t)(ch + *pSrc++ - 0x2080); |
701 | } else /* f0..f4 */ { /* U+10000..U+10FFFF */ |
702 | /* 0x3c82080 = (0xf0 << 18) + (0x80 << 12) + (0x80 << 6) + 0x80 */ |
703 | ch = (ch << 18) + (*pSrc++ << 12); |
704 | ch += *pSrc++ << 6; |
705 | ch += *pSrc++ - 0x3c82080; |
706 | *(pDest++) = U16_LEAD(ch); |
707 | *(pDest++) = U16_TRAIL(ch); |
708 | } |
709 | } while(pSrc < pSrcLimit); |
710 | |
711 | pSrcLimit += 3; /* restore original pSrcLimit */ |
712 | } |
713 | |
714 | while(pSrc < pSrcLimit) { |
715 | ch = *pSrc++; |
716 | if(ch < 0xc0) { |
717 | /* |
718 | * ASCII, or a trail byte in lead position which is treated like |
719 | * a single-byte sequence for better character boundary |
720 | * resynchronization after illegal sequences. |
721 | */ |
722 | *pDest++=(char16_t)ch; |
723 | continue; |
724 | } else if(ch < 0xe0) { /* U+0080..U+07FF */ |
725 | if(pSrc < pSrcLimit) { |
726 | /* 0x3080 = (0xc0 << 6) + 0x80 */ |
727 | *pDest++ = (char16_t)((ch << 6) + *pSrc++ - 0x3080); |
728 | continue; |
729 | } |
730 | } else if(ch < 0xf0) { /* U+0800..U+FFFF */ |
731 | if((pSrcLimit - pSrc) >= 2) { |
732 | /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) */ |
733 | /* 0x2080 = (0x80 << 6) + 0x80 */ |
734 | ch = (ch << 12) + (*pSrc++ << 6); |
735 | *pDest++ = (char16_t)(ch + *pSrc++ - 0x2080); |
736 | pSrc += 3; |
737 | continue; |
738 | } |
739 | } else /* f0..f4 */ { /* U+10000..U+10FFFF */ |
740 | if((pSrcLimit - pSrc) >= 3) { |
741 | /* 0x3c82080 = (0xf0 << 18) + (0x80 << 12) + (0x80 << 6) + 0x80 */ |
742 | ch = (ch << 18) + (*pSrc++ << 12); |
743 | ch += *pSrc++ << 6; |
744 | ch += *pSrc++ - 0x3c82080; |
745 | *(pDest++) = U16_LEAD(ch); |
746 | *(pDest++) = U16_TRAIL(ch); |
747 | pSrc += 4; |
748 | continue; |
749 | } |
750 | } |
751 | |
752 | /* truncated character at the end */ |
753 | *pDest++ = 0xfffd; |
754 | break; |
755 | } |
756 | } |
757 | |
758 | reqLength+=(int32_t)(pDest - dest); |
759 | |
760 | if(pDestLength){ |
761 | *pDestLength = reqLength; |
762 | } |
763 | |
764 | /* Terminate the buffer */ |
765 | u_terminateUChars(dest,destCapacity,reqLength,pErrorCode); |
766 | |
767 | return dest; |
768 | } |
769 | |
770 | static inline uint8_t * |
771 | _appendUTF8(uint8_t *pDest, UChar32 c) { |
772 | /* it is 0<=c<=0x10ffff and not a surrogate if called by a validating function */ |
773 | if((c)<=0x7f) { |
774 | *pDest++=(uint8_t)c; |
775 | } else if(c<=0x7ff) { |
776 | *pDest++=(uint8_t)((c>>6)|0xc0); |
777 | *pDest++=(uint8_t)((c&0x3f)|0x80); |
778 | } else if(c<=0xffff) { |
779 | *pDest++=(uint8_t)((c>>12)|0xe0); |
780 | *pDest++=(uint8_t)(((c>>6)&0x3f)|0x80); |
781 | *pDest++=(uint8_t)(((c)&0x3f)|0x80); |
782 | } else /* if((uint32_t)(c)<=0x10ffff) */ { |
783 | *pDest++=(uint8_t)(((c)>>18)|0xf0); |
784 | *pDest++=(uint8_t)((((c)>>12)&0x3f)|0x80); |
785 | *pDest++=(uint8_t)((((c)>>6)&0x3f)|0x80); |
786 | *pDest++=(uint8_t)(((c)&0x3f)|0x80); |
787 | } |
788 | return pDest; |
789 | } |
790 | |
791 | |
792 | U_CAPI char* U_EXPORT2 |
793 | u_strToUTF8WithSub(char *dest, |
794 | int32_t destCapacity, |
795 | int32_t *pDestLength, |
796 | const char16_t *pSrc, |
797 | int32_t srcLength, |
798 | UChar32 subchar, int32_t *pNumSubstitutions, |
799 | UErrorCode *pErrorCode){ |
800 | int32_t reqLength=0; |
801 | uint32_t ch=0,ch2=0; |
802 | uint8_t *pDest = (uint8_t *)dest; |
803 | uint8_t *pDestLimit = (pDest!=nullptr)?(pDest + destCapacity):nullptr; |
804 | int32_t numSubstitutions; |
805 | |
806 | /* args check */ |
807 | if(U_FAILURE(*pErrorCode)){ |
808 | return nullptr; |
809 | } |
810 | |
811 | if( (pSrc==nullptr && srcLength!=0) || srcLength < -1 || |
812 | (destCapacity<0) || (dest == nullptr && destCapacity > 0) || |
813 | subchar > 0x10ffff || U_IS_SURROGATE(subchar) |
814 | ) { |
815 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
816 | return nullptr; |
817 | } |
818 | |
819 | if(pNumSubstitutions!=nullptr) { |
820 | *pNumSubstitutions=0; |
821 | } |
822 | numSubstitutions=0; |
823 | |
824 | if(srcLength==-1) { |
825 | while((ch=*pSrc)!=0) { |
826 | ++pSrc; |
827 | if(ch <= 0x7f) { |
828 | if(pDest<pDestLimit) { |
829 | *pDest++ = (uint8_t)ch; |
830 | } else { |
831 | reqLength = 1; |
832 | break; |
833 | } |
834 | } else if(ch <= 0x7ff) { |
835 | if((pDestLimit - pDest) >= 2) { |
836 | *pDest++=(uint8_t)((ch>>6)|0xc0); |
837 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
838 | } else { |
839 | reqLength = 2; |
840 | break; |
841 | } |
842 | } else if(ch <= 0xd7ff || ch >= 0xe000) { |
843 | if((pDestLimit - pDest) >= 3) { |
844 | *pDest++=(uint8_t)((ch>>12)|0xe0); |
845 | *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80); |
846 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
847 | } else { |
848 | reqLength = 3; |
849 | break; |
850 | } |
851 | } else /* ch is a surrogate */ { |
852 | int32_t length; |
853 | |
854 | /*need not check for NUL because NUL fails U16_IS_TRAIL() anyway*/ |
855 | if(U16_IS_SURROGATE_LEAD(ch) && U16_IS_TRAIL(ch2=*pSrc)) { |
856 | ++pSrc; |
857 | ch=U16_GET_SUPPLEMENTARY(ch, ch2); |
858 | } else if(subchar>=0) { |
859 | ch=subchar; |
860 | ++numSubstitutions; |
861 | } else { |
862 | /* Unicode 3.2 forbids surrogate code points in UTF-8 */ |
863 | *pErrorCode = U_INVALID_CHAR_FOUND; |
864 | return nullptr; |
865 | } |
866 | |
867 | length = U8_LENGTH(ch); |
868 | if((pDestLimit - pDest) >= length) { |
869 | /* convert and append*/ |
870 | pDest=_appendUTF8(pDest, ch); |
871 | } else { |
872 | reqLength = length; |
873 | break; |
874 | } |
875 | } |
876 | } |
877 | while((ch=*pSrc++)!=0) { |
878 | if(ch<=0x7f) { |
879 | ++reqLength; |
880 | } else if(ch<=0x7ff) { |
881 | reqLength+=2; |
882 | } else if(!U16_IS_SURROGATE(ch)) { |
883 | reqLength+=3; |
884 | } else if(U16_IS_SURROGATE_LEAD(ch) && U16_IS_TRAIL(ch2=*pSrc)) { |
885 | ++pSrc; |
886 | reqLength+=4; |
887 | } else if(subchar>=0) { |
888 | reqLength+=U8_LENGTH(subchar); |
889 | ++numSubstitutions; |
890 | } else { |
891 | /* Unicode 3.2 forbids surrogate code points in UTF-8 */ |
892 | *pErrorCode = U_INVALID_CHAR_FOUND; |
893 | return nullptr; |
894 | } |
895 | } |
896 | } else { |
897 | const char16_t *pSrcLimit = (pSrc!=nullptr)?(pSrc+srcLength):nullptr; |
898 | int32_t count; |
899 | |
900 | /* Faster loop without ongoing checking for pSrcLimit and pDestLimit. */ |
901 | for(;;) { |
902 | /* |
903 | * Each iteration of the inner loop progresses by at most 3 UTF-8 |
904 | * bytes and one char16_t, for most characters. |
905 | * For supplementary code points (4 & 2), which are rare, |
906 | * there is an additional adjustment. |
907 | */ |
908 | count = (int32_t)((pDestLimit - pDest) / 3); |
909 | srcLength = (int32_t)(pSrcLimit - pSrc); |
910 | if(count > srcLength) { |
911 | count = srcLength; /* min(remaining dest/3, remaining src) */ |
912 | } |
913 | if(count < 3) { |
914 | /* |
915 | * Too much overhead if we get near the end of the string, |
916 | * continue with the next loop. |
917 | */ |
918 | break; |
919 | } |
920 | do { |
921 | ch=*pSrc++; |
922 | if(ch <= 0x7f) { |
923 | *pDest++ = (uint8_t)ch; |
924 | } else if(ch <= 0x7ff) { |
925 | *pDest++=(uint8_t)((ch>>6)|0xc0); |
926 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
927 | } else if(ch <= 0xd7ff || ch >= 0xe000) { |
928 | *pDest++=(uint8_t)((ch>>12)|0xe0); |
929 | *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80); |
930 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
931 | } else /* ch is a surrogate */ { |
932 | /* |
933 | * We will read two UChars and probably output four bytes, |
934 | * which we didn't account for with computing count, |
935 | * so we adjust it here. |
936 | */ |
937 | if(--count == 0) { |
938 | --pSrc; /* undo ch=*pSrc++ for the lead surrogate */ |
939 | break; /* recompute count */ |
940 | } |
941 | |
942 | if(U16_IS_SURROGATE_LEAD(ch) && U16_IS_TRAIL(ch2=*pSrc)) { |
943 | ++pSrc; |
944 | ch=U16_GET_SUPPLEMENTARY(ch, ch2); |
945 | |
946 | /* writing 4 bytes per 2 UChars is ok */ |
947 | *pDest++=(uint8_t)((ch>>18)|0xf0); |
948 | *pDest++=(uint8_t)(((ch>>12)&0x3f)|0x80); |
949 | *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80); |
950 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
951 | } else { |
952 | /* Unicode 3.2 forbids surrogate code points in UTF-8 */ |
953 | if(subchar>=0) { |
954 | ch=subchar; |
955 | ++numSubstitutions; |
956 | } else { |
957 | *pErrorCode = U_INVALID_CHAR_FOUND; |
958 | return nullptr; |
959 | } |
960 | |
961 | /* convert and append*/ |
962 | pDest=_appendUTF8(pDest, ch); |
963 | } |
964 | } |
965 | } while(--count > 0); |
966 | } |
967 | |
968 | while(pSrc<pSrcLimit) { |
969 | ch=*pSrc++; |
970 | if(ch <= 0x7f) { |
971 | if(pDest<pDestLimit) { |
972 | *pDest++ = (uint8_t)ch; |
973 | } else { |
974 | reqLength = 1; |
975 | break; |
976 | } |
977 | } else if(ch <= 0x7ff) { |
978 | if((pDestLimit - pDest) >= 2) { |
979 | *pDest++=(uint8_t)((ch>>6)|0xc0); |
980 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
981 | } else { |
982 | reqLength = 2; |
983 | break; |
984 | } |
985 | } else if(ch <= 0xd7ff || ch >= 0xe000) { |
986 | if((pDestLimit - pDest) >= 3) { |
987 | *pDest++=(uint8_t)((ch>>12)|0xe0); |
988 | *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80); |
989 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
990 | } else { |
991 | reqLength = 3; |
992 | break; |
993 | } |
994 | } else /* ch is a surrogate */ { |
995 | int32_t length; |
996 | |
997 | if(U16_IS_SURROGATE_LEAD(ch) && pSrc<pSrcLimit && U16_IS_TRAIL(ch2=*pSrc)) { |
998 | ++pSrc; |
999 | ch=U16_GET_SUPPLEMENTARY(ch, ch2); |
1000 | } else if(subchar>=0) { |
1001 | ch=subchar; |
1002 | ++numSubstitutions; |
1003 | } else { |
1004 | /* Unicode 3.2 forbids surrogate code points in UTF-8 */ |
1005 | *pErrorCode = U_INVALID_CHAR_FOUND; |
1006 | return nullptr; |
1007 | } |
1008 | |
1009 | length = U8_LENGTH(ch); |
1010 | if((pDestLimit - pDest) >= length) { |
1011 | /* convert and append*/ |
1012 | pDest=_appendUTF8(pDest, ch); |
1013 | } else { |
1014 | reqLength = length; |
1015 | break; |
1016 | } |
1017 | } |
1018 | } |
1019 | while(pSrc<pSrcLimit) { |
1020 | ch=*pSrc++; |
1021 | if(ch<=0x7f) { |
1022 | ++reqLength; |
1023 | } else if(ch<=0x7ff) { |
1024 | reqLength+=2; |
1025 | } else if(!U16_IS_SURROGATE(ch)) { |
1026 | reqLength+=3; |
1027 | } else if(U16_IS_SURROGATE_LEAD(ch) && pSrc<pSrcLimit && U16_IS_TRAIL(ch2=*pSrc)) { |
1028 | ++pSrc; |
1029 | reqLength+=4; |
1030 | } else if(subchar>=0) { |
1031 | reqLength+=U8_LENGTH(subchar); |
1032 | ++numSubstitutions; |
1033 | } else { |
1034 | /* Unicode 3.2 forbids surrogate code points in UTF-8 */ |
1035 | *pErrorCode = U_INVALID_CHAR_FOUND; |
1036 | return nullptr; |
1037 | } |
1038 | } |
1039 | } |
1040 | |
1041 | reqLength+=(int32_t)(pDest - (uint8_t *)dest); |
1042 | |
1043 | if(pNumSubstitutions!=nullptr) { |
1044 | *pNumSubstitutions=numSubstitutions; |
1045 | } |
1046 | |
1047 | if(pDestLength){ |
1048 | *pDestLength = reqLength; |
1049 | } |
1050 | |
1051 | /* Terminate the buffer */ |
1052 | u_terminateChars(dest, destCapacity, reqLength, pErrorCode); |
1053 | return dest; |
1054 | } |
1055 | |
1056 | U_CAPI char* U_EXPORT2 |
1057 | u_strToUTF8(char *dest, |
1058 | int32_t destCapacity, |
1059 | int32_t *pDestLength, |
1060 | const char16_t *pSrc, |
1061 | int32_t srcLength, |
1062 | UErrorCode *pErrorCode){ |
1063 | return u_strToUTF8WithSub( |
1064 | dest, destCapacity, pDestLength, |
1065 | pSrc, srcLength, |
1066 | U_SENTINEL, nullptr, |
1067 | pErrorCode); |
1068 | } |
1069 | |
1070 | U_CAPI char16_t* U_EXPORT2 |
1071 | u_strFromJavaModifiedUTF8WithSub( |
1072 | char16_t *dest, |
1073 | int32_t destCapacity, |
1074 | int32_t *pDestLength, |
1075 | const char *src, |
1076 | int32_t srcLength, |
1077 | UChar32 subchar, int32_t *pNumSubstitutions, |
1078 | UErrorCode *pErrorCode) { |
1079 | /* args check */ |
1080 | if(U_FAILURE(*pErrorCode)) { |
1081 | return nullptr; |
1082 | } |
1083 | if( (src==nullptr && srcLength!=0) || srcLength < -1 || |
1084 | (dest==nullptr && destCapacity!=0) || destCapacity<0 || |
1085 | subchar > 0x10ffff || U_IS_SURROGATE(subchar) |
1086 | ) { |
1087 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
1088 | return nullptr; |
1089 | } |
1090 | |
1091 | if(pNumSubstitutions!=nullptr) { |
1092 | *pNumSubstitutions=0; |
1093 | } |
1094 | char16_t *pDest = dest; |
1095 | char16_t *pDestLimit = dest+destCapacity; |
1096 | int32_t reqLength = 0; |
1097 | int32_t numSubstitutions=0; |
1098 | |
1099 | if(srcLength < 0) { |
1100 | /* |
1101 | * Transform a NUL-terminated ASCII string. |
1102 | * Handle non-ASCII strings with slower code. |
1103 | */ |
1104 | UChar32 c; |
1105 | while(((c = (uint8_t)*src) != 0) && c <= 0x7f && (pDest < pDestLimit)) { |
1106 | *pDest++=(char16_t)c; |
1107 | ++src; |
1108 | } |
1109 | if(c == 0) { |
1110 | reqLength=(int32_t)(pDest - dest); |
1111 | if(pDestLength) { |
1112 | *pDestLength = reqLength; |
1113 | } |
1114 | |
1115 | /* Terminate the buffer */ |
1116 | u_terminateUChars(dest, destCapacity, reqLength, pErrorCode); |
1117 | return dest; |
1118 | } |
1119 | srcLength = static_cast<int32_t>(uprv_strlen(src)); |
1120 | } |
1121 | |
1122 | /* Faster loop without ongoing checking for srcLength and pDestLimit. */ |
1123 | UChar32 ch; |
1124 | uint8_t t1, t2; |
1125 | int32_t i = 0; |
1126 | for(;;) { |
1127 | int32_t count = (int32_t)(pDestLimit - pDest); |
1128 | int32_t count2 = srcLength - i; |
1129 | if(count >= count2 && srcLength > 0 && U8_IS_SINGLE(*src)) { |
1130 | /* fast ASCII loop */ |
1131 | int32_t start = i; |
1132 | uint8_t b; |
1133 | while(i < srcLength && U8_IS_SINGLE(b = src[i])) { |
1134 | *pDest++=b; |
1135 | ++i; |
1136 | } |
1137 | int32_t delta = i - start; |
1138 | count -= delta; |
1139 | count2 -= delta; |
1140 | } |
1141 | /* |
1142 | * Each iteration of the inner loop progresses by at most 3 UTF-8 |
1143 | * bytes and one char16_t. |
1144 | */ |
1145 | if(subchar > 0xFFFF) { |
1146 | break; |
1147 | } |
1148 | count2 /= 3; |
1149 | if(count > count2) { |
1150 | count = count2; /* min(remaining dest, remaining src/3) */ |
1151 | } |
1152 | if(count < 3) { |
1153 | /* |
1154 | * Too much overhead if we get near the end of the string, |
1155 | * continue with the next loop. |
1156 | */ |
1157 | break; |
1158 | } |
1159 | do { |
1160 | ch = (uint8_t)src[i++]; |
1161 | if(U8_IS_SINGLE(ch)) { |
1162 | *pDest++=(char16_t)ch; |
1163 | } else { |
1164 | if(ch >= 0xe0) { |
1165 | if( /* handle U+0000..U+FFFF inline */ |
1166 | ch <= 0xef && |
1167 | (t1 = (uint8_t)(src[i] - 0x80)) <= 0x3f && |
1168 | (t2 = (uint8_t)(src[i+1] - 0x80)) <= 0x3f |
1169 | ) { |
1170 | /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) */ |
1171 | *pDest++ = (char16_t)((ch << 12) | (t1 << 6) | t2); |
1172 | i += 2; |
1173 | continue; |
1174 | } |
1175 | } else { |
1176 | if( /* handle U+0000..U+07FF inline */ |
1177 | ch >= 0xc0 && |
1178 | (t1 = (uint8_t)(src[i] - 0x80)) <= 0x3f |
1179 | ) { |
1180 | *pDest++ = (char16_t)(((ch & 0x1f) << 6) | t1); |
1181 | ++i; |
1182 | continue; |
1183 | } |
1184 | } |
1185 | |
1186 | if(subchar < 0) { |
1187 | *pErrorCode = U_INVALID_CHAR_FOUND; |
1188 | return nullptr; |
1189 | } else if(subchar > 0xffff && --count == 0) { |
1190 | /* |
1191 | * We need to write two UChars, adjusted count for that, |
1192 | * and ran out of space. |
1193 | */ |
1194 | --i; // back out byte ch |
1195 | break; |
1196 | } else { |
1197 | /* function call for error cases */ |
1198 | utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, ch, -1); |
1199 | ++numSubstitutions; |
1200 | *(pDest++)=(char16_t)subchar; |
1201 | } |
1202 | } |
1203 | } while(--count > 0); |
1204 | } |
1205 | |
1206 | while(i < srcLength && (pDest < pDestLimit)) { |
1207 | ch = (uint8_t)src[i++]; |
1208 | if(U8_IS_SINGLE(ch)){ |
1209 | *pDest++=(char16_t)ch; |
1210 | } else { |
1211 | if(ch >= 0xe0) { |
1212 | if( /* handle U+0000..U+FFFF inline */ |
1213 | ch <= 0xef && |
1214 | (i+1) < srcLength && |
1215 | (t1 = (uint8_t)(src[i] - 0x80)) <= 0x3f && |
1216 | (t2 = (uint8_t)(src[i+1] - 0x80)) <= 0x3f |
1217 | ) { |
1218 | /* no need for (ch & 0xf) because the upper bits are truncated after <<12 in the cast to (char16_t) */ |
1219 | *pDest++ = (char16_t)((ch << 12) | (t1 << 6) | t2); |
1220 | i += 2; |
1221 | continue; |
1222 | } |
1223 | } else { |
1224 | if( /* handle U+0000..U+07FF inline */ |
1225 | ch >= 0xc0 && |
1226 | i < srcLength && |
1227 | (t1 = (uint8_t)(src[i] - 0x80)) <= 0x3f |
1228 | ) { |
1229 | *pDest++ = (char16_t)(((ch & 0x1f) << 6) | t1); |
1230 | ++i; |
1231 | continue; |
1232 | } |
1233 | } |
1234 | |
1235 | if(subchar < 0) { |
1236 | *pErrorCode = U_INVALID_CHAR_FOUND; |
1237 | return nullptr; |
1238 | } else { |
1239 | /* function call for error cases */ |
1240 | utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, ch, -1); |
1241 | ++numSubstitutions; |
1242 | if(subchar<=0xFFFF) { |
1243 | *(pDest++)=(char16_t)subchar; |
1244 | } else { |
1245 | *(pDest++)=U16_LEAD(subchar); |
1246 | if(pDest<pDestLimit) { |
1247 | *(pDest++)=U16_TRAIL(subchar); |
1248 | } else { |
1249 | reqLength++; |
1250 | break; |
1251 | } |
1252 | } |
1253 | } |
1254 | } |
1255 | } |
1256 | |
1257 | /* Pre-flight the rest of the string. */ |
1258 | while(i < srcLength) { |
1259 | ch = (uint8_t)src[i++]; |
1260 | if(U8_IS_SINGLE(ch)) { |
1261 | reqLength++; |
1262 | } else { |
1263 | if(ch >= 0xe0) { |
1264 | if( /* handle U+0000..U+FFFF inline */ |
1265 | ch <= 0xef && |
1266 | (i+1) < srcLength && |
1267 | (uint8_t)(src[i] - 0x80) <= 0x3f && |
1268 | (uint8_t)(src[i+1] - 0x80) <= 0x3f |
1269 | ) { |
1270 | reqLength++; |
1271 | i += 2; |
1272 | continue; |
1273 | } |
1274 | } else { |
1275 | if( /* handle U+0000..U+07FF inline */ |
1276 | ch >= 0xc0 && |
1277 | i < srcLength && |
1278 | (uint8_t)(src[i] - 0x80) <= 0x3f |
1279 | ) { |
1280 | reqLength++; |
1281 | ++i; |
1282 | continue; |
1283 | } |
1284 | } |
1285 | |
1286 | if(subchar < 0) { |
1287 | *pErrorCode = U_INVALID_CHAR_FOUND; |
1288 | return nullptr; |
1289 | } else { |
1290 | /* function call for error cases */ |
1291 | utf8_nextCharSafeBody((const uint8_t *)src, &(i), srcLength, ch, -1); |
1292 | ++numSubstitutions; |
1293 | reqLength+=U16_LENGTH(ch); |
1294 | } |
1295 | } |
1296 | } |
1297 | |
1298 | if(pNumSubstitutions!=nullptr) { |
1299 | *pNumSubstitutions=numSubstitutions; |
1300 | } |
1301 | |
1302 | reqLength+=(int32_t)(pDest - dest); |
1303 | if(pDestLength) { |
1304 | *pDestLength = reqLength; |
1305 | } |
1306 | |
1307 | /* Terminate the buffer */ |
1308 | u_terminateUChars(dest, destCapacity, reqLength, pErrorCode); |
1309 | return dest; |
1310 | } |
1311 | |
1312 | U_CAPI char* U_EXPORT2 |
1313 | u_strToJavaModifiedUTF8( |
1314 | char *dest, |
1315 | int32_t destCapacity, |
1316 | int32_t *pDestLength, |
1317 | const char16_t *src, |
1318 | int32_t srcLength, |
1319 | UErrorCode *pErrorCode) { |
1320 | int32_t reqLength=0; |
1321 | uint32_t ch=0; |
1322 | const char16_t *pSrcLimit; |
1323 | int32_t count; |
1324 | |
1325 | /* args check */ |
1326 | if(U_FAILURE(*pErrorCode)){ |
1327 | return nullptr; |
1328 | } |
1329 | if( (src==nullptr && srcLength!=0) || srcLength < -1 || |
1330 | (dest==nullptr && destCapacity!=0) || destCapacity<0 |
1331 | ) { |
1332 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
1333 | return nullptr; |
1334 | } |
1335 | uint8_t *pDest = (uint8_t *)dest; |
1336 | uint8_t *pDestLimit = pDest + destCapacity; |
1337 | |
1338 | if(srcLength==-1) { |
1339 | /* Convert NUL-terminated ASCII, then find the string length. */ |
1340 | while((ch=*src)<=0x7f && ch != 0 && pDest<pDestLimit) { |
1341 | *pDest++ = (uint8_t)ch; |
1342 | ++src; |
1343 | } |
1344 | if(ch == 0) { |
1345 | reqLength=(int32_t)(pDest - (uint8_t *)dest); |
1346 | if(pDestLength) { |
1347 | *pDestLength = reqLength; |
1348 | } |
1349 | |
1350 | /* Terminate the buffer */ |
1351 | u_terminateChars(dest, destCapacity, reqLength, pErrorCode); |
1352 | return dest; |
1353 | } |
1354 | srcLength = u_strlen(src); |
1355 | } |
1356 | |
1357 | /* Faster loop without ongoing checking for pSrcLimit and pDestLimit. */ |
1358 | pSrcLimit = (src!=nullptr)?(src+srcLength):nullptr; |
1359 | for(;;) { |
1360 | count = (int32_t)(pDestLimit - pDest); |
1361 | srcLength = (int32_t)(pSrcLimit - src); |
1362 | if(count >= srcLength && srcLength > 0 && *src <= 0x7f) { |
1363 | /* fast ASCII loop */ |
1364 | const char16_t *prevSrc = src; |
1365 | int32_t delta; |
1366 | while(src < pSrcLimit && (ch = *src) <= 0x7f && ch != 0) { |
1367 | *pDest++=(uint8_t)ch; |
1368 | ++src; |
1369 | } |
1370 | delta = (int32_t)(src - prevSrc); |
1371 | count -= delta; |
1372 | srcLength -= delta; |
1373 | } |
1374 | /* |
1375 | * Each iteration of the inner loop progresses by at most 3 UTF-8 |
1376 | * bytes and one char16_t. |
1377 | */ |
1378 | count /= 3; |
1379 | if(count > srcLength) { |
1380 | count = srcLength; /* min(remaining dest/3, remaining src) */ |
1381 | } |
1382 | if(count < 3) { |
1383 | /* |
1384 | * Too much overhead if we get near the end of the string, |
1385 | * continue with the next loop. |
1386 | */ |
1387 | break; |
1388 | } |
1389 | do { |
1390 | ch=*src++; |
1391 | if(ch <= 0x7f && ch != 0) { |
1392 | *pDest++ = (uint8_t)ch; |
1393 | } else if(ch <= 0x7ff) { |
1394 | *pDest++=(uint8_t)((ch>>6)|0xc0); |
1395 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
1396 | } else { |
1397 | *pDest++=(uint8_t)((ch>>12)|0xe0); |
1398 | *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80); |
1399 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
1400 | } |
1401 | } while(--count > 0); |
1402 | } |
1403 | |
1404 | while(src<pSrcLimit) { |
1405 | ch=*src++; |
1406 | if(ch <= 0x7f && ch != 0) { |
1407 | if(pDest<pDestLimit) { |
1408 | *pDest++ = (uint8_t)ch; |
1409 | } else { |
1410 | reqLength = 1; |
1411 | break; |
1412 | } |
1413 | } else if(ch <= 0x7ff) { |
1414 | if((pDestLimit - pDest) >= 2) { |
1415 | *pDest++=(uint8_t)((ch>>6)|0xc0); |
1416 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
1417 | } else { |
1418 | reqLength = 2; |
1419 | break; |
1420 | } |
1421 | } else { |
1422 | if((pDestLimit - pDest) >= 3) { |
1423 | *pDest++=(uint8_t)((ch>>12)|0xe0); |
1424 | *pDest++=(uint8_t)(((ch>>6)&0x3f)|0x80); |
1425 | *pDest++=(uint8_t)((ch&0x3f)|0x80); |
1426 | } else { |
1427 | reqLength = 3; |
1428 | break; |
1429 | } |
1430 | } |
1431 | } |
1432 | while(src<pSrcLimit) { |
1433 | ch=*src++; |
1434 | if(ch <= 0x7f && ch != 0) { |
1435 | ++reqLength; |
1436 | } else if(ch<=0x7ff) { |
1437 | reqLength+=2; |
1438 | } else { |
1439 | reqLength+=3; |
1440 | } |
1441 | } |
1442 | |
1443 | reqLength+=(int32_t)(pDest - (uint8_t *)dest); |
1444 | if(pDestLength){ |
1445 | *pDestLength = reqLength; |
1446 | } |
1447 | |
1448 | /* Terminate the buffer */ |
1449 | u_terminateChars(dest, destCapacity, reqLength, pErrorCode); |
1450 | return dest; |
1451 | } |
1452 | |