1 /*
2 * Copyright 2001-2004 Unicode, Inc.
3 *
4 * Disclaimer
5 *
6 * This source code is provided as is by Unicode, Inc. No claims are
7 * made as to fitness for any particular purpose. No warranties of any
8 * kind are expressed or implied. The recipient agrees to determine
9 * applicability of information provided. If this file has been
10 * purchased on magnetic or optical media from Unicode, Inc., the
11 * sole remedy for any claim will be exchange of defective media
12 * within 90 days of receipt.
13 *
14 * Limitations on Rights to Redistribute This Code
15 *
16 * Unicode, Inc. hereby grants the right to freely use the information
17 * supplied in this file in the creation of products supporting the
18 * Unicode Standard, and to make copies of this file in any form
19 * for internal or external distribution as long as this notice
20 * remains attached.
21 */
22
23 /* ---------------------------------------------------------------------
24
25 Conversions between UTF32, UTF-16, and UTF-8. Source code file.
26 Author: Mark E. Davis, 1994.
27 Rev History: Rick McGowan, fixes & updates May 2001.
28 Sept 2001: fixed const & error conditions per
29 mods suggested by S. Parent & A. Lillich.
30 June 2002: Tim Dodd added detection and handling of incomplete
31 source sequences, enhanced error detection, added casts
32 to eliminate compiler warnings.
33 July 2003: slight mods to back out aggressive FFFE detection.
34 Jan 2004: updated switches in from-UTF8 conversions.
35 Oct 2004: updated to use UNI_MAX_LEGAL_UTF32 in UTF-32 conversions.
36 May 2006: updated isLegalUTF8Sequence.
37
38 See the header file "ConvertUTF.h" for complete documentation.
39
40 ------------------------------------------------------------------------ */
41
42 #ifdef CVTUTF_DEBUG
43 #include <stdio.h>
44 #endif
45 #include <string.h> /* strlen () */
46
47 #include "transmission.h"
48
49 #include "ConvertUTF.h"
50
51 static const int halfShift = 10; /* used for shifting by 10 bits */
52
53 static const UTF32 halfBase = 0x0010000UL;
54 static const UTF32 halfMask = 0x3FFUL;
55
56 #define UNI_SUR_HIGH_START (UTF32)0xD800
57 #define UNI_SUR_HIGH_END (UTF32)0xDBFF
58 #define UNI_SUR_LOW_START (UTF32)0xDC00
59 #define UNI_SUR_LOW_END (UTF32)0xDFFF
60 #define false 0
61 #define true 1
62
63 /* --------------------------------------------------------------------- */
64
ConvertUTF32toUTF16(const UTF32 ** sourceStart,const UTF32 * sourceEnd,UTF16 ** targetStart,UTF16 * targetEnd,ConversionFlags flags)65 ConversionResult ConvertUTF32toUTF16 (
66 const UTF32** sourceStart, const UTF32* sourceEnd,
67 UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags) {
68 ConversionResult result = conversionOK;
69 const UTF32* source = *sourceStart;
70 UTF16* target = *targetStart;
71 while (source < sourceEnd) {
72 UTF32 ch;
73 if (target >= targetEnd) {
74 result = targetExhausted; break;
75 }
76 ch = *source++;
77 if (ch <= UNI_MAX_BMP) { /* Target is a character <= 0xFFFF */
78 /* UTF-16 surrogate values are illegal in UTF-32; 0xffff or 0xfffe are both reserved values */
79 if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) {
80 if (flags == strictConversion) {
81 --source; /* return to the illegal value itself */
82 result = sourceIllegal;
83 break;
84 } else {
85 *target++ = UNI_REPLACEMENT_CHAR;
86 }
87 } else {
88 *target++ = (UTF16)ch; /* normal case */
89 }
90 } else if (ch > UNI_MAX_LEGAL_UTF32) {
91 if (flags == strictConversion) {
92 result = sourceIllegal;
93 } else {
94 *target++ = UNI_REPLACEMENT_CHAR;
95 }
96 } else {
97 /* target is a character in range 0xFFFF - 0x10FFFF. */
98 if (target + 1 >= targetEnd) {
99 --source; /* Back up source pointer! */
100 result = targetExhausted; break;
101 }
102 ch -= halfBase;
103 *target++ = (UTF16)((ch >> halfShift) + UNI_SUR_HIGH_START);
104 *target++ = (UTF16)((ch & halfMask) + UNI_SUR_LOW_START);
105 }
106 }
107 *sourceStart = source;
108 *targetStart = target;
109 return result;
110 }
111
112 /* --------------------------------------------------------------------- */
113
ConvertUTF16toUTF32(const UTF16 ** sourceStart,const UTF16 * sourceEnd,UTF32 ** targetStart,UTF32 * targetEnd,ConversionFlags flags)114 ConversionResult ConvertUTF16toUTF32 (
115 const UTF16** sourceStart, const UTF16* sourceEnd,
116 UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags) {
117 ConversionResult result = conversionOK;
118 const UTF16* source = *sourceStart;
119 UTF32* target = *targetStart;
120 UTF32 ch, ch2;
121 while (source < sourceEnd) {
122 const UTF16* oldSource = source; /* In case we have to back up because of target overflow. */
123 ch = *source++;
124 /* If we have a surrogate pair, convert to UTF32 first. */
125 if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_HIGH_END) {
126 /* If the 16 bits following the high surrogate are in the source buffer... */
127 if (source < sourceEnd) {
128 ch2 = *source;
129 /* If it's a low surrogate, convert to UTF32. */
130 if (ch2 >= UNI_SUR_LOW_START && ch2 <= UNI_SUR_LOW_END) {
131 ch = ((ch - UNI_SUR_HIGH_START) << halfShift)
132 + (ch2 - UNI_SUR_LOW_START) + halfBase;
133 ++source;
134 } else if (flags == strictConversion) { /* it's an unpaired high surrogate */
135 --source; /* return to the illegal value itself */
136 result = sourceIllegal;
137 break;
138 }
139 } else { /* We don't have the 16 bits following the high surrogate. */
140 --source; /* return to the high surrogate */
141 result = sourceExhausted;
142 break;
143 }
144 } else if (flags == strictConversion) {
145 /* UTF-16 surrogate values are illegal in UTF-32 */
146 if (ch >= UNI_SUR_LOW_START && ch <= UNI_SUR_LOW_END) {
147 --source; /* return to the illegal value itself */
148 result = sourceIllegal;
149 break;
150 }
151 }
152 if (target >= targetEnd) {
153 source = oldSource; /* Back up source pointer! */
154 result = targetExhausted; break;
155 }
156 *target++ = ch;
157 }
158 *sourceStart = source;
159 *targetStart = target;
160 #ifdef CVTUTF_DEBUG
161 if (result == sourceIllegal) {
162 fprintf (stderr, "ConvertUTF16toUTF32 illegal seq 0x%04x,%04x\n", ch, ch2);
163 fflush (stderr);
164 }
165 #endif
166 return result;
167 }
168
169 /* --------------------------------------------------------------------- */
170
171 /*
172 * Index into the table below with the first byte of a UTF-8 sequence to
173 * get the number of trailing bytes that are supposed to follow it.
174 * Note that *legal* UTF-8 values can't have 4 or 5-bytes. The table is
175 * left as-is for anyone who may want to do such conversion, which was
176 * allowed in earlier algorithms.
177 */
178 static const char trailingBytesForUTF8[256] = {
179 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,
180 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,
181 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,
182 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,
183 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,
184 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,
185 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,
186 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 3,3,3,3,3,3,3,3,4,4,4,4,5,5,5,5
187 };
188
189 /*
190 * Magic values subtracted from a buffer value during UTF8 conversion.
191 * This table contains as many values as there might be trailing bytes
192 * in a UTF-8 sequence.
193 */
194 static const UTF32 offsetsFromUTF8[6] = { 0x00000000UL, 0x00003080UL, 0x000E2080UL,
195 0x03C82080UL, 0xFA082080UL, 0x82082080UL };
196
197 /*
198 * Once the bits are split out into bytes of UTF-8, this is a mask OR-ed
199 * into the first byte, depending on how many bytes follow. There are
200 * as many entries in this table as there are UTF-8 sequence types.
201 * (I.e., one byte sequence, two byte... etc.). Remember that sequencs
202 * for *legal* UTF-8 will be 4 or fewer bytes total.
203 */
204 static const UTF8 firstByteMark[7] = { 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC };
205
206 /* --------------------------------------------------------------------- */
207
208 /* The interface converts a whole buffer to avoid function-call overhead.
209 * Constants have been gathered. Loops & conditionals have been removed as
210 * much as possible for efficiency, in favor of drop-through switches.
211 * (See "Note A" at the bottom of the file for equivalent code.)
212 * If your compiler supports it, the "isLegalUTF8" call can be turned
213 * into an inline function.
214 */
215
216 /* --------------------------------------------------------------------- */
217
ConvertUTF16toUTF8(const UTF16 ** sourceStart,const UTF16 * sourceEnd,UTF8 ** targetStart,UTF8 * targetEnd,ConversionFlags flags)218 ConversionResult ConvertUTF16toUTF8 (
219 const UTF16** sourceStart, const UTF16* sourceEnd,
220 UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags) {
221 ConversionResult result = conversionOK;
222 const UTF16* source = *sourceStart;
223 UTF8* target = *targetStart;
224 while (source < sourceEnd) {
225 UTF32 ch;
226 unsigned short bytesToWrite = 0;
227 const UTF32 byteMask = 0xBF;
228 const UTF32 byteMark = 0x80;
229 const UTF16* oldSource = source; /* In case we have to back up because of target overflow. */
230 ch = *source++;
231 /* If we have a surrogate pair, convert to UTF32 first. */
232 if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_HIGH_END) {
233 /* If the 16 bits following the high surrogate are in the source buffer... */
234 if (source < sourceEnd) {
235 UTF32 ch2 = *source;
236 /* If it's a low surrogate, convert to UTF32. */
237 if (ch2 >= UNI_SUR_LOW_START && ch2 <= UNI_SUR_LOW_END) {
238 ch = ((ch - UNI_SUR_HIGH_START) << halfShift)
239 + (ch2 - UNI_SUR_LOW_START) + halfBase;
240 ++source;
241 } else if (flags == strictConversion) { /* it's an unpaired high surrogate */
242 --source; /* return to the illegal value itself */
243 result = sourceIllegal;
244 break;
245 }
246 } else { /* We don't have the 16 bits following the high surrogate. */
247 --source; /* return to the high surrogate */
248 result = sourceExhausted;
249 break;
250 }
251 } else if (flags == strictConversion) {
252 /* UTF-16 surrogate values are illegal in UTF-32 */
253 if (ch >= UNI_SUR_LOW_START && ch <= UNI_SUR_LOW_END) {
254 --source; /* return to the illegal value itself */
255 result = sourceIllegal;
256 break;
257 }
258 }
259 /* Figure out how many bytes the result will require */
260 if (ch < (UTF32)0x80) { bytesToWrite = 1;
261 } else if (ch < (UTF32)0x800) { bytesToWrite = 2;
262 } else if (ch < (UTF32)0x10000) { bytesToWrite = 3;
263 } else if (ch < (UTF32)0x110000) { bytesToWrite = 4;
264 } else { bytesToWrite = 3;
265 ch = UNI_REPLACEMENT_CHAR;
266 }
267
268 target += bytesToWrite;
269 if (target > targetEnd) {
270 source = oldSource; /* Back up source pointer! */
271 target -= bytesToWrite; result = targetExhausted; break;
272 }
273 switch (bytesToWrite) { /* note: everything falls through. */
274 case 4: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
275 case 3: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
276 case 2: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
277 case 1: *--target = (UTF8)(ch | firstByteMark[bytesToWrite]);
278 }
279 target += bytesToWrite;
280 }
281 *sourceStart = source;
282 *targetStart = target;
283 return result;
284 }
285
286 /* --------------------------------------------------------------------- */
287
288 /*
289 * Utility routine to tell whether a sequence of bytes is legal UTF-8.
290 * This must be called with the length pre-determined by the first byte.
291 * If not calling this from ConvertUTF8to*, then the length can be set by:
292 * length = trailingBytesForUTF8[*source]+1;
293 * and the sequence is illegal right away if there aren't that many bytes
294 * available.
295 * If presented with a length > 4, this returns false. The Unicode
296 * definition of UTF-8 goes up to 4-byte sequences.
297 */
298
isLegalUTF8(const UTF8 * source,int length)299 static Boolean isLegalUTF8 (const UTF8 *source, int length) {
300 UTF8 a;
301 const UTF8 *srcptr = source+length;
302 switch (length) {
303 default: return false;
304 /* Everything else falls through when "true"... */
305 case 4: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
306 case 3: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
307 case 2: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
308
309 switch (*source) {
310 /* no fall-through in this inner switch */
311 case 0xE0: if (a < 0xA0) return false; break;
312 case 0xED: if (a > 0x9F) return false; break;
313 case 0xF0: if (a < 0x90) return false; break;
314 case 0xF4: if (a > 0x8F) return false; break;
315 default: if (a < 0x80) return false;
316 }
317
318 case 1: if (*source >= 0x80 && *source < 0xC2) return false;
319 }
320 if (*source > 0xF4) return false;
321 return true;
322 }
323
324 /* --------------------------------------------------------------------- */
325
326 /*
327 * Exported function to return whether a UTF-8 sequence is legal or not.
328 * This is not used here; it's just exported.
329 */
330
isLegalUTF8Sequence(const UTF8 * source,const UTF8 * sourceEnd)331 Boolean isLegalUTF8Sequence (const UTF8 *source, const UTF8 *sourceEnd) {
332 int length;
333 if (source == sourceEnd) {
334 return true;
335 }
336 while (true) {
337 length = trailingBytesForUTF8[*source]+1;
338 if (source+length > sourceEnd) {
339 return false;
340 }
341 if (!isLegalUTF8 (source, length)) {
342 return false;
343 }
344 source += length;
345 if (source >= sourceEnd) {
346 return true;
347 }
348 }
349 }
350
351 /**
352 * This is a variation of isLegalUTF8Sequence () that behaves like g_utf8_validate ().
353 * In addition to knowing if the sequence is legal, it also tells you the last good character.
354 */
355 Boolean
tr_utf8_validate(const char * str,size_t max_len,const char ** end)356 tr_utf8_validate (const char * str, size_t max_len, const char ** end)
357 {
358 const UTF8* source = (const UTF8*) str;
359 const UTF8* sourceEnd;
360
361 if (max_len == 0)
362 return true;
363
364 if (str == NULL)
365 return false;
366
367 sourceEnd = source + (max_len == TR_BAD_SIZE ? strlen (str) : max_len);
368
369 if (source == sourceEnd)
370 {
371 if (end != NULL)
372 *end = (const char*) source;
373 return true;
374 }
375
376 for (;;)
377 {
378 const int length = trailingBytesForUTF8[*source] + 1;
379 if (source + length > sourceEnd) {
380 if (end != NULL)
381 *end = (const char*) source;
382 return false;
383 }
384 if (!isLegalUTF8 (source, length)) {
385 if (end != NULL)
386 *end = (const char*) source;
387 return false;
388 }
389 source += length;
390 if (source >= sourceEnd) {
391 if (end != NULL)
392 *end = (const char*) source;
393 return true;
394 }
395 }
396 }
397
398
399 /* --------------------------------------------------------------------- */
400
ConvertUTF8toUTF16(const UTF8 ** sourceStart,const UTF8 * sourceEnd,UTF16 ** targetStart,UTF16 * targetEnd,ConversionFlags flags)401 ConversionResult ConvertUTF8toUTF16 (
402 const UTF8** sourceStart, const UTF8* sourceEnd,
403 UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags) {
404 ConversionResult result = conversionOK;
405 const UTF8* source = *sourceStart;
406 UTF16* target = *targetStart;
407 while (source < sourceEnd) {
408 UTF32 ch = 0;
409 unsigned short extraBytesToRead = trailingBytesForUTF8[*source];
410 if (extraBytesToRead >= sourceEnd - source) {
411 result = sourceExhausted; break;
412 }
413 /* Do this check whether lenient or strict */
414 if (! isLegalUTF8 (source, extraBytesToRead+1)) {
415 result = sourceIllegal;
416 break;
417 }
418 /*
419 * The cases all fall through. See "Note A" below.
420 */
421 switch (extraBytesToRead) {
422 case 5: ch += *source++; ch <<= 6; /* remember, illegal UTF-8 */
423 case 4: ch += *source++; ch <<= 6; /* remember, illegal UTF-8 */
424 case 3: ch += *source++; ch <<= 6;
425 case 2: ch += *source++; ch <<= 6;
426 case 1: ch += *source++; ch <<= 6;
427 case 0: ch += *source++;
428 }
429 ch -= offsetsFromUTF8[extraBytesToRead];
430
431 if (target >= targetEnd) {
432 source -= (extraBytesToRead+1); /* Back up source pointer! */
433 result = targetExhausted; break;
434 }
435 if (ch <= UNI_MAX_BMP) { /* Target is a character <= 0xFFFF */
436 /* UTF-16 surrogate values are illegal in UTF-32 */
437 if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) {
438 if (flags == strictConversion) {
439 source -= (extraBytesToRead+1); /* return to the illegal value itself */
440 result = sourceIllegal;
441 break;
442 } else {
443 *target++ = UNI_REPLACEMENT_CHAR;
444 }
445 } else {
446 *target++ = (UTF16)ch; /* normal case */
447 }
448 } else if (ch > UNI_MAX_UTF16) {
449 if (flags == strictConversion) {
450 result = sourceIllegal;
451 source -= (extraBytesToRead+1); /* return to the start */
452 break; /* Bail out; shouldn't continue */
453 } else {
454 *target++ = UNI_REPLACEMENT_CHAR;
455 }
456 } else {
457 /* target is a character in range 0xFFFF - 0x10FFFF. */
458 if (target + 1 >= targetEnd) {
459 source -= (extraBytesToRead+1); /* Back up source pointer! */
460 result = targetExhausted; break;
461 }
462 ch -= halfBase;
463 *target++ = (UTF16)((ch >> halfShift) + UNI_SUR_HIGH_START);
464 *target++ = (UTF16)((ch & halfMask) + UNI_SUR_LOW_START);
465 }
466 }
467 *sourceStart = source;
468 *targetStart = target;
469 return result;
470 }
471
472 /* --------------------------------------------------------------------- */
473
ConvertUTF32toUTF8(const UTF32 ** sourceStart,const UTF32 * sourceEnd,UTF8 ** targetStart,UTF8 * targetEnd,ConversionFlags flags)474 ConversionResult ConvertUTF32toUTF8 (
475 const UTF32** sourceStart, const UTF32* sourceEnd,
476 UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags) {
477 ConversionResult result = conversionOK;
478 const UTF32* source = *sourceStart;
479 UTF8* target = *targetStart;
480 while (source < sourceEnd) {
481 UTF32 ch;
482 unsigned short bytesToWrite = 0;
483 const UTF32 byteMask = 0xBF;
484 const UTF32 byteMark = 0x80;
485 ch = *source++;
486 if (flags == strictConversion) {
487 /* UTF-16 surrogate values are illegal in UTF-32 */
488 if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) {
489 --source; /* return to the illegal value itself */
490 result = sourceIllegal;
491 break;
492 }
493 }
494 /*
495 * Figure out how many bytes the result will require. Turn any
496 * illegally large UTF32 things (> Plane 17) into replacement chars.
497 */
498 if (ch < (UTF32)0x80) { bytesToWrite = 1;
499 } else if (ch < (UTF32)0x800) { bytesToWrite = 2;
500 } else if (ch < (UTF32)0x10000) { bytesToWrite = 3;
501 } else if (ch <= UNI_MAX_LEGAL_UTF32) { bytesToWrite = 4;
502 } else { bytesToWrite = 3;
503 ch = UNI_REPLACEMENT_CHAR;
504 result = sourceIllegal;
505 }
506
507 target += bytesToWrite;
508 if (target > targetEnd) {
509 --source; /* Back up source pointer! */
510 target -= bytesToWrite; result = targetExhausted; break;
511 }
512 switch (bytesToWrite) { /* note: everything falls through. */
513 case 4: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
514 case 3: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
515 case 2: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
516 case 1: *--target = (UTF8)(ch | firstByteMark[bytesToWrite]);
517 }
518 target += bytesToWrite;
519 }
520 *sourceStart = source;
521 *targetStart = target;
522 return result;
523 }
524
525 /* --------------------------------------------------------------------- */
526
ConvertUTF8toUTF32(const UTF8 ** sourceStart,const UTF8 * sourceEnd,UTF32 ** targetStart,UTF32 * targetEnd,ConversionFlags flags)527 ConversionResult ConvertUTF8toUTF32 (
528 const UTF8** sourceStart, const UTF8* sourceEnd,
529 UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags) {
530 ConversionResult result = conversionOK;
531 const UTF8* source = *sourceStart;
532 UTF32* target = *targetStart;
533 while (source < sourceEnd) {
534 UTF32 ch = 0;
535 unsigned short extraBytesToRead = trailingBytesForUTF8[*source];
536 if (extraBytesToRead >= sourceEnd - source) {
537 result = sourceExhausted; break;
538 }
539 /* Do this check whether lenient or strict */
540 if (! isLegalUTF8 (source, extraBytesToRead+1)) {
541 result = sourceIllegal;
542 break;
543 }
544 /*
545 * The cases all fall through. See "Note A" below.
546 */
547 switch (extraBytesToRead) {
548 case 5: ch += *source++; ch <<= 6;
549 case 4: ch += *source++; ch <<= 6;
550 case 3: ch += *source++; ch <<= 6;
551 case 2: ch += *source++; ch <<= 6;
552 case 1: ch += *source++; ch <<= 6;
553 case 0: ch += *source++;
554 }
555 ch -= offsetsFromUTF8[extraBytesToRead];
556
557 if (target >= targetEnd) {
558 source -= (extraBytesToRead+1); /* Back up the source pointer! */
559 result = targetExhausted; break;
560 }
561 if (ch <= UNI_MAX_LEGAL_UTF32) {
562 /*
563 * UTF-16 surrogate values are illegal in UTF-32, and anything
564 * over Plane 17 (> 0x10FFFF) is illegal.
565 */
566 if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) {
567 if (flags == strictConversion) {
568 source -= (extraBytesToRead+1); /* return to the illegal value itself */
569 result = sourceIllegal;
570 break;
571 } else {
572 *target++ = UNI_REPLACEMENT_CHAR;
573 }
574 } else {
575 *target++ = ch;
576 }
577 } else { /* i.e., ch > UNI_MAX_LEGAL_UTF32 */
578 result = sourceIllegal;
579 *target++ = UNI_REPLACEMENT_CHAR;
580 }
581 }
582 *sourceStart = source;
583 *targetStart = target;
584 return result;
585 }
586
587 /* ---------------------------------------------------------------------
588
589 Note A.
590 The fall-through switches in UTF-8 reading code save a
591 temp variable, some decrements & conditionals. The switches
592 are equivalent to the following loop:
593 {
594 int tmpBytesToRead = extraBytesToRead+1;
595 do {
596 ch += *source++;
597 --tmpBytesToRead;
598 if (tmpBytesToRead) ch <<= 6;
599 } while (tmpBytesToRead > 0);
600 }
601 In UTF-8 writing code, the switches on "bytesToWrite" are
602 similarly unrolled loops.
603
604 --------------------------------------------------------------------- */
605