1 /*
2 **********************************************************************
3 * Copyright (C) 2002-2012, International Business Machines
4 * Corporation and others. All Rights Reserved.
5 **********************************************************************
6 * file name: ucnv_u8.c
7 * encoding: US-ASCII
8 * tab size: 8 (not used)
9 * indentation:4
10 *
11 * created on: 2002jul01
12 * created by: Markus W. Scherer
13 *
14 * UTF-8 converter implementation. Used to be in ucnv_utf.c.
15 *
16 * Also, CESU-8 implementation, see UTR 26.
17 * The CESU-8 converter uses all the same functions as the
18 * UTF-8 converter, with a branch for converting supplementary code points.
19 */
20
21 #include "unicode/utypes.h"
22
23 #if !UCONFIG_NO_CONVERSION
24
25 #include "unicode/ucnv.h"
26 #include "unicode/utf.h"
27 #include "unicode/utf8.h"
28 #include "unicode/utf16.h"
29 #include "ucnv_bld.h"
30 #include "ucnv_cnv.h"
31 #include "cmemory.h"
32
33 /* Prototypes --------------------------------------------------------------- */
34
35 /* Keep these here to make finicky compilers happy */
36
37 U_CFUNC void ucnv_fromUnicode_UTF8(UConverterFromUnicodeArgs *args,
38 UErrorCode *err);
39 U_CFUNC void ucnv_fromUnicode_UTF8_OFFSETS_LOGIC(UConverterFromUnicodeArgs *args,
40 UErrorCode *err);
41
42
43 /* UTF-8 -------------------------------------------------------------------- */
44
45 /* UTF-8 Conversion DATA
46 * for more information see Unicode Standard 2.0, Transformation Formats Appendix A-9
47 */
48 /*static const uint32_t REPLACEMENT_CHARACTER = 0x0000FFFD;*/
49 #define MAXIMUM_UCS2 0x0000FFFF
50 #define MAXIMUM_UTF 0x0010FFFF
51 #define MAXIMUM_UCS4 0x7FFFFFFF
52 #define HALF_SHIFT 10
53 #define HALF_BASE 0x0010000
54 #define HALF_MASK 0x3FF
55 #define SURROGATE_HIGH_START 0xD800
56 #define SURROGATE_HIGH_END 0xDBFF
57 #define SURROGATE_LOW_START 0xDC00
58 #define SURROGATE_LOW_END 0xDFFF
59
60 /* -SURROGATE_LOW_START + HALF_BASE */
61 #define SURROGATE_LOW_BASE 9216
62
63 static const uint32_t offsetsFromUTF8[7] = {0,
64 (uint32_t) 0x00000000, (uint32_t) 0x00003080, (uint32_t) 0x000E2080,
65 (uint32_t) 0x03C82080, (uint32_t) 0xFA082080, (uint32_t) 0x82082080
66 };
67
68 /* END OF UTF-8 Conversion DATA */
69
70 static const int8_t bytesFromUTF8[256] = {
71 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
72 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
73 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
74 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
75 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
76 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
77 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
78 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 6, 6, 0, 0
79 };
80
81 /*
82 * Starting with Unicode 3.0.1:
83 * UTF-8 byte sequences of length N _must_ encode code points of or above utf8_minChar32[N];
84 * byte sequences with more than 4 bytes are illegal in UTF-8,
85 * which is tested with impossible values for them
86 */
87 static const uint32_t
88 utf8_minChar32[7]={ 0, 0, 0x80, 0x800, 0x10000, 0xffffffff, 0xffffffff };
89
ucnv_toUnicode_UTF8(UConverterToUnicodeArgs * args,UErrorCode * err)90 static void ucnv_toUnicode_UTF8 (UConverterToUnicodeArgs * args,
91 UErrorCode * err)
92 {
93 UConverter *cnv = args->converter;
94 const unsigned char *mySource = (unsigned char *) args->source;
95 UChar *myTarget = args->target;
96 const unsigned char *sourceLimit = (unsigned char *) args->sourceLimit;
97 const UChar *targetLimit = args->targetLimit;
98 unsigned char *toUBytes = cnv->toUBytes;
99 UBool isCESU8 = (UBool)(cnv->sharedData == &_CESU8Data);
100 uint32_t ch, ch2 = 0;
101 int32_t i, inBytes;
102
103 /* Restore size of current sequence */
104 if (cnv->toUnicodeStatus && myTarget < targetLimit)
105 {
106 inBytes = cnv->mode; /* restore # of bytes to consume */
107 i = cnv->toULength; /* restore # of bytes consumed */
108 cnv->toULength = 0;
109
110 ch = cnv->toUnicodeStatus;/*Stores the previously calculated ch from a previous call*/
111 cnv->toUnicodeStatus = 0;
112 goto morebytes;
113 }
114
115
116 while (mySource < sourceLimit && myTarget < targetLimit)
117 {
118 ch = *(mySource++);
119 if (ch < 0x80) /* Simple case */
120 {
121 *(myTarget++) = (UChar) ch;
122 }
123 else
124 {
125 /* store the first char */
126 toUBytes[0] = (char)ch;
127 inBytes = bytesFromUTF8[ch]; /* lookup current sequence length */
128 i = 1;
129
130 morebytes:
131 while (i < inBytes)
132 {
133 if (mySource < sourceLimit)
134 {
135 toUBytes[i] = (char) (ch2 = *mySource);
136 if (!U8_IS_TRAIL(ch2))
137 {
138 break; /* i < inBytes */
139 }
140 ch = (ch << 6) + ch2;
141 ++mySource;
142 i++;
143 }
144 else
145 {
146 /* stores a partially calculated target*/
147 cnv->toUnicodeStatus = ch;
148 cnv->mode = inBytes;
149 cnv->toULength = (int8_t) i;
150 goto donefornow;
151 }
152 }
153
154 /* Remove the accumulated high bits */
155 ch -= offsetsFromUTF8[inBytes];
156
157 /*
158 * Legal UTF-8 byte sequences in Unicode 3.0.1 and up:
159 * - use only trail bytes after a lead byte (checked above)
160 * - use the right number of trail bytes for a given lead byte
161 * - encode a code point <= U+10ffff
162 * - use the fewest possible number of bytes for their code points
163 * - use at most 4 bytes (for i>=5 it is 0x10ffff<utf8_minChar32[])
164 *
165 * Starting with Unicode 3.2, surrogate code points must not be encoded in UTF-8.
166 * There are no irregular sequences any more.
167 * In CESU-8, only surrogates, not supplementary code points, are encoded directly.
168 */
169 if (i == inBytes && ch <= MAXIMUM_UTF && ch >= utf8_minChar32[i] &&
170 (isCESU8 ? i <= 3 : !U_IS_SURROGATE(ch)))
171 {
172 /* Normal valid byte when the loop has not prematurely terminated (i < inBytes) */
173 if (ch <= MAXIMUM_UCS2)
174 {
175 /* fits in 16 bits */
176 *(myTarget++) = (UChar) ch;
177 }
178 else
179 {
180 /* write out the surrogates */
181 ch -= HALF_BASE;
182 *(myTarget++) = (UChar) ((ch >> HALF_SHIFT) + SURROGATE_HIGH_START);
183 ch = (ch & HALF_MASK) + SURROGATE_LOW_START;
184 if (myTarget < targetLimit)
185 {
186 *(myTarget++) = (UChar)ch;
187 }
188 else
189 {
190 /* Put in overflow buffer (not handled here) */
191 cnv->UCharErrorBuffer[0] = (UChar) ch;
192 cnv->UCharErrorBufferLength = 1;
193 *err = U_BUFFER_OVERFLOW_ERROR;
194 break;
195 }
196 }
197 }
198 else
199 {
200 cnv->toULength = (int8_t)i;
201 *err = U_ILLEGAL_CHAR_FOUND;
202 break;
203 }
204 }
205 }
206
207 donefornow:
208 if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err))
209 {
210 /* End of target buffer */
211 *err = U_BUFFER_OVERFLOW_ERROR;
212 }
213
214 args->target = myTarget;
215 args->source = (const char *) mySource;
216 }
217
ucnv_toUnicode_UTF8_OFFSETS_LOGIC(UConverterToUnicodeArgs * args,UErrorCode * err)218 static void ucnv_toUnicode_UTF8_OFFSETS_LOGIC (UConverterToUnicodeArgs * args,
219 UErrorCode * err)
220 {
221 UConverter *cnv = args->converter;
222 const unsigned char *mySource = (unsigned char *) args->source;
223 UChar *myTarget = args->target;
224 int32_t *myOffsets = args->offsets;
225 int32_t offsetNum = 0;
226 const unsigned char *sourceLimit = (unsigned char *) args->sourceLimit;
227 const UChar *targetLimit = args->targetLimit;
228 unsigned char *toUBytes = cnv->toUBytes;
229 UBool isCESU8 = (UBool)(cnv->sharedData == &_CESU8Data);
230 uint32_t ch, ch2 = 0;
231 int32_t i, inBytes;
232
233 /* Restore size of current sequence */
234 if (cnv->toUnicodeStatus && myTarget < targetLimit)
235 {
236 inBytes = cnv->mode; /* restore # of bytes to consume */
237 i = cnv->toULength; /* restore # of bytes consumed */
238 cnv->toULength = 0;
239
240 ch = cnv->toUnicodeStatus;/*Stores the previously calculated ch from a previous call*/
241 cnv->toUnicodeStatus = 0;
242 goto morebytes;
243 }
244
245 while (mySource < sourceLimit && myTarget < targetLimit)
246 {
247 ch = *(mySource++);
248 if (ch < 0x80) /* Simple case */
249 {
250 *(myTarget++) = (UChar) ch;
251 *(myOffsets++) = offsetNum++;
252 }
253 else
254 {
255 toUBytes[0] = (char)ch;
256 inBytes = bytesFromUTF8[ch];
257 i = 1;
258
259 morebytes:
260 while (i < inBytes)
261 {
262 if (mySource < sourceLimit)
263 {
264 toUBytes[i] = (char) (ch2 = *mySource);
265 if (!U8_IS_TRAIL(ch2))
266 {
267 break; /* i < inBytes */
268 }
269 ch = (ch << 6) + ch2;
270 ++mySource;
271 i++;
272 }
273 else
274 {
275 cnv->toUnicodeStatus = ch;
276 cnv->mode = inBytes;
277 cnv->toULength = (int8_t)i;
278 goto donefornow;
279 }
280 }
281
282 /* Remove the accumulated high bits */
283 ch -= offsetsFromUTF8[inBytes];
284
285 /*
286 * Legal UTF-8 byte sequences in Unicode 3.0.1 and up:
287 * - use only trail bytes after a lead byte (checked above)
288 * - use the right number of trail bytes for a given lead byte
289 * - encode a code point <= U+10ffff
290 * - use the fewest possible number of bytes for their code points
291 * - use at most 4 bytes (for i>=5 it is 0x10ffff<utf8_minChar32[])
292 *
293 * Starting with Unicode 3.2, surrogate code points must not be encoded in UTF-8.
294 * There are no irregular sequences any more.
295 * In CESU-8, only surrogates, not supplementary code points, are encoded directly.
296 */
297 if (i == inBytes && ch <= MAXIMUM_UTF && ch >= utf8_minChar32[i] &&
298 (isCESU8 ? i <= 3 : !U_IS_SURROGATE(ch)))
299 {
300 /* Normal valid byte when the loop has not prematurely terminated (i < inBytes) */
301 if (ch <= MAXIMUM_UCS2)
302 {
303 /* fits in 16 bits */
304 *(myTarget++) = (UChar) ch;
305 *(myOffsets++) = offsetNum;
306 }
307 else
308 {
309 /* write out the surrogates */
310 ch -= HALF_BASE;
311 *(myTarget++) = (UChar) ((ch >> HALF_SHIFT) + SURROGATE_HIGH_START);
312 *(myOffsets++) = offsetNum;
313 ch = (ch & HALF_MASK) + SURROGATE_LOW_START;
314 if (myTarget < targetLimit)
315 {
316 *(myTarget++) = (UChar)ch;
317 *(myOffsets++) = offsetNum;
318 }
319 else
320 {
321 cnv->UCharErrorBuffer[0] = (UChar) ch;
322 cnv->UCharErrorBufferLength = 1;
323 *err = U_BUFFER_OVERFLOW_ERROR;
324 }
325 }
326 offsetNum += i;
327 }
328 else
329 {
330 cnv->toULength = (int8_t)i;
331 *err = U_ILLEGAL_CHAR_FOUND;
332 break;
333 }
334 }
335 }
336
337 donefornow:
338 if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err))
339 { /* End of target buffer */
340 *err = U_BUFFER_OVERFLOW_ERROR;
341 }
342
343 args->target = myTarget;
344 args->source = (const char *) mySource;
345 args->offsets = myOffsets;
346 }
347
ucnv_fromUnicode_UTF8(UConverterFromUnicodeArgs * args,UErrorCode * err)348 U_CFUNC void ucnv_fromUnicode_UTF8 (UConverterFromUnicodeArgs * args,
349 UErrorCode * err)
350 {
351 UConverter *cnv = args->converter;
352 const UChar *mySource = args->source;
353 const UChar *sourceLimit = args->sourceLimit;
354 uint8_t *myTarget = (uint8_t *) args->target;
355 const uint8_t *targetLimit = (uint8_t *) args->targetLimit;
356 uint8_t *tempPtr;
357 UChar32 ch;
358 uint8_t tempBuf[4];
359 int32_t indexToWrite;
360 UBool isNotCESU8 = (UBool)(cnv->sharedData != &_CESU8Data);
361
362 if (cnv->fromUChar32 && myTarget < targetLimit)
363 {
364 ch = cnv->fromUChar32;
365 cnv->fromUChar32 = 0;
366 goto lowsurrogate;
367 }
368
369 while (mySource < sourceLimit && myTarget < targetLimit)
370 {
371 ch = *(mySource++);
372
373 if (ch < 0x80) /* Single byte */
374 {
375 *(myTarget++) = (uint8_t) ch;
376 }
377 else if (ch < 0x800) /* Double byte */
378 {
379 *(myTarget++) = (uint8_t) ((ch >> 6) | 0xc0);
380 if (myTarget < targetLimit)
381 {
382 *(myTarget++) = (uint8_t) ((ch & 0x3f) | 0x80);
383 }
384 else
385 {
386 cnv->charErrorBuffer[0] = (uint8_t) ((ch & 0x3f) | 0x80);
387 cnv->charErrorBufferLength = 1;
388 *err = U_BUFFER_OVERFLOW_ERROR;
389 }
390 }
391 else {
392 /* Check for surrogates */
393 if(U16_IS_SURROGATE(ch) && isNotCESU8) {
394 lowsurrogate:
395 if (mySource < sourceLimit) {
396 /* test both code units */
397 if(U16_IS_SURROGATE_LEAD(ch) && U16_IS_TRAIL(*mySource)) {
398 /* convert and consume this supplementary code point */
399 ch=U16_GET_SUPPLEMENTARY(ch, *mySource);
400 ++mySource;
401 /* exit this condition tree */
402 }
403 else {
404 /* this is an unpaired trail or lead code unit */
405 /* callback(illegal) */
406 cnv->fromUChar32 = ch;
407 *err = U_ILLEGAL_CHAR_FOUND;
408 break;
409 }
410 }
411 else {
412 /* no more input */
413 cnv->fromUChar32 = ch;
414 break;
415 }
416 }
417
418 /* Do we write the buffer directly for speed,
419 or do we have to be careful about target buffer space? */
420 tempPtr = (((targetLimit - myTarget) >= 4) ? myTarget : tempBuf);
421
422 if (ch <= MAXIMUM_UCS2) {
423 indexToWrite = 2;
424 tempPtr[0] = (uint8_t) ((ch >> 12) | 0xe0);
425 }
426 else {
427 indexToWrite = 3;
428 tempPtr[0] = (uint8_t) ((ch >> 18) | 0xf0);
429 tempPtr[1] = (uint8_t) (((ch >> 12) & 0x3f) | 0x80);
430 }
431 tempPtr[indexToWrite-1] = (uint8_t) (((ch >> 6) & 0x3f) | 0x80);
432 tempPtr[indexToWrite] = (uint8_t) ((ch & 0x3f) | 0x80);
433
434 if (tempPtr == myTarget) {
435 /* There was enough space to write the codepoint directly. */
436 myTarget += (indexToWrite + 1);
437 }
438 else {
439 /* We might run out of room soon. Write it slowly. */
440 for (; tempPtr <= (tempBuf + indexToWrite); tempPtr++) {
441 if (myTarget < targetLimit) {
442 *(myTarget++) = *tempPtr;
443 }
444 else {
445 cnv->charErrorBuffer[cnv->charErrorBufferLength++] = *tempPtr;
446 *err = U_BUFFER_OVERFLOW_ERROR;
447 }
448 }
449 }
450 }
451 }
452
453 if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err))
454 {
455 *err = U_BUFFER_OVERFLOW_ERROR;
456 }
457
458 args->target = (char *) myTarget;
459 args->source = mySource;
460 }
461
ucnv_fromUnicode_UTF8_OFFSETS_LOGIC(UConverterFromUnicodeArgs * args,UErrorCode * err)462 U_CFUNC void ucnv_fromUnicode_UTF8_OFFSETS_LOGIC (UConverterFromUnicodeArgs * args,
463 UErrorCode * err)
464 {
465 UConverter *cnv = args->converter;
466 const UChar *mySource = args->source;
467 int32_t *myOffsets = args->offsets;
468 const UChar *sourceLimit = args->sourceLimit;
469 uint8_t *myTarget = (uint8_t *) args->target;
470 const uint8_t *targetLimit = (uint8_t *) args->targetLimit;
471 uint8_t *tempPtr;
472 UChar32 ch;
473 int32_t offsetNum, nextSourceIndex;
474 int32_t indexToWrite;
475 uint8_t tempBuf[4];
476 UBool isNotCESU8 = (UBool)(cnv->sharedData != &_CESU8Data);
477
478 if (cnv->fromUChar32 && myTarget < targetLimit)
479 {
480 ch = cnv->fromUChar32;
481 cnv->fromUChar32 = 0;
482 offsetNum = -1;
483 nextSourceIndex = 0;
484 goto lowsurrogate;
485 } else {
486 offsetNum = 0;
487 }
488
489 while (mySource < sourceLimit && myTarget < targetLimit)
490 {
491 ch = *(mySource++);
492
493 if (ch < 0x80) /* Single byte */
494 {
495 *(myOffsets++) = offsetNum++;
496 *(myTarget++) = (char) ch;
497 }
498 else if (ch < 0x800) /* Double byte */
499 {
500 *(myOffsets++) = offsetNum;
501 *(myTarget++) = (uint8_t) ((ch >> 6) | 0xc0);
502 if (myTarget < targetLimit)
503 {
504 *(myOffsets++) = offsetNum++;
505 *(myTarget++) = (uint8_t) ((ch & 0x3f) | 0x80);
506 }
507 else
508 {
509 cnv->charErrorBuffer[0] = (uint8_t) ((ch & 0x3f) | 0x80);
510 cnv->charErrorBufferLength = 1;
511 *err = U_BUFFER_OVERFLOW_ERROR;
512 }
513 }
514 else
515 /* Check for surrogates */
516 {
517 nextSourceIndex = offsetNum + 1;
518
519 if(U16_IS_SURROGATE(ch) && isNotCESU8) {
520 lowsurrogate:
521 if (mySource < sourceLimit) {
522 /* test both code units */
523 if(U16_IS_SURROGATE_LEAD(ch) && U16_IS_TRAIL(*mySource)) {
524 /* convert and consume this supplementary code point */
525 ch=U16_GET_SUPPLEMENTARY(ch, *mySource);
526 ++mySource;
527 ++nextSourceIndex;
528 /* exit this condition tree */
529 }
530 else {
531 /* this is an unpaired trail or lead code unit */
532 /* callback(illegal) */
533 cnv->fromUChar32 = ch;
534 *err = U_ILLEGAL_CHAR_FOUND;
535 break;
536 }
537 }
538 else {
539 /* no more input */
540 cnv->fromUChar32 = ch;
541 break;
542 }
543 }
544
545 /* Do we write the buffer directly for speed,
546 or do we have to be careful about target buffer space? */
547 tempPtr = (((targetLimit - myTarget) >= 4) ? myTarget : tempBuf);
548
549 if (ch <= MAXIMUM_UCS2) {
550 indexToWrite = 2;
551 tempPtr[0] = (uint8_t) ((ch >> 12) | 0xe0);
552 }
553 else {
554 indexToWrite = 3;
555 tempPtr[0] = (uint8_t) ((ch >> 18) | 0xf0);
556 tempPtr[1] = (uint8_t) (((ch >> 12) & 0x3f) | 0x80);
557 }
558 tempPtr[indexToWrite-1] = (uint8_t) (((ch >> 6) & 0x3f) | 0x80);
559 tempPtr[indexToWrite] = (uint8_t) ((ch & 0x3f) | 0x80);
560
561 if (tempPtr == myTarget) {
562 /* There was enough space to write the codepoint directly. */
563 myTarget += (indexToWrite + 1);
564 myOffsets[0] = offsetNum;
565 myOffsets[1] = offsetNum;
566 myOffsets[2] = offsetNum;
567 if (indexToWrite >= 3) {
568 myOffsets[3] = offsetNum;
569 }
570 myOffsets += (indexToWrite + 1);
571 }
572 else {
573 /* We might run out of room soon. Write it slowly. */
574 for (; tempPtr <= (tempBuf + indexToWrite); tempPtr++) {
575 if (myTarget < targetLimit)
576 {
577 *(myOffsets++) = offsetNum;
578 *(myTarget++) = *tempPtr;
579 }
580 else
581 {
582 cnv->charErrorBuffer[cnv->charErrorBufferLength++] = *tempPtr;
583 *err = U_BUFFER_OVERFLOW_ERROR;
584 }
585 }
586 }
587 offsetNum = nextSourceIndex;
588 }
589 }
590
591 if (mySource < sourceLimit && myTarget >= targetLimit && U_SUCCESS(*err))
592 {
593 *err = U_BUFFER_OVERFLOW_ERROR;
594 }
595
596 args->target = (char *) myTarget;
597 args->source = mySource;
598 args->offsets = myOffsets;
599 }
600
ucnv_getNextUChar_UTF8(UConverterToUnicodeArgs * args,UErrorCode * err)601 static UChar32 ucnv_getNextUChar_UTF8(UConverterToUnicodeArgs *args,
602 UErrorCode *err) {
603 UConverter *cnv;
604 const uint8_t *sourceInitial;
605 const uint8_t *source;
606 uint16_t extraBytesToWrite;
607 uint8_t myByte;
608 UChar32 ch;
609 int8_t i, isLegalSequence;
610
611 /* UTF-8 only here, the framework handles CESU-8 to combine surrogate pairs */
612
613 cnv = args->converter;
614 sourceInitial = source = (const uint8_t *)args->source;
615 if (source >= (const uint8_t *)args->sourceLimit)
616 {
617 /* no input */
618 *err = U_INDEX_OUTOFBOUNDS_ERROR;
619 return 0xffff;
620 }
621
622 myByte = (uint8_t)*(source++);
623 if (myByte < 0x80)
624 {
625 args->source = (const char *)source;
626 return (UChar32)myByte;
627 }
628
629 extraBytesToWrite = (uint16_t)bytesFromUTF8[myByte];
630 if (extraBytesToWrite == 0) {
631 cnv->toUBytes[0] = myByte;
632 cnv->toULength = 1;
633 *err = U_ILLEGAL_CHAR_FOUND;
634 args->source = (const char *)source;
635 return 0xffff;
636 }
637
638 /*The byte sequence is longer than the buffer area passed*/
639 if (((const char *)source + extraBytesToWrite - 1) > args->sourceLimit)
640 {
641 /* check if all of the remaining bytes are trail bytes */
642 cnv->toUBytes[0] = myByte;
643 i = 1;
644 *err = U_TRUNCATED_CHAR_FOUND;
645 while(source < (const uint8_t *)args->sourceLimit) {
646 if(U8_IS_TRAIL(myByte = *source)) {
647 cnv->toUBytes[i++] = myByte;
648 ++source;
649 } else {
650 /* error even before we run out of input */
651 *err = U_ILLEGAL_CHAR_FOUND;
652 break;
653 }
654 }
655 cnv->toULength = i;
656 args->source = (const char *)source;
657 return 0xffff;
658 }
659
660 isLegalSequence = 1;
661 ch = myByte << 6;
662 switch(extraBytesToWrite)
663 {
664 /* note: code falls through cases! (sic)*/
665 case 6:
666 ch += (myByte = *source);
667 ch <<= 6;
668 if (!U8_IS_TRAIL(myByte))
669 {
670 isLegalSequence = 0;
671 break;
672 }
673 ++source;
674 case 5: /*fall through*/
675 ch += (myByte = *source);
676 ch <<= 6;
677 if (!U8_IS_TRAIL(myByte))
678 {
679 isLegalSequence = 0;
680 break;
681 }
682 ++source;
683 case 4: /*fall through*/
684 ch += (myByte = *source);
685 ch <<= 6;
686 if (!U8_IS_TRAIL(myByte))
687 {
688 isLegalSequence = 0;
689 break;
690 }
691 ++source;
692 case 3: /*fall through*/
693 ch += (myByte = *source);
694 ch <<= 6;
695 if (!U8_IS_TRAIL(myByte))
696 {
697 isLegalSequence = 0;
698 break;
699 }
700 ++source;
701 case 2: /*fall through*/
702 ch += (myByte = *source);
703 if (!U8_IS_TRAIL(myByte))
704 {
705 isLegalSequence = 0;
706 break;
707 }
708 ++source;
709 };
710 ch -= offsetsFromUTF8[extraBytesToWrite];
711 args->source = (const char *)source;
712
713 /*
714 * Legal UTF-8 byte sequences in Unicode 3.0.1 and up:
715 * - use only trail bytes after a lead byte (checked above)
716 * - use the right number of trail bytes for a given lead byte
717 * - encode a code point <= U+10ffff
718 * - use the fewest possible number of bytes for their code points
719 * - use at most 4 bytes (for i>=5 it is 0x10ffff<utf8_minChar32[])
720 *
721 * Starting with Unicode 3.2, surrogate code points must not be encoded in UTF-8.
722 * There are no irregular sequences any more.
723 */
724 if (isLegalSequence &&
725 (uint32_t)ch <= MAXIMUM_UTF &&
726 (uint32_t)ch >= utf8_minChar32[extraBytesToWrite] &&
727 !U_IS_SURROGATE(ch)
728 ) {
729 return ch; /* return the code point */
730 }
731
732 for(i = 0; sourceInitial < source; ++i) {
733 cnv->toUBytes[i] = *sourceInitial++;
734 }
735 cnv->toULength = i;
736 *err = U_ILLEGAL_CHAR_FOUND;
737 return 0xffff;
738 }
739
740 /* UTF-8-from-UTF-8 conversion functions ------------------------------------ */
741
742 /* minimum code point values for n-byte UTF-8 sequences, n=0..4 */
743 static const UChar32
744 utf8_minLegal[5]={ 0, 0, 0x80, 0x800, 0x10000 };
745
746 /* offsets for n-byte UTF-8 sequences that were calculated with ((lead<<6)+trail)<<6+trail... */
747 static const UChar32
748 utf8_offsets[7]={ 0, 0, 0x3080, 0xE2080, 0x3C82080 };
749
750 /* "Convert" UTF-8 to UTF-8: Validate and copy. Modified from ucnv_DBCSFromUTF8(). */
751 static void
ucnv_UTF8FromUTF8(UConverterFromUnicodeArgs * pFromUArgs,UConverterToUnicodeArgs * pToUArgs,UErrorCode * pErrorCode)752 ucnv_UTF8FromUTF8(UConverterFromUnicodeArgs *pFromUArgs,
753 UConverterToUnicodeArgs *pToUArgs,
754 UErrorCode *pErrorCode) {
755 UConverter *utf8;
756 const uint8_t *source, *sourceLimit;
757 uint8_t *target;
758 int32_t targetCapacity;
759 int32_t count;
760
761 int8_t oldToULength, toULength, toULimit;
762
763 UChar32 c;
764 uint8_t b, t1, t2;
765
766 /* set up the local pointers */
767 utf8=pToUArgs->converter;
768 source=(uint8_t *)pToUArgs->source;
769 sourceLimit=(uint8_t *)pToUArgs->sourceLimit;
770 target=(uint8_t *)pFromUArgs->target;
771 targetCapacity=(int32_t)(pFromUArgs->targetLimit-pFromUArgs->target);
772
773 /* get the converter state from the UTF-8 UConverter */
774 c=(UChar32)utf8->toUnicodeStatus;
775 if(c!=0) {
776 toULength=oldToULength=utf8->toULength;
777 toULimit=(int8_t)utf8->mode;
778 } else {
779 toULength=oldToULength=toULimit=0;
780 }
781
782 count=(int32_t)(sourceLimit-source)+oldToULength;
783 if(count<toULimit) {
784 /*
785 * Not enough input to complete the partial character.
786 * Jump to moreBytes below - it will not output to target.
787 */
788 } else if(targetCapacity<toULimit) {
789 /*
790 * Not enough target capacity to output the partial character.
791 * Let the standard converter handle this.
792 */
793 *pErrorCode=U_USING_DEFAULT_WARNING;
794 return;
795 } else {
796 /*
797 * Use a single counter for source and target, counting the minimum of
798 * the source length and the target capacity.
799 * As a result, the source length is checked only once per multi-byte
800 * character instead of twice.
801 *
802 * Make sure that the last byte sequence is complete, or else
803 * stop just before it.
804 * (The longest legal byte sequence has 3 trail bytes.)
805 * Count oldToULength (number of source bytes from a previous buffer)
806 * into the source length but reduce the source index by toULimit
807 * while going back over trail bytes in order to not go back into
808 * the bytes that will be read for finishing a partial
809 * sequence from the previous buffer.
810 * Let the standard converter handle edge cases.
811 */
812 int32_t i;
813
814 if(count>targetCapacity) {
815 count=targetCapacity;
816 }
817
818 i=0;
819 while(i<3 && i<(count-toULimit)) {
820 b=source[count-oldToULength-i-1];
821 if(U8_IS_TRAIL(b)) {
822 ++i;
823 } else {
824 if(i<U8_COUNT_TRAIL_BYTES(b)) {
825 /* stop converting before the lead byte if there are not enough trail bytes for it */
826 count-=i+1;
827 }
828 break;
829 }
830 }
831 }
832
833 if(c!=0) {
834 utf8->toUnicodeStatus=0;
835 utf8->toULength=0;
836 goto moreBytes;
837 /* See note in ucnv_SBCSFromUTF8() about this goto. */
838 }
839
840 /* conversion loop */
841 while(count>0) {
842 b=*source++;
843 if((int8_t)b>=0) {
844 /* convert ASCII */
845 *target++=b;
846 --count;
847 continue;
848 } else {
849 if(b>0xe0) {
850 if( /* handle U+1000..U+D7FF inline */
851 (t1=source[0]) >= 0x80 && ((b<0xed && (t1 <= 0xbf)) ||
852 (b==0xed && (t1 <= 0x9f))) &&
853 (t2=source[1]) >= 0x80 && t2 <= 0xbf
854 ) {
855 source+=2;
856 *target++=b;
857 *target++=t1;
858 *target++=t2;
859 count-=3;
860 continue;
861 }
862 } else if(b<0xe0) {
863 if( /* handle U+0080..U+07FF inline */
864 b>=0xc2 &&
865 (t1=*source) >= 0x80 && t1 <= 0xbf
866 ) {
867 ++source;
868 *target++=b;
869 *target++=t1;
870 count-=2;
871 continue;
872 }
873 } else if(b==0xe0) {
874 if( /* handle U+0800..U+0FFF inline */
875 (t1=source[0]) >= 0xa0 && t1 <= 0xbf &&
876 (t2=source[1]) >= 0x80 && t2 <= 0xbf
877 ) {
878 source+=2;
879 *target++=b;
880 *target++=t1;
881 *target++=t2;
882 count-=3;
883 continue;
884 }
885 }
886
887 /* handle "complicated" and error cases, and continuing partial characters */
888 oldToULength=0;
889 toULength=1;
890 toULimit=U8_COUNT_TRAIL_BYTES(b)+1;
891 c=b;
892 moreBytes:
893 while(toULength<toULimit) {
894 if(source<sourceLimit) {
895 b=*source;
896 if(U8_IS_TRAIL(b)) {
897 ++source;
898 ++toULength;
899 c=(c<<6)+b;
900 } else {
901 break; /* sequence too short, stop with toULength<toULimit */
902 }
903 } else {
904 /* store the partial UTF-8 character, compatible with the regular UTF-8 converter */
905 source-=(toULength-oldToULength);
906 while(oldToULength<toULength) {
907 utf8->toUBytes[oldToULength++]=*source++;
908 }
909 utf8->toUnicodeStatus=c;
910 utf8->toULength=toULength;
911 utf8->mode=toULimit;
912 pToUArgs->source=(char *)source;
913 pFromUArgs->target=(char *)target;
914 return;
915 }
916 }
917
918 if( toULength==toULimit && /* consumed all trail bytes */
919 (toULength==3 || toULength==2) && /* BMP */
920 (c-=utf8_offsets[toULength])>=utf8_minLegal[toULength] &&
921 (c<=0xd7ff || 0xe000<=c) /* not a surrogate */
922 ) {
923 /* legal byte sequence for BMP code point */
924 } else if(
925 toULength==toULimit && toULength==4 &&
926 (0x10000<=(c-=utf8_offsets[4]) && c<=0x10ffff)
927 ) {
928 /* legal byte sequence for supplementary code point */
929 } else {
930 /* error handling: illegal UTF-8 byte sequence */
931 source-=(toULength-oldToULength);
932 while(oldToULength<toULength) {
933 utf8->toUBytes[oldToULength++]=*source++;
934 }
935 utf8->toULength=toULength;
936 pToUArgs->source=(char *)source;
937 pFromUArgs->target=(char *)target;
938 *pErrorCode=U_ILLEGAL_CHAR_FOUND;
939 return;
940 }
941
942 /* copy the legal byte sequence to the target */
943 {
944 int8_t i;
945
946 for(i=0; i<oldToULength; ++i) {
947 *target++=utf8->toUBytes[i];
948 }
949 source-=(toULength-oldToULength);
950 for(; i<toULength; ++i) {
951 *target++=*source++;
952 }
953 count-=toULength;
954 }
955 }
956 }
957
958 if(U_SUCCESS(*pErrorCode) && source<sourceLimit) {
959 if(target==(const uint8_t *)pFromUArgs->targetLimit) {
960 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
961 } else {
962 b=*source;
963 toULimit=U8_COUNT_TRAIL_BYTES(b)+1;
964 if(toULimit>(sourceLimit-source)) {
965 /* collect a truncated byte sequence */
966 toULength=0;
967 c=b;
968 for(;;) {
969 utf8->toUBytes[toULength++]=b;
970 if(++source==sourceLimit) {
971 /* partial byte sequence at end of source */
972 utf8->toUnicodeStatus=c;
973 utf8->toULength=toULength;
974 utf8->mode=toULimit;
975 break;
976 } else if(!U8_IS_TRAIL(b=*source)) {
977 /* lead byte in trail byte position */
978 utf8->toULength=toULength;
979 *pErrorCode=U_ILLEGAL_CHAR_FOUND;
980 break;
981 }
982 c=(c<<6)+b;
983 }
984 } else {
985 /* partial-sequence target overflow: fall back to the pivoting implementation */
986 *pErrorCode=U_USING_DEFAULT_WARNING;
987 }
988 }
989 }
990
991 /* write back the updated pointers */
992 pToUArgs->source=(char *)source;
993 pFromUArgs->target=(char *)target;
994 }
995
996 /* UTF-8 converter data ----------------------------------------------------- */
997
998 static const UConverterImpl _UTF8Impl={
999 UCNV_UTF8,
1000
1001 NULL,
1002 NULL,
1003
1004 NULL,
1005 NULL,
1006 NULL,
1007
1008 ucnv_toUnicode_UTF8,
1009 ucnv_toUnicode_UTF8_OFFSETS_LOGIC,
1010 ucnv_fromUnicode_UTF8,
1011 ucnv_fromUnicode_UTF8_OFFSETS_LOGIC,
1012 ucnv_getNextUChar_UTF8,
1013
1014 NULL,
1015 NULL,
1016 NULL,
1017 NULL,
1018 ucnv_getNonSurrogateUnicodeSet,
1019
1020 ucnv_UTF8FromUTF8,
1021 ucnv_UTF8FromUTF8
1022 };
1023
1024 /* The 1208 CCSID refers to any version of Unicode of UTF-8 */
1025 static const UConverterStaticData _UTF8StaticData={
1026 sizeof(UConverterStaticData),
1027 "UTF-8",
1028 1208, UCNV_IBM, UCNV_UTF8,
1029 1, 3, /* max 3 bytes per UChar from UTF-8 (4 bytes from surrogate _pair_) */
1030 { 0xef, 0xbf, 0xbd, 0 },3,FALSE,FALSE,
1031 0,
1032 0,
1033 { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */
1034 };
1035
1036
1037 const UConverterSharedData _UTF8Data={
1038 sizeof(UConverterSharedData), ~((uint32_t) 0),
1039 NULL, NULL, &_UTF8StaticData, FALSE, &_UTF8Impl,
1040 0
1041 };
1042
1043 /* CESU-8 converter data ---------------------------------------------------- */
1044
1045 static const UConverterImpl _CESU8Impl={
1046 UCNV_CESU8,
1047
1048 NULL,
1049 NULL,
1050
1051 NULL,
1052 NULL,
1053 NULL,
1054
1055 ucnv_toUnicode_UTF8,
1056 ucnv_toUnicode_UTF8_OFFSETS_LOGIC,
1057 ucnv_fromUnicode_UTF8,
1058 ucnv_fromUnicode_UTF8_OFFSETS_LOGIC,
1059 NULL,
1060
1061 NULL,
1062 NULL,
1063 NULL,
1064 NULL,
1065 ucnv_getCompleteUnicodeSet
1066 };
1067
1068 static const UConverterStaticData _CESU8StaticData={
1069 sizeof(UConverterStaticData),
1070 "CESU-8",
1071 9400, /* CCSID for CESU-8 */
1072 UCNV_UNKNOWN, UCNV_CESU8, 1, 3,
1073 { 0xef, 0xbf, 0xbd, 0 },3,FALSE,FALSE,
1074 0,
1075 0,
1076 { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */
1077 };
1078
1079
1080 const UConverterSharedData _CESU8Data={
1081 sizeof(UConverterSharedData), ~((uint32_t) 0),
1082 NULL, NULL, &_CESU8StaticData, FALSE, &_CESU8Impl,
1083 0
1084 };
1085
1086 #endif
1087