1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | ********************************************************************** |
5 | * Copyright (C) 2002-2015, International Business Machines |
6 | * Corporation and others. All Rights Reserved. |
7 | ********************************************************************** |
8 | * file name: ucnv_u16.c |
9 | * encoding: UTF-8 |
10 | * tab size: 8 (not used) |
11 | * indentation:4 |
12 | * |
13 | * created on: 2002jul01 |
14 | * created by: Markus W. Scherer |
15 | * |
16 | * UTF-16 converter implementation. Used to be in ucnv_utf.c. |
17 | */ |
18 | |
19 | #include "unicode/utypes.h" |
20 | |
21 | #if !UCONFIG_NO_CONVERSION |
22 | |
23 | #include "unicode/ucnv.h" |
24 | #include "unicode/uversion.h" |
25 | #include "ucnv_bld.h" |
26 | #include "ucnv_cnv.h" |
27 | #include "cmemory.h" |
28 | |
29 | enum { |
30 | UCNV_NEED_TO_WRITE_BOM=1 |
31 | }; |
32 | |
33 | U_CDECL_BEGIN |
34 | /* |
35 | * The UTF-16 toUnicode implementation is also used for the Java-specific |
36 | * "with BOM" variants of UTF-16BE and UTF-16LE. |
37 | */ |
38 | static void U_CALLCONV |
39 | _UTF16ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, |
40 | UErrorCode *pErrorCode); |
41 | |
42 | /* UTF-16BE ----------------------------------------------------------------- */ |
43 | |
44 | #if U_IS_BIG_ENDIAN |
45 | # define _UTF16PEFromUnicodeWithOffsets _UTF16BEFromUnicodeWithOffsets |
46 | #else |
47 | # define _UTF16PEFromUnicodeWithOffsets _UTF16LEFromUnicodeWithOffsets |
48 | #endif |
49 | |
50 | |
51 | static void U_CALLCONV |
52 | _UTF16BEFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, |
53 | UErrorCode *pErrorCode) { |
54 | UConverter *cnv; |
55 | const char16_t *source; |
56 | char *target; |
57 | int32_t *offsets; |
58 | |
59 | uint32_t targetCapacity, length, sourceIndex; |
60 | char16_t c, trail; |
61 | char overflow[4]; |
62 | |
63 | source=pArgs->source; |
64 | length=(int32_t)(pArgs->sourceLimit-source); |
65 | if(length<=0) { |
66 | /* no input, nothing to do */ |
67 | return; |
68 | } |
69 | |
70 | cnv=pArgs->converter; |
71 | |
72 | /* write the BOM if necessary */ |
73 | if(cnv->fromUnicodeStatus==UCNV_NEED_TO_WRITE_BOM) { |
74 | static const char bom[]={ (char)0xfeu, (char)0xffu }; |
75 | ucnv_fromUWriteBytes(cnv, |
76 | bom, 2, |
77 | &pArgs->target, pArgs->targetLimit, |
78 | &pArgs->offsets, -1, |
79 | pErrorCode); |
80 | cnv->fromUnicodeStatus=0; |
81 | } |
82 | |
83 | target=pArgs->target; |
84 | if(target >= pArgs->targetLimit) { |
85 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
86 | return; |
87 | } |
88 | |
89 | targetCapacity=(uint32_t)(pArgs->targetLimit-target); |
90 | offsets=pArgs->offsets; |
91 | sourceIndex=0; |
92 | |
93 | /* c!=0 indicates in several places outside the main loops that a surrogate was found */ |
94 | |
95 | if((c=(char16_t)cnv->fromUChar32)!=0 && U16_IS_TRAIL(trail=*source) && targetCapacity>=4) { |
96 | /* the last buffer ended with a lead surrogate, output the surrogate pair */ |
97 | ++source; |
98 | --length; |
99 | target[0]=(uint8_t)(c>>8); |
100 | target[1]=(uint8_t)c; |
101 | target[2]=(uint8_t)(trail>>8); |
102 | target[3]=(uint8_t)trail; |
103 | target+=4; |
104 | targetCapacity-=4; |
105 | if(offsets!=nullptr) { |
106 | *offsets++=-1; |
107 | *offsets++=-1; |
108 | *offsets++=-1; |
109 | *offsets++=-1; |
110 | } |
111 | sourceIndex=1; |
112 | cnv->fromUChar32=c=0; |
113 | } |
114 | |
115 | if(c==0) { |
116 | /* copy an even number of bytes for complete UChars */ |
117 | uint32_t count=2*length; |
118 | if(count>targetCapacity) { |
119 | count=targetCapacity&~1; |
120 | } |
121 | /* count is even */ |
122 | targetCapacity-=count; |
123 | count>>=1; |
124 | length-=count; |
125 | |
126 | if(offsets==nullptr) { |
127 | while(count>0) { |
128 | c=*source++; |
129 | if(U16_IS_SINGLE(c)) { |
130 | target[0]=(uint8_t)(c>>8); |
131 | target[1]=(uint8_t)c; |
132 | target+=2; |
133 | } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && U16_IS_TRAIL(trail=*source)) { |
134 | ++source; |
135 | --count; |
136 | target[0]=(uint8_t)(c>>8); |
137 | target[1]=(uint8_t)c; |
138 | target[2]=(uint8_t)(trail>>8); |
139 | target[3]=(uint8_t)trail; |
140 | target+=4; |
141 | } else { |
142 | break; |
143 | } |
144 | --count; |
145 | } |
146 | } else { |
147 | while(count>0) { |
148 | c=*source++; |
149 | if(U16_IS_SINGLE(c)) { |
150 | target[0]=(uint8_t)(c>>8); |
151 | target[1]=(uint8_t)c; |
152 | target+=2; |
153 | *offsets++=sourceIndex; |
154 | *offsets++=sourceIndex++; |
155 | } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && U16_IS_TRAIL(trail=*source)) { |
156 | ++source; |
157 | --count; |
158 | target[0]=(uint8_t)(c>>8); |
159 | target[1]=(uint8_t)c; |
160 | target[2]=(uint8_t)(trail>>8); |
161 | target[3]=(uint8_t)trail; |
162 | target+=4; |
163 | *offsets++=sourceIndex; |
164 | *offsets++=sourceIndex; |
165 | *offsets++=sourceIndex; |
166 | *offsets++=sourceIndex; |
167 | sourceIndex+=2; |
168 | } else { |
169 | break; |
170 | } |
171 | --count; |
172 | } |
173 | } |
174 | |
175 | if(count==0) { |
176 | /* done with the loop for complete UChars */ |
177 | if(length>0 && targetCapacity>0) { |
178 | /* |
179 | * there is more input and some target capacity - |
180 | * it must be targetCapacity==1 because otherwise |
181 | * the above would have copied more; |
182 | * prepare for overflow output |
183 | */ |
184 | if(U16_IS_SINGLE(c=*source++)) { |
185 | overflow[0]=(char)(c>>8); |
186 | overflow[1]=(char)c; |
187 | length=2; /* 2 bytes to output */ |
188 | c=0; |
189 | /* } else { keep c for surrogate handling, length will be set there */ |
190 | } |
191 | } else { |
192 | length=0; |
193 | c=0; |
194 | } |
195 | } else { |
196 | /* keep c for surrogate handling, length will be set there */ |
197 | targetCapacity+=2*count; |
198 | } |
199 | } else { |
200 | length=0; /* from here on, length counts the bytes in overflow[] */ |
201 | } |
202 | |
203 | if(c!=0) { |
204 | /* |
205 | * c is a surrogate, and |
206 | * - source or target too short |
207 | * - or the surrogate is unmatched |
208 | */ |
209 | length=0; |
210 | if(U16_IS_SURROGATE_LEAD(c)) { |
211 | if(source<pArgs->sourceLimit) { |
212 | if(U16_IS_TRAIL(trail=*source)) { |
213 | /* output the surrogate pair, will overflow (see conditions comment above) */ |
214 | ++source; |
215 | overflow[0]=(char)(c>>8); |
216 | overflow[1]=(char)c; |
217 | overflow[2]=(char)(trail>>8); |
218 | overflow[3]=(char)trail; |
219 | length=4; /* 4 bytes to output */ |
220 | c=0; |
221 | } else { |
222 | /* unmatched lead surrogate */ |
223 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
224 | } |
225 | } else { |
226 | /* see if the trail surrogate is in the next buffer */ |
227 | } |
228 | } else { |
229 | /* unmatched trail surrogate */ |
230 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
231 | } |
232 | cnv->fromUChar32=c; |
233 | } |
234 | |
235 | if(length>0) { |
236 | /* output length bytes with overflow (length>targetCapacity>0) */ |
237 | ucnv_fromUWriteBytes(cnv, |
238 | overflow, length, |
239 | (char **)&target, pArgs->targetLimit, |
240 | &offsets, sourceIndex, |
241 | pErrorCode); |
242 | targetCapacity=(uint32_t)(pArgs->targetLimit-(char *)target); |
243 | } |
244 | |
245 | if(U_SUCCESS(*pErrorCode) && source<pArgs->sourceLimit && targetCapacity==0) { |
246 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
247 | } |
248 | |
249 | /* write back the updated pointers */ |
250 | pArgs->source=source; |
251 | pArgs->target=(char *)target; |
252 | pArgs->offsets=offsets; |
253 | } |
254 | |
255 | static void U_CALLCONV |
256 | _UTF16BEToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, |
257 | UErrorCode *pErrorCode) { |
258 | UConverter *cnv; |
259 | const uint8_t *source; |
260 | char16_t *target; |
261 | int32_t *offsets; |
262 | |
263 | uint32_t targetCapacity, length, count, sourceIndex; |
264 | char16_t c, trail; |
265 | |
266 | if(pArgs->converter->mode<8) { |
267 | _UTF16ToUnicodeWithOffsets(pArgs, pErrorCode); |
268 | return; |
269 | } |
270 | |
271 | cnv=pArgs->converter; |
272 | source=(const uint8_t *)pArgs->source; |
273 | length=(int32_t)((const uint8_t *)pArgs->sourceLimit-source); |
274 | if(length<=0 && cnv->toUnicodeStatus==0) { |
275 | /* no input, nothing to do */ |
276 | return; |
277 | } |
278 | |
279 | target=pArgs->target; |
280 | if(target >= pArgs->targetLimit) { |
281 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
282 | return; |
283 | } |
284 | |
285 | targetCapacity=(uint32_t)(pArgs->targetLimit-target); |
286 | offsets=pArgs->offsets; |
287 | sourceIndex=0; |
288 | c=0; |
289 | |
290 | /* complete a partial char16_t or pair from the last call */ |
291 | if(cnv->toUnicodeStatus!=0) { |
292 | /* |
293 | * special case: single byte from a previous buffer, |
294 | * where the byte turned out not to belong to a trail surrogate |
295 | * and the preceding, unmatched lead surrogate was put into toUBytes[] |
296 | * for error handling |
297 | */ |
298 | cnv->toUBytes[0]=(uint8_t)cnv->toUnicodeStatus; |
299 | cnv->toULength=1; |
300 | cnv->toUnicodeStatus=0; |
301 | } |
302 | if((count=cnv->toULength)!=0) { |
303 | uint8_t *p=cnv->toUBytes; |
304 | do { |
305 | p[count++]=*source++; |
306 | ++sourceIndex; |
307 | --length; |
308 | if(count==2) { |
309 | c=((char16_t)p[0]<<8)|p[1]; |
310 | if(U16_IS_SINGLE(c)) { |
311 | /* output the BMP code point */ |
312 | *target++=c; |
313 | if(offsets!=nullptr) { |
314 | *offsets++=-1; |
315 | } |
316 | --targetCapacity; |
317 | count=0; |
318 | c=0; |
319 | break; |
320 | } else if(U16_IS_SURROGATE_LEAD(c)) { |
321 | /* continue collecting bytes for the trail surrogate */ |
322 | c=0; /* avoid unnecessary surrogate handling below */ |
323 | } else { |
324 | /* fall through to error handling for an unmatched trail surrogate */ |
325 | break; |
326 | } |
327 | } else if(count==4) { |
328 | c=((char16_t)p[0]<<8)|p[1]; |
329 | trail=((char16_t)p[2]<<8)|p[3]; |
330 | if(U16_IS_TRAIL(trail)) { |
331 | /* output the surrogate pair */ |
332 | *target++=c; |
333 | if(targetCapacity>=2) { |
334 | *target++=trail; |
335 | if(offsets!=nullptr) { |
336 | *offsets++=-1; |
337 | *offsets++=-1; |
338 | } |
339 | targetCapacity-=2; |
340 | } else /* targetCapacity==1 */ { |
341 | targetCapacity=0; |
342 | cnv->UCharErrorBuffer[0]=trail; |
343 | cnv->UCharErrorBufferLength=1; |
344 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
345 | } |
346 | count=0; |
347 | c=0; |
348 | break; |
349 | } else { |
350 | /* unmatched lead surrogate, handle here for consistent toUBytes[] */ |
351 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
352 | |
353 | /* back out reading the code unit after it */ |
354 | if(((const uint8_t *)pArgs->source-source)>=2) { |
355 | source-=2; |
356 | } else { |
357 | /* |
358 | * if the trail unit's first byte was in a previous buffer, then |
359 | * we need to put it into a special place because toUBytes[] will be |
360 | * used for the lead unit's bytes |
361 | */ |
362 | cnv->toUnicodeStatus=0x100|p[2]; |
363 | --source; |
364 | } |
365 | cnv->toULength=2; |
366 | |
367 | /* write back the updated pointers */ |
368 | pArgs->source=(const char *)source; |
369 | pArgs->target=target; |
370 | pArgs->offsets=offsets; |
371 | return; |
372 | } |
373 | } |
374 | } while(length>0); |
375 | cnv->toULength=(int8_t)count; |
376 | } |
377 | |
378 | /* copy an even number of bytes for complete UChars */ |
379 | count=2*targetCapacity; |
380 | if(count>length) { |
381 | count=length&~1; |
382 | } |
383 | if(c==0 && count>0) { |
384 | length-=count; |
385 | count>>=1; |
386 | targetCapacity-=count; |
387 | if(offsets==nullptr) { |
388 | do { |
389 | c=((char16_t)source[0]<<8)|source[1]; |
390 | source+=2; |
391 | if(U16_IS_SINGLE(c)) { |
392 | *target++=c; |
393 | } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && |
394 | U16_IS_TRAIL(trail=((char16_t)source[0]<<8)|source[1]) |
395 | ) { |
396 | source+=2; |
397 | --count; |
398 | *target++=c; |
399 | *target++=trail; |
400 | } else { |
401 | break; |
402 | } |
403 | } while(--count>0); |
404 | } else { |
405 | do { |
406 | c=((char16_t)source[0]<<8)|source[1]; |
407 | source+=2; |
408 | if(U16_IS_SINGLE(c)) { |
409 | *target++=c; |
410 | *offsets++=sourceIndex; |
411 | sourceIndex+=2; |
412 | } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && |
413 | U16_IS_TRAIL(trail=((char16_t)source[0]<<8)|source[1]) |
414 | ) { |
415 | source+=2; |
416 | --count; |
417 | *target++=c; |
418 | *target++=trail; |
419 | *offsets++=sourceIndex; |
420 | *offsets++=sourceIndex; |
421 | sourceIndex+=4; |
422 | } else { |
423 | break; |
424 | } |
425 | } while(--count>0); |
426 | } |
427 | |
428 | if(count==0) { |
429 | /* done with the loop for complete UChars */ |
430 | c=0; |
431 | } else { |
432 | /* keep c for surrogate handling, trail will be set there */ |
433 | length+=2*(count-1); /* one more byte pair was consumed than count decremented */ |
434 | targetCapacity+=count; |
435 | } |
436 | } |
437 | |
438 | if(c!=0) { |
439 | /* |
440 | * c is a surrogate, and |
441 | * - source or target too short |
442 | * - or the surrogate is unmatched |
443 | */ |
444 | cnv->toUBytes[0]=(uint8_t)(c>>8); |
445 | cnv->toUBytes[1]=(uint8_t)c; |
446 | cnv->toULength=2; |
447 | |
448 | if(U16_IS_SURROGATE_LEAD(c)) { |
449 | if(length>=2) { |
450 | if(U16_IS_TRAIL(trail=((char16_t)source[0]<<8)|source[1])) { |
451 | /* output the surrogate pair, will overflow (see conditions comment above) */ |
452 | source+=2; |
453 | length-=2; |
454 | *target++=c; |
455 | if(offsets!=nullptr) { |
456 | *offsets++=sourceIndex; |
457 | } |
458 | cnv->UCharErrorBuffer[0]=trail; |
459 | cnv->UCharErrorBufferLength=1; |
460 | cnv->toULength=0; |
461 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
462 | } else { |
463 | /* unmatched lead surrogate */ |
464 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
465 | } |
466 | } else { |
467 | /* see if the trail surrogate is in the next buffer */ |
468 | } |
469 | } else { |
470 | /* unmatched trail surrogate */ |
471 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
472 | } |
473 | } |
474 | |
475 | if(U_SUCCESS(*pErrorCode)) { |
476 | /* check for a remaining source byte */ |
477 | if(length>0) { |
478 | if(targetCapacity==0) { |
479 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
480 | } else { |
481 | /* it must be length==1 because otherwise the above would have copied more */ |
482 | cnv->toUBytes[cnv->toULength++]=*source++; |
483 | } |
484 | } |
485 | } |
486 | |
487 | /* write back the updated pointers */ |
488 | pArgs->source=(const char *)source; |
489 | pArgs->target=target; |
490 | pArgs->offsets=offsets; |
491 | } |
492 | |
493 | static UChar32 U_CALLCONV |
494 | _UTF16BEGetNextUChar(UConverterToUnicodeArgs *pArgs, UErrorCode *err) { |
495 | const uint8_t *s, *sourceLimit; |
496 | UChar32 c; |
497 | |
498 | if(pArgs->converter->mode<8) { |
499 | return UCNV_GET_NEXT_UCHAR_USE_TO_U; |
500 | } |
501 | |
502 | s=(const uint8_t *)pArgs->source; |
503 | sourceLimit=(const uint8_t *)pArgs->sourceLimit; |
504 | |
505 | if(s>=sourceLimit) { |
506 | /* no input */ |
507 | *err=U_INDEX_OUTOFBOUNDS_ERROR; |
508 | return 0xffff; |
509 | } |
510 | |
511 | if(s+2>sourceLimit) { |
512 | /* only one byte: truncated char16_t */ |
513 | pArgs->converter->toUBytes[0]=*s++; |
514 | pArgs->converter->toULength=1; |
515 | pArgs->source=(const char *)s; |
516 | *err = U_TRUNCATED_CHAR_FOUND; |
517 | return 0xffff; |
518 | } |
519 | |
520 | /* get one char16_t */ |
521 | c=((UChar32)*s<<8)|s[1]; |
522 | s+=2; |
523 | |
524 | /* check for a surrogate pair */ |
525 | if(U_IS_SURROGATE(c)) { |
526 | if(U16_IS_SURROGATE_LEAD(c)) { |
527 | if(s+2<=sourceLimit) { |
528 | char16_t trail; |
529 | |
530 | /* get a second char16_t and see if it is a trail surrogate */ |
531 | trail=((char16_t)*s<<8)|s[1]; |
532 | if(U16_IS_TRAIL(trail)) { |
533 | c=U16_GET_SUPPLEMENTARY(c, trail); |
534 | s+=2; |
535 | } else { |
536 | /* unmatched lead surrogate */ |
537 | c=-2; |
538 | } |
539 | } else { |
540 | /* too few (2 or 3) bytes for a surrogate pair: truncated code point */ |
541 | uint8_t *bytes=pArgs->converter->toUBytes; |
542 | s-=2; |
543 | pArgs->converter->toULength=(int8_t)(sourceLimit-s); |
544 | do { |
545 | *bytes++=*s++; |
546 | } while(s<sourceLimit); |
547 | |
548 | c=0xffff; |
549 | *err=U_TRUNCATED_CHAR_FOUND; |
550 | } |
551 | } else { |
552 | /* unmatched trail surrogate */ |
553 | c=-2; |
554 | } |
555 | |
556 | if(c<0) { |
557 | /* write the unmatched surrogate */ |
558 | uint8_t *bytes=pArgs->converter->toUBytes; |
559 | pArgs->converter->toULength=2; |
560 | *bytes=*(s-2); |
561 | bytes[1]=*(s-1); |
562 | |
563 | c=0xffff; |
564 | *err=U_ILLEGAL_CHAR_FOUND; |
565 | } |
566 | } |
567 | |
568 | pArgs->source=(const char *)s; |
569 | return c; |
570 | } |
571 | |
572 | static void U_CALLCONV |
573 | _UTF16BEReset(UConverter *cnv, UConverterResetChoice choice) { |
574 | if(choice<=UCNV_RESET_TO_UNICODE) { |
575 | /* reset toUnicode state */ |
576 | if(UCNV_GET_VERSION(cnv)==0) { |
577 | cnv->mode=8; /* no BOM handling */ |
578 | } else { |
579 | cnv->mode=0; /* Java-specific "UnicodeBig" requires BE BOM or no BOM */ |
580 | } |
581 | } |
582 | if(choice!=UCNV_RESET_TO_UNICODE && UCNV_GET_VERSION(cnv)==1) { |
583 | /* reset fromUnicode for "UnicodeBig": prepare to output the UTF-16BE BOM */ |
584 | cnv->fromUnicodeStatus=UCNV_NEED_TO_WRITE_BOM; |
585 | } |
586 | } |
587 | |
588 | static void U_CALLCONV |
589 | _UTF16BEOpen(UConverter *cnv, |
590 | UConverterLoadArgs *pArgs, |
591 | UErrorCode *pErrorCode) { |
592 | (void)pArgs; |
593 | if(UCNV_GET_VERSION(cnv)<=1) { |
594 | _UTF16BEReset(cnv, UCNV_RESET_BOTH); |
595 | } else { |
596 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
597 | } |
598 | } |
599 | |
600 | static const char * U_CALLCONV |
601 | _UTF16BEGetName(const UConverter *cnv) { |
602 | if(UCNV_GET_VERSION(cnv)==0) { |
603 | return "UTF-16BE" ; |
604 | } else { |
605 | return "UTF-16BE,version=1" ; |
606 | } |
607 | } |
608 | U_CDECL_END |
609 | |
610 | static const UConverterImpl _UTF16BEImpl={ |
611 | UCNV_UTF16_BigEndian, |
612 | |
613 | nullptr, |
614 | nullptr, |
615 | |
616 | _UTF16BEOpen, |
617 | nullptr, |
618 | _UTF16BEReset, |
619 | |
620 | _UTF16BEToUnicodeWithOffsets, |
621 | _UTF16BEToUnicodeWithOffsets, |
622 | _UTF16BEFromUnicodeWithOffsets, |
623 | _UTF16BEFromUnicodeWithOffsets, |
624 | _UTF16BEGetNextUChar, |
625 | |
626 | nullptr, |
627 | _UTF16BEGetName, |
628 | nullptr, |
629 | nullptr, |
630 | ucnv_getNonSurrogateUnicodeSet, |
631 | |
632 | nullptr, |
633 | nullptr |
634 | }; |
635 | |
636 | static const UConverterStaticData _UTF16BEStaticData={ |
637 | sizeof(UConverterStaticData), |
638 | "UTF-16BE" , |
639 | 1200, UCNV_IBM, UCNV_UTF16_BigEndian, 2, 2, |
640 | { 0xff, 0xfd, 0, 0 },2,false,false, |
641 | 0, |
642 | 0, |
643 | { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ |
644 | }; |
645 | |
646 | |
647 | const UConverterSharedData _UTF16BEData= |
648 | UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF16BEStaticData, &_UTF16BEImpl); |
649 | |
650 | /* UTF-16LE ----------------------------------------------------------------- */ |
651 | U_CDECL_BEGIN |
652 | static void U_CALLCONV |
653 | _UTF16LEFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, |
654 | UErrorCode *pErrorCode) { |
655 | UConverter *cnv; |
656 | const char16_t *source; |
657 | char *target; |
658 | int32_t *offsets; |
659 | |
660 | uint32_t targetCapacity, length, sourceIndex; |
661 | char16_t c, trail; |
662 | char overflow[4]; |
663 | |
664 | source=pArgs->source; |
665 | length=(int32_t)(pArgs->sourceLimit-source); |
666 | if(length<=0) { |
667 | /* no input, nothing to do */ |
668 | return; |
669 | } |
670 | |
671 | cnv=pArgs->converter; |
672 | |
673 | /* write the BOM if necessary */ |
674 | if(cnv->fromUnicodeStatus==UCNV_NEED_TO_WRITE_BOM) { |
675 | static const char bom[]={ (char)0xffu, (char)0xfeu }; |
676 | ucnv_fromUWriteBytes(cnv, |
677 | bom, 2, |
678 | &pArgs->target, pArgs->targetLimit, |
679 | &pArgs->offsets, -1, |
680 | pErrorCode); |
681 | cnv->fromUnicodeStatus=0; |
682 | } |
683 | |
684 | target=pArgs->target; |
685 | if(target >= pArgs->targetLimit) { |
686 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
687 | return; |
688 | } |
689 | |
690 | targetCapacity=(uint32_t)(pArgs->targetLimit-pArgs->target); |
691 | offsets=pArgs->offsets; |
692 | sourceIndex=0; |
693 | |
694 | /* c!=0 indicates in several places outside the main loops that a surrogate was found */ |
695 | |
696 | if((c=(char16_t)cnv->fromUChar32)!=0 && U16_IS_TRAIL(trail=*source) && targetCapacity>=4) { |
697 | /* the last buffer ended with a lead surrogate, output the surrogate pair */ |
698 | ++source; |
699 | --length; |
700 | target[0]=(uint8_t)c; |
701 | target[1]=(uint8_t)(c>>8); |
702 | target[2]=(uint8_t)trail; |
703 | target[3]=(uint8_t)(trail>>8); |
704 | target+=4; |
705 | targetCapacity-=4; |
706 | if(offsets!=nullptr) { |
707 | *offsets++=-1; |
708 | *offsets++=-1; |
709 | *offsets++=-1; |
710 | *offsets++=-1; |
711 | } |
712 | sourceIndex=1; |
713 | cnv->fromUChar32=c=0; |
714 | } |
715 | |
716 | if(c==0) { |
717 | /* copy an even number of bytes for complete UChars */ |
718 | uint32_t count=2*length; |
719 | if(count>targetCapacity) { |
720 | count=targetCapacity&~1; |
721 | } |
722 | /* count is even */ |
723 | targetCapacity-=count; |
724 | count>>=1; |
725 | length-=count; |
726 | |
727 | if(offsets==nullptr) { |
728 | while(count>0) { |
729 | c=*source++; |
730 | if(U16_IS_SINGLE(c)) { |
731 | target[0]=(uint8_t)c; |
732 | target[1]=(uint8_t)(c>>8); |
733 | target+=2; |
734 | } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && U16_IS_TRAIL(trail=*source)) { |
735 | ++source; |
736 | --count; |
737 | target[0]=(uint8_t)c; |
738 | target[1]=(uint8_t)(c>>8); |
739 | target[2]=(uint8_t)trail; |
740 | target[3]=(uint8_t)(trail>>8); |
741 | target+=4; |
742 | } else { |
743 | break; |
744 | } |
745 | --count; |
746 | } |
747 | } else { |
748 | while(count>0) { |
749 | c=*source++; |
750 | if(U16_IS_SINGLE(c)) { |
751 | target[0]=(uint8_t)c; |
752 | target[1]=(uint8_t)(c>>8); |
753 | target+=2; |
754 | *offsets++=sourceIndex; |
755 | *offsets++=sourceIndex++; |
756 | } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && U16_IS_TRAIL(trail=*source)) { |
757 | ++source; |
758 | --count; |
759 | target[0]=(uint8_t)c; |
760 | target[1]=(uint8_t)(c>>8); |
761 | target[2]=(uint8_t)trail; |
762 | target[3]=(uint8_t)(trail>>8); |
763 | target+=4; |
764 | *offsets++=sourceIndex; |
765 | *offsets++=sourceIndex; |
766 | *offsets++=sourceIndex; |
767 | *offsets++=sourceIndex; |
768 | sourceIndex+=2; |
769 | } else { |
770 | break; |
771 | } |
772 | --count; |
773 | } |
774 | } |
775 | |
776 | if(count==0) { |
777 | /* done with the loop for complete UChars */ |
778 | if(length>0 && targetCapacity>0) { |
779 | /* |
780 | * there is more input and some target capacity - |
781 | * it must be targetCapacity==1 because otherwise |
782 | * the above would have copied more; |
783 | * prepare for overflow output |
784 | */ |
785 | if(U16_IS_SINGLE(c=*source++)) { |
786 | overflow[0]=(char)c; |
787 | overflow[1]=(char)(c>>8); |
788 | length=2; /* 2 bytes to output */ |
789 | c=0; |
790 | /* } else { keep c for surrogate handling, length will be set there */ |
791 | } |
792 | } else { |
793 | length=0; |
794 | c=0; |
795 | } |
796 | } else { |
797 | /* keep c for surrogate handling, length will be set there */ |
798 | targetCapacity+=2*count; |
799 | } |
800 | } else { |
801 | length=0; /* from here on, length counts the bytes in overflow[] */ |
802 | } |
803 | |
804 | if(c!=0) { |
805 | /* |
806 | * c is a surrogate, and |
807 | * - source or target too short |
808 | * - or the surrogate is unmatched |
809 | */ |
810 | length=0; |
811 | if(U16_IS_SURROGATE_LEAD(c)) { |
812 | if(source<pArgs->sourceLimit) { |
813 | if(U16_IS_TRAIL(trail=*source)) { |
814 | /* output the surrogate pair, will overflow (see conditions comment above) */ |
815 | ++source; |
816 | overflow[0]=(char)c; |
817 | overflow[1]=(char)(c>>8); |
818 | overflow[2]=(char)trail; |
819 | overflow[3]=(char)(trail>>8); |
820 | length=4; /* 4 bytes to output */ |
821 | c=0; |
822 | } else { |
823 | /* unmatched lead surrogate */ |
824 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
825 | } |
826 | } else { |
827 | /* see if the trail surrogate is in the next buffer */ |
828 | } |
829 | } else { |
830 | /* unmatched trail surrogate */ |
831 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
832 | } |
833 | cnv->fromUChar32=c; |
834 | } |
835 | |
836 | if(length>0) { |
837 | /* output length bytes with overflow (length>targetCapacity>0) */ |
838 | ucnv_fromUWriteBytes(cnv, |
839 | overflow, length, |
840 | &target, pArgs->targetLimit, |
841 | &offsets, sourceIndex, |
842 | pErrorCode); |
843 | targetCapacity=(uint32_t)(pArgs->targetLimit-(char *)target); |
844 | } |
845 | |
846 | if(U_SUCCESS(*pErrorCode) && source<pArgs->sourceLimit && targetCapacity==0) { |
847 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
848 | } |
849 | |
850 | /* write back the updated pointers */ |
851 | pArgs->source=source; |
852 | pArgs->target=target; |
853 | pArgs->offsets=offsets; |
854 | } |
855 | |
856 | static void U_CALLCONV |
857 | _UTF16LEToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, |
858 | UErrorCode *pErrorCode) { |
859 | UConverter *cnv; |
860 | const uint8_t *source; |
861 | char16_t *target; |
862 | int32_t *offsets; |
863 | |
864 | uint32_t targetCapacity, length, count, sourceIndex; |
865 | char16_t c, trail; |
866 | |
867 | if(pArgs->converter->mode<8) { |
868 | _UTF16ToUnicodeWithOffsets(pArgs, pErrorCode); |
869 | return; |
870 | } |
871 | |
872 | cnv=pArgs->converter; |
873 | source=(const uint8_t *)pArgs->source; |
874 | length=(int32_t)((const uint8_t *)pArgs->sourceLimit-source); |
875 | if(length<=0 && cnv->toUnicodeStatus==0) { |
876 | /* no input, nothing to do */ |
877 | return; |
878 | } |
879 | |
880 | target=pArgs->target; |
881 | if(target >= pArgs->targetLimit) { |
882 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
883 | return; |
884 | } |
885 | |
886 | targetCapacity=(uint32_t)(pArgs->targetLimit-pArgs->target); |
887 | offsets=pArgs->offsets; |
888 | sourceIndex=0; |
889 | c=0; |
890 | |
891 | /* complete a partial char16_t or pair from the last call */ |
892 | if(cnv->toUnicodeStatus!=0) { |
893 | /* |
894 | * special case: single byte from a previous buffer, |
895 | * where the byte turned out not to belong to a trail surrogate |
896 | * and the preceding, unmatched lead surrogate was put into toUBytes[] |
897 | * for error handling |
898 | */ |
899 | cnv->toUBytes[0]=(uint8_t)cnv->toUnicodeStatus; |
900 | cnv->toULength=1; |
901 | cnv->toUnicodeStatus=0; |
902 | } |
903 | if((count=cnv->toULength)!=0) { |
904 | uint8_t *p=cnv->toUBytes; |
905 | do { |
906 | p[count++]=*source++; |
907 | ++sourceIndex; |
908 | --length; |
909 | if(count==2) { |
910 | c=((char16_t)p[1]<<8)|p[0]; |
911 | if(U16_IS_SINGLE(c)) { |
912 | /* output the BMP code point */ |
913 | *target++=c; |
914 | if(offsets!=nullptr) { |
915 | *offsets++=-1; |
916 | } |
917 | --targetCapacity; |
918 | count=0; |
919 | c=0; |
920 | break; |
921 | } else if(U16_IS_SURROGATE_LEAD(c)) { |
922 | /* continue collecting bytes for the trail surrogate */ |
923 | c=0; /* avoid unnecessary surrogate handling below */ |
924 | } else { |
925 | /* fall through to error handling for an unmatched trail surrogate */ |
926 | break; |
927 | } |
928 | } else if(count==4) { |
929 | c=((char16_t)p[1]<<8)|p[0]; |
930 | trail=((char16_t)p[3]<<8)|p[2]; |
931 | if(U16_IS_TRAIL(trail)) { |
932 | /* output the surrogate pair */ |
933 | *target++=c; |
934 | if(targetCapacity>=2) { |
935 | *target++=trail; |
936 | if(offsets!=nullptr) { |
937 | *offsets++=-1; |
938 | *offsets++=-1; |
939 | } |
940 | targetCapacity-=2; |
941 | } else /* targetCapacity==1 */ { |
942 | targetCapacity=0; |
943 | cnv->UCharErrorBuffer[0]=trail; |
944 | cnv->UCharErrorBufferLength=1; |
945 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
946 | } |
947 | count=0; |
948 | c=0; |
949 | break; |
950 | } else { |
951 | /* unmatched lead surrogate, handle here for consistent toUBytes[] */ |
952 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
953 | |
954 | /* back out reading the code unit after it */ |
955 | if(((const uint8_t *)pArgs->source-source)>=2) { |
956 | source-=2; |
957 | } else { |
958 | /* |
959 | * if the trail unit's first byte was in a previous buffer, then |
960 | * we need to put it into a special place because toUBytes[] will be |
961 | * used for the lead unit's bytes |
962 | */ |
963 | cnv->toUnicodeStatus=0x100|p[2]; |
964 | --source; |
965 | } |
966 | cnv->toULength=2; |
967 | |
968 | /* write back the updated pointers */ |
969 | pArgs->source=(const char *)source; |
970 | pArgs->target=target; |
971 | pArgs->offsets=offsets; |
972 | return; |
973 | } |
974 | } |
975 | } while(length>0); |
976 | cnv->toULength=(int8_t)count; |
977 | } |
978 | |
979 | /* copy an even number of bytes for complete UChars */ |
980 | count=2*targetCapacity; |
981 | if(count>length) { |
982 | count=length&~1; |
983 | } |
984 | if(c==0 && count>0) { |
985 | length-=count; |
986 | count>>=1; |
987 | targetCapacity-=count; |
988 | if(offsets==nullptr) { |
989 | do { |
990 | c=((char16_t)source[1]<<8)|source[0]; |
991 | source+=2; |
992 | if(U16_IS_SINGLE(c)) { |
993 | *target++=c; |
994 | } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && |
995 | U16_IS_TRAIL(trail=((char16_t)source[1]<<8)|source[0]) |
996 | ) { |
997 | source+=2; |
998 | --count; |
999 | *target++=c; |
1000 | *target++=trail; |
1001 | } else { |
1002 | break; |
1003 | } |
1004 | } while(--count>0); |
1005 | } else { |
1006 | do { |
1007 | c=((char16_t)source[1]<<8)|source[0]; |
1008 | source+=2; |
1009 | if(U16_IS_SINGLE(c)) { |
1010 | *target++=c; |
1011 | *offsets++=sourceIndex; |
1012 | sourceIndex+=2; |
1013 | } else if(U16_IS_SURROGATE_LEAD(c) && count>=2 && |
1014 | U16_IS_TRAIL(trail=((char16_t)source[1]<<8)|source[0]) |
1015 | ) { |
1016 | source+=2; |
1017 | --count; |
1018 | *target++=c; |
1019 | *target++=trail; |
1020 | *offsets++=sourceIndex; |
1021 | *offsets++=sourceIndex; |
1022 | sourceIndex+=4; |
1023 | } else { |
1024 | break; |
1025 | } |
1026 | } while(--count>0); |
1027 | } |
1028 | |
1029 | if(count==0) { |
1030 | /* done with the loop for complete UChars */ |
1031 | c=0; |
1032 | } else { |
1033 | /* keep c for surrogate handling, trail will be set there */ |
1034 | length+=2*(count-1); /* one more byte pair was consumed than count decremented */ |
1035 | targetCapacity+=count; |
1036 | } |
1037 | } |
1038 | |
1039 | if(c!=0) { |
1040 | /* |
1041 | * c is a surrogate, and |
1042 | * - source or target too short |
1043 | * - or the surrogate is unmatched |
1044 | */ |
1045 | cnv->toUBytes[0]=(uint8_t)c; |
1046 | cnv->toUBytes[1]=(uint8_t)(c>>8); |
1047 | cnv->toULength=2; |
1048 | |
1049 | if(U16_IS_SURROGATE_LEAD(c)) { |
1050 | if(length>=2) { |
1051 | if(U16_IS_TRAIL(trail=((char16_t)source[1]<<8)|source[0])) { |
1052 | /* output the surrogate pair, will overflow (see conditions comment above) */ |
1053 | source+=2; |
1054 | length-=2; |
1055 | *target++=c; |
1056 | if(offsets!=nullptr) { |
1057 | *offsets++=sourceIndex; |
1058 | } |
1059 | cnv->UCharErrorBuffer[0]=trail; |
1060 | cnv->UCharErrorBufferLength=1; |
1061 | cnv->toULength=0; |
1062 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1063 | } else { |
1064 | /* unmatched lead surrogate */ |
1065 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
1066 | } |
1067 | } else { |
1068 | /* see if the trail surrogate is in the next buffer */ |
1069 | } |
1070 | } else { |
1071 | /* unmatched trail surrogate */ |
1072 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
1073 | } |
1074 | } |
1075 | |
1076 | if(U_SUCCESS(*pErrorCode)) { |
1077 | /* check for a remaining source byte */ |
1078 | if(length>0) { |
1079 | if(targetCapacity==0) { |
1080 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1081 | } else { |
1082 | /* it must be length==1 because otherwise the above would have copied more */ |
1083 | cnv->toUBytes[cnv->toULength++]=*source++; |
1084 | } |
1085 | } |
1086 | } |
1087 | |
1088 | /* write back the updated pointers */ |
1089 | pArgs->source=(const char *)source; |
1090 | pArgs->target=target; |
1091 | pArgs->offsets=offsets; |
1092 | } |
1093 | |
1094 | static UChar32 U_CALLCONV |
1095 | _UTF16LEGetNextUChar(UConverterToUnicodeArgs *pArgs, UErrorCode *err) { |
1096 | const uint8_t *s, *sourceLimit; |
1097 | UChar32 c; |
1098 | |
1099 | if(pArgs->converter->mode<8) { |
1100 | return UCNV_GET_NEXT_UCHAR_USE_TO_U; |
1101 | } |
1102 | |
1103 | s=(const uint8_t *)pArgs->source; |
1104 | sourceLimit=(const uint8_t *)pArgs->sourceLimit; |
1105 | |
1106 | if(s>=sourceLimit) { |
1107 | /* no input */ |
1108 | *err=U_INDEX_OUTOFBOUNDS_ERROR; |
1109 | return 0xffff; |
1110 | } |
1111 | |
1112 | if(s+2>sourceLimit) { |
1113 | /* only one byte: truncated char16_t */ |
1114 | pArgs->converter->toUBytes[0]=*s++; |
1115 | pArgs->converter->toULength=1; |
1116 | pArgs->source=(const char *)s; |
1117 | *err = U_TRUNCATED_CHAR_FOUND; |
1118 | return 0xffff; |
1119 | } |
1120 | |
1121 | /* get one char16_t */ |
1122 | c=((UChar32)s[1]<<8)|*s; |
1123 | s+=2; |
1124 | |
1125 | /* check for a surrogate pair */ |
1126 | if(U_IS_SURROGATE(c)) { |
1127 | if(U16_IS_SURROGATE_LEAD(c)) { |
1128 | if(s+2<=sourceLimit) { |
1129 | char16_t trail; |
1130 | |
1131 | /* get a second char16_t and see if it is a trail surrogate */ |
1132 | trail=((char16_t)s[1]<<8)|*s; |
1133 | if(U16_IS_TRAIL(trail)) { |
1134 | c=U16_GET_SUPPLEMENTARY(c, trail); |
1135 | s+=2; |
1136 | } else { |
1137 | /* unmatched lead surrogate */ |
1138 | c=-2; |
1139 | } |
1140 | } else { |
1141 | /* too few (2 or 3) bytes for a surrogate pair: truncated code point */ |
1142 | uint8_t *bytes=pArgs->converter->toUBytes; |
1143 | s-=2; |
1144 | pArgs->converter->toULength=(int8_t)(sourceLimit-s); |
1145 | do { |
1146 | *bytes++=*s++; |
1147 | } while(s<sourceLimit); |
1148 | |
1149 | c=0xffff; |
1150 | *err=U_TRUNCATED_CHAR_FOUND; |
1151 | } |
1152 | } else { |
1153 | /* unmatched trail surrogate */ |
1154 | c=-2; |
1155 | } |
1156 | |
1157 | if(c<0) { |
1158 | /* write the unmatched surrogate */ |
1159 | uint8_t *bytes=pArgs->converter->toUBytes; |
1160 | pArgs->converter->toULength=2; |
1161 | *bytes=*(s-2); |
1162 | bytes[1]=*(s-1); |
1163 | |
1164 | c=0xffff; |
1165 | *err=U_ILLEGAL_CHAR_FOUND; |
1166 | } |
1167 | } |
1168 | |
1169 | pArgs->source=(const char *)s; |
1170 | return c; |
1171 | } |
1172 | |
1173 | static void U_CALLCONV |
1174 | _UTF16LEReset(UConverter *cnv, UConverterResetChoice choice) { |
1175 | if(choice<=UCNV_RESET_TO_UNICODE) { |
1176 | /* reset toUnicode state */ |
1177 | if(UCNV_GET_VERSION(cnv)==0) { |
1178 | cnv->mode=8; /* no BOM handling */ |
1179 | } else { |
1180 | cnv->mode=0; /* Java-specific "UnicodeLittle" requires LE BOM or no BOM */ |
1181 | } |
1182 | } |
1183 | if(choice!=UCNV_RESET_TO_UNICODE && UCNV_GET_VERSION(cnv)==1) { |
1184 | /* reset fromUnicode for "UnicodeLittle": prepare to output the UTF-16LE BOM */ |
1185 | cnv->fromUnicodeStatus=UCNV_NEED_TO_WRITE_BOM; |
1186 | } |
1187 | } |
1188 | |
1189 | static void U_CALLCONV |
1190 | _UTF16LEOpen(UConverter *cnv, |
1191 | UConverterLoadArgs *pArgs, |
1192 | UErrorCode *pErrorCode) { |
1193 | (void)pArgs; |
1194 | if(UCNV_GET_VERSION(cnv)<=1) { |
1195 | _UTF16LEReset(cnv, UCNV_RESET_BOTH); |
1196 | } else { |
1197 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
1198 | } |
1199 | } |
1200 | |
1201 | static const char * U_CALLCONV |
1202 | _UTF16LEGetName(const UConverter *cnv) { |
1203 | if(UCNV_GET_VERSION(cnv)==0) { |
1204 | return "UTF-16LE" ; |
1205 | } else { |
1206 | return "UTF-16LE,version=1" ; |
1207 | } |
1208 | } |
1209 | U_CDECL_END |
1210 | |
1211 | static const UConverterImpl _UTF16LEImpl={ |
1212 | UCNV_UTF16_LittleEndian, |
1213 | |
1214 | nullptr, |
1215 | nullptr, |
1216 | |
1217 | _UTF16LEOpen, |
1218 | nullptr, |
1219 | _UTF16LEReset, |
1220 | |
1221 | _UTF16LEToUnicodeWithOffsets, |
1222 | _UTF16LEToUnicodeWithOffsets, |
1223 | _UTF16LEFromUnicodeWithOffsets, |
1224 | _UTF16LEFromUnicodeWithOffsets, |
1225 | _UTF16LEGetNextUChar, |
1226 | |
1227 | nullptr, |
1228 | _UTF16LEGetName, |
1229 | nullptr, |
1230 | nullptr, |
1231 | ucnv_getNonSurrogateUnicodeSet, |
1232 | |
1233 | nullptr, |
1234 | nullptr |
1235 | }; |
1236 | |
1237 | |
1238 | static const UConverterStaticData _UTF16LEStaticData={ |
1239 | sizeof(UConverterStaticData), |
1240 | "UTF-16LE" , |
1241 | 1202, UCNV_IBM, UCNV_UTF16_LittleEndian, 2, 2, |
1242 | { 0xfd, 0xff, 0, 0 },2,false,false, |
1243 | 0, |
1244 | 0, |
1245 | { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ |
1246 | }; |
1247 | |
1248 | |
1249 | const UConverterSharedData _UTF16LEData= |
1250 | UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF16LEStaticData, &_UTF16LEImpl); |
1251 | |
1252 | /* UTF-16 (Detect BOM) ------------------------------------------------------ */ |
1253 | |
1254 | /* |
1255 | * Detect a BOM at the beginning of the stream and select UTF-16BE or UTF-16LE |
1256 | * accordingly. |
1257 | * This is a simpler version of the UTF-32 converter, with |
1258 | * fewer states for shorter BOMs. |
1259 | * |
1260 | * State values: |
1261 | * 0 initial state |
1262 | * 1 saw first byte |
1263 | * 2..5 - |
1264 | * 6..7 see _UTF16ToUnicodeWithOffsets() comments in state 1 |
1265 | * 8 UTF-16BE mode |
1266 | * 9 UTF-16LE mode |
1267 | * |
1268 | * During detection: state==number of initial bytes seen so far. |
1269 | * |
1270 | * On output, emit U+FEFF as the first code point. |
1271 | * |
1272 | * Variants: |
1273 | * - UTF-16,version=1 (Java "Unicode" encoding) treats a missing BOM as an error. |
1274 | * - UTF-16BE,version=1 (Java "UnicodeBig" encoding) and |
1275 | * UTF-16LE,version=1 (Java "UnicodeLittle" encoding) treat a reverse BOM as an error. |
1276 | */ |
1277 | U_CDECL_BEGIN |
1278 | static void U_CALLCONV |
1279 | _UTF16Reset(UConverter *cnv, UConverterResetChoice choice) { |
1280 | if(choice<=UCNV_RESET_TO_UNICODE) { |
1281 | /* reset toUnicode: state=0 */ |
1282 | cnv->mode=0; |
1283 | } |
1284 | if(choice!=UCNV_RESET_TO_UNICODE) { |
1285 | /* reset fromUnicode: prepare to output the UTF-16PE BOM */ |
1286 | cnv->fromUnicodeStatus=UCNV_NEED_TO_WRITE_BOM; |
1287 | } |
1288 | } |
1289 | U_CDECL_END |
1290 | extern const UConverterSharedData _UTF16v2Data; |
1291 | U_CDECL_BEGIN |
1292 | static void U_CALLCONV |
1293 | _UTF16Open(UConverter *cnv, |
1294 | UConverterLoadArgs *pArgs, |
1295 | UErrorCode *pErrorCode) { |
1296 | if(UCNV_GET_VERSION(cnv)<=2) { |
1297 | if(UCNV_GET_VERSION(cnv)==2 && !pArgs->onlyTestIsLoadable) { |
1298 | /* |
1299 | * Switch implementation, and switch the staticData that's different |
1300 | * and was copied into the UConverter. |
1301 | * (See ucnv_createConverterFromSharedData() in ucnv_bld.c.) |
1302 | * UTF-16,version=2 fromUnicode() always writes a big-endian byte stream. |
1303 | */ |
1304 | cnv->sharedData=(UConverterSharedData*)&_UTF16v2Data; |
1305 | uprv_memcpy(cnv->subChars, _UTF16v2Data.staticData->subChar, UCNV_MAX_SUBCHAR_LEN); |
1306 | } |
1307 | _UTF16Reset(cnv, UCNV_RESET_BOTH); |
1308 | } else { |
1309 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
1310 | } |
1311 | } |
1312 | |
1313 | static const char * U_CALLCONV |
1314 | _UTF16GetName(const UConverter *cnv) { |
1315 | if(UCNV_GET_VERSION(cnv)==0) { |
1316 | return "UTF-16" ; |
1317 | } else if(UCNV_GET_VERSION(cnv)==1) { |
1318 | return "UTF-16,version=1" ; |
1319 | } else { |
1320 | return "UTF-16,version=2" ; |
1321 | } |
1322 | } |
1323 | U_CDECL_END |
1324 | extern const UConverterSharedData _UTF16Data; |
1325 | |
1326 | static inline bool IS_UTF16BE(const UConverter *cnv) { |
1327 | return ((cnv)->sharedData == &_UTF16BEData); |
1328 | } |
1329 | |
1330 | static inline bool IS_UTF16LE(const UConverter *cnv) { |
1331 | return ((cnv)->sharedData == &_UTF16LEData); |
1332 | } |
1333 | |
1334 | static inline bool IS_UTF16(const UConverter *cnv) { |
1335 | return ((cnv)->sharedData==&_UTF16Data) || ((cnv)->sharedData == &_UTF16v2Data); |
1336 | } |
1337 | |
1338 | U_CDECL_BEGIN |
1339 | static void U_CALLCONV |
1340 | _UTF16ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, |
1341 | UErrorCode *pErrorCode) { |
1342 | UConverter *cnv=pArgs->converter; |
1343 | const char *source=pArgs->source; |
1344 | const char *sourceLimit=pArgs->sourceLimit; |
1345 | int32_t *offsets=pArgs->offsets; |
1346 | |
1347 | int32_t state, offsetDelta; |
1348 | uint8_t b; |
1349 | |
1350 | state=cnv->mode; |
1351 | |
1352 | /* |
1353 | * If we detect a BOM in this buffer, then we must add the BOM size to the |
1354 | * offsets because the actual converter function will not see and count the BOM. |
1355 | * offsetDelta will have the number of the BOM bytes that are in the current buffer. |
1356 | */ |
1357 | offsetDelta=0; |
1358 | |
1359 | while(source<sourceLimit && U_SUCCESS(*pErrorCode)) { |
1360 | switch(state) { |
1361 | case 0: |
1362 | cnv->toUBytes[0]=(uint8_t)*source++; |
1363 | cnv->toULength=1; |
1364 | state=1; |
1365 | break; |
1366 | case 1: |
1367 | /* |
1368 | * Only inside this switch case can the state variable |
1369 | * temporarily take two additional values: |
1370 | * 6: BOM error, continue with BE |
1371 | * 7: BOM error, continue with LE |
1372 | */ |
1373 | b=*source; |
1374 | if(cnv->toUBytes[0]==0xfe && b==0xff) { |
1375 | if(IS_UTF16LE(cnv)) { |
1376 | state=7; /* illegal reverse BOM for Java "UnicodeLittle" */ |
1377 | } else { |
1378 | state=8; /* detect UTF-16BE */ |
1379 | } |
1380 | } else if(cnv->toUBytes[0]==0xff && b==0xfe) { |
1381 | if(IS_UTF16BE(cnv)) { |
1382 | state=6; /* illegal reverse BOM for Java "UnicodeBig" */ |
1383 | } else { |
1384 | state=9; /* detect UTF-16LE */ |
1385 | } |
1386 | } else if((IS_UTF16(cnv) && UCNV_GET_VERSION(cnv)==1)) { |
1387 | state=6; /* illegal missing BOM for Java "Unicode" */ |
1388 | } |
1389 | if(state>=8) { |
1390 | /* BOM detected, consume it */ |
1391 | ++source; |
1392 | cnv->toULength=0; |
1393 | offsetDelta=(int32_t)(source-pArgs->source); |
1394 | } else if(state<6) { |
1395 | /* ok: no BOM, and not a reverse BOM */ |
1396 | if(source!=pArgs->source) { |
1397 | /* reset the source for a correct first offset */ |
1398 | source=pArgs->source; |
1399 | cnv->toULength=0; |
1400 | } |
1401 | if(IS_UTF16LE(cnv)) { |
1402 | /* Make Java "UnicodeLittle" default to LE. */ |
1403 | state=9; |
1404 | } else { |
1405 | /* Make standard UTF-16 and Java "UnicodeBig" default to BE. */ |
1406 | state=8; |
1407 | } |
1408 | } else { |
1409 | /* |
1410 | * error: missing BOM, or reverse BOM |
1411 | * UTF-16,version=1: Java-specific "Unicode" requires a BOM. |
1412 | * UTF-16BE,version=1: Java-specific "UnicodeBig" requires a BE BOM or no BOM. |
1413 | * UTF-16LE,version=1: Java-specific "UnicodeLittle" requires an LE BOM or no BOM. |
1414 | */ |
1415 | /* report the non-BOM or reverse BOM as an illegal sequence */ |
1416 | cnv->toUBytes[1]=b; |
1417 | cnv->toULength=2; |
1418 | pArgs->source=source+1; |
1419 | /* continue with conversion if the callback resets the error */ |
1420 | /* |
1421 | * Make Java "Unicode" default to BE like standard UTF-16. |
1422 | * Make Java "UnicodeBig" and "UnicodeLittle" default |
1423 | * to their normal endiannesses. |
1424 | */ |
1425 | cnv->mode=state+2; |
1426 | *pErrorCode=U_ILLEGAL_ESCAPE_SEQUENCE; |
1427 | return; |
1428 | } |
1429 | /* convert the rest of the stream */ |
1430 | cnv->mode=state; |
1431 | continue; |
1432 | case 8: |
1433 | /* call UTF-16BE */ |
1434 | pArgs->source=source; |
1435 | _UTF16BEToUnicodeWithOffsets(pArgs, pErrorCode); |
1436 | source=pArgs->source; |
1437 | break; |
1438 | case 9: |
1439 | /* call UTF-16LE */ |
1440 | pArgs->source=source; |
1441 | _UTF16LEToUnicodeWithOffsets(pArgs, pErrorCode); |
1442 | source=pArgs->source; |
1443 | break; |
1444 | default: |
1445 | break; /* does not occur */ |
1446 | } |
1447 | } |
1448 | |
1449 | /* add BOM size to offsets - see comment at offsetDelta declaration */ |
1450 | if(offsets!=nullptr && offsetDelta!=0) { |
1451 | int32_t *offsetsLimit=pArgs->offsets; |
1452 | while(offsets<offsetsLimit) { |
1453 | *offsets++ += offsetDelta; |
1454 | } |
1455 | } |
1456 | |
1457 | pArgs->source=source; |
1458 | |
1459 | if(source==sourceLimit && pArgs->flush) { |
1460 | /* handle truncated input */ |
1461 | switch(state) { |
1462 | case 0: |
1463 | break; /* no input at all, nothing to do */ |
1464 | case 8: |
1465 | _UTF16BEToUnicodeWithOffsets(pArgs, pErrorCode); |
1466 | break; |
1467 | case 9: |
1468 | _UTF16LEToUnicodeWithOffsets(pArgs, pErrorCode); |
1469 | break; |
1470 | default: |
1471 | /* 0<state<8: framework will report truncation, nothing to do here */ |
1472 | break; |
1473 | } |
1474 | } |
1475 | |
1476 | cnv->mode=state; |
1477 | } |
1478 | |
1479 | static UChar32 U_CALLCONV |
1480 | _UTF16GetNextUChar(UConverterToUnicodeArgs *pArgs, |
1481 | UErrorCode *pErrorCode) { |
1482 | switch(pArgs->converter->mode) { |
1483 | case 8: |
1484 | return _UTF16BEGetNextUChar(pArgs, pErrorCode); |
1485 | case 9: |
1486 | return _UTF16LEGetNextUChar(pArgs, pErrorCode); |
1487 | default: |
1488 | return UCNV_GET_NEXT_UCHAR_USE_TO_U; |
1489 | } |
1490 | } |
1491 | U_CDECL_END |
1492 | |
1493 | static const UConverterImpl _UTF16Impl = { |
1494 | UCNV_UTF16, |
1495 | |
1496 | nullptr, |
1497 | nullptr, |
1498 | |
1499 | _UTF16Open, |
1500 | nullptr, |
1501 | _UTF16Reset, |
1502 | |
1503 | _UTF16ToUnicodeWithOffsets, |
1504 | _UTF16ToUnicodeWithOffsets, |
1505 | _UTF16PEFromUnicodeWithOffsets, |
1506 | _UTF16PEFromUnicodeWithOffsets, |
1507 | _UTF16GetNextUChar, |
1508 | |
1509 | nullptr, /* ### TODO implement getStarters for all Unicode encodings?! */ |
1510 | _UTF16GetName, |
1511 | nullptr, |
1512 | nullptr, |
1513 | ucnv_getNonSurrogateUnicodeSet, |
1514 | |
1515 | nullptr, |
1516 | nullptr |
1517 | }; |
1518 | |
1519 | static const UConverterStaticData _UTF16StaticData = { |
1520 | sizeof(UConverterStaticData), |
1521 | "UTF-16" , |
1522 | 1204, /* CCSID for BOM sensitive UTF-16 */ |
1523 | UCNV_IBM, UCNV_UTF16, 2, 2, |
1524 | #if U_IS_BIG_ENDIAN |
1525 | { 0xff, 0xfd, 0, 0 }, 2, |
1526 | #else |
1527 | { 0xfd, 0xff, 0, 0 }, 2, |
1528 | #endif |
1529 | false, false, |
1530 | 0, |
1531 | 0, |
1532 | { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ |
1533 | }; |
1534 | |
1535 | const UConverterSharedData _UTF16Data = |
1536 | UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF16StaticData, &_UTF16Impl); |
1537 | |
1538 | static const UConverterImpl _UTF16v2Impl = { |
1539 | UCNV_UTF16, |
1540 | |
1541 | nullptr, |
1542 | nullptr, |
1543 | |
1544 | _UTF16Open, |
1545 | nullptr, |
1546 | _UTF16Reset, |
1547 | |
1548 | _UTF16ToUnicodeWithOffsets, |
1549 | _UTF16ToUnicodeWithOffsets, |
1550 | _UTF16BEFromUnicodeWithOffsets, |
1551 | _UTF16BEFromUnicodeWithOffsets, |
1552 | _UTF16GetNextUChar, |
1553 | |
1554 | nullptr, /* ### TODO implement getStarters for all Unicode encodings?! */ |
1555 | _UTF16GetName, |
1556 | nullptr, |
1557 | nullptr, |
1558 | ucnv_getNonSurrogateUnicodeSet, |
1559 | |
1560 | nullptr, |
1561 | nullptr |
1562 | }; |
1563 | |
1564 | static const UConverterStaticData _UTF16v2StaticData = { |
1565 | sizeof(UConverterStaticData), |
1566 | "UTF-16,version=2" , |
1567 | 1204, /* CCSID for BOM sensitive UTF-16 */ |
1568 | UCNV_IBM, UCNV_UTF16, 2, 2, |
1569 | { 0xff, 0xfd, 0, 0 }, 2, |
1570 | false, false, |
1571 | 0, |
1572 | 0, |
1573 | { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ |
1574 | }; |
1575 | |
1576 | const UConverterSharedData _UTF16v2Data = |
1577 | UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF16v2StaticData, &_UTF16v2Impl); |
1578 | |
1579 | #endif |
1580 | |