]>
Commit | Line | Data |
---|---|---|
ef416fc2 | 1 | /* |
31db8ded | 2 | * "$Id: transcode.c 9306 2010-09-16 21:43:57Z mike $" |
ef416fc2 | 3 | * |
71e16022 | 4 | * Transcoding support for CUPS. |
ef416fc2 | 5 | * |
71e16022 | 6 | * Copyright 2007-2010 by Apple Inc. |
b86bc4cf | 7 | * Copyright 1997-2007 by Easy Software Products. |
ef416fc2 | 8 | * |
bc44d920 | 9 | * These coded instructions, statements, and computer programs are the |
10 | * property of Apple Inc. and are protected by Federal copyright | |
11 | * law. Distribution and use rights are outlined in the file "LICENSE.txt" | |
12 | * which should have been included with this file. If this file is | |
13 | * file is missing or damaged, see the license at "http://www.cups.org/". | |
ef416fc2 | 14 | * |
bc44d920 | 15 | * This file is subject to the Apple OS-Developed Software exception. |
ef416fc2 | 16 | * |
17 | * Contents: | |
18 | * | |
fa73b229 | 19 | * _cupsCharmapFlush() - Flush all character set maps out of cache. |
ef416fc2 | 20 | * cupsCharsetToUTF8() - Convert legacy character set to UTF-8. |
e1d6a774 | 21 | * cupsUTF8ToCharset() - Convert UTF-8 to legacy character set. |
ef416fc2 | 22 | * cupsUTF8ToUTF32() - Convert UTF-8 to UTF-32. |
23 | * cupsUTF32ToUTF8() - Convert UTF-32 to UTF-8. | |
ef416fc2 | 24 | */ |
25 | ||
26 | /* | |
27 | * Include necessary headers... | |
28 | */ | |
29 | ||
71e16022 | 30 | #include "cups-private.h" |
e53920b9 | 31 | #include <limits.h> |
ef416fc2 | 32 | #include <time.h> |
cc754834 MS |
33 | #ifdef HAVE_ICONV_H |
34 | # include <iconv.h> | |
35 | #endif /* HAVE_ICONV_H */ | |
ef416fc2 | 36 | |
37 | ||
d6ae789d | 38 | /* |
39 | * Local globals... | |
40 | */ | |
41 | ||
cc754834 | 42 | #ifdef HAVE_ICONV_H |
6d2f911b | 43 | static _cups_mutex_t map_mutex = _CUPS_MUTEX_INITIALIZER; |
d6ae789d | 44 | /* Mutex to control access to maps */ |
cc754834 MS |
45 | static iconv_t map_from_utf8 = (iconv_t)-1; |
46 | /* Convert from UTF-8 to charset */ | |
47 | static iconv_t map_to_utf8 = (iconv_t)-1; | |
48 | /* Convert from charset to UTF-8 */ | |
49 | static cups_encoding_t map_encoding = CUPS_AUTO_ENCODING; | |
50 | /* Which charset is cached */ | |
51 | #endif /* HAVE_ICONV_H */ | |
e1d6a774 | 52 | |
ef416fc2 | 53 | |
54 | /* | |
e1d6a774 | 55 | * '_cupsCharmapFlush()' - Flush all character set maps out of cache. |
ef416fc2 | 56 | */ |
57 | ||
e1d6a774 | 58 | void |
d6ae789d | 59 | _cupsCharmapFlush(void) |
ef416fc2 | 60 | { |
cc754834 MS |
61 | #ifdef HAVE_ICONV_H |
62 | if (map_from_utf8 != (iconv_t)-1) | |
ef416fc2 | 63 | { |
cc754834 MS |
64 | iconv_close(map_from_utf8); |
65 | map_from_utf8 = (iconv_t)-1; | |
ef416fc2 | 66 | } |
d6ae789d | 67 | |
cc754834 | 68 | if (map_to_utf8 != (iconv_t)-1) |
ef416fc2 | 69 | { |
cc754834 MS |
70 | iconv_close(map_to_utf8); |
71 | map_to_utf8 = (iconv_t)-1; | |
ef416fc2 | 72 | } |
ef416fc2 | 73 | |
cc754834 MS |
74 | map_encoding = CUPS_AUTO_ENCODING; |
75 | #endif /* HAVE_ICONV_H */ | |
ef416fc2 | 76 | } |
77 | ||
e1d6a774 | 78 | |
ef416fc2 | 79 | /* |
e1d6a774 | 80 | * 'cupsCharsetToUTF8()' - Convert legacy character set to UTF-8. |
ef416fc2 | 81 | */ |
e1d6a774 | 82 | |
83 | int /* O - Count or -1 on error */ | |
84 | cupsCharsetToUTF8( | |
cc754834 MS |
85 | cups_utf8_t *dest, /* O - Target string */ |
86 | const char *src, /* I - Source string */ | |
87 | const int maxout, /* I - Max output */ | |
e1d6a774 | 88 | const cups_encoding_t encoding) /* I - Encoding */ |
ef416fc2 | 89 | { |
cc754834 | 90 | cups_utf8_t *destptr; /* Pointer into UTF-8 buffer */ |
7cf5915e | 91 | #ifdef HAVE_ICONV_H |
cc754834 MS |
92 | size_t srclen, /* Length of source string */ |
93 | outBytesLeft; /* Bytes remaining in output buffer */ | |
7cf5915e | 94 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 95 | |
96 | ||
ef416fc2 | 97 | /* |
98 | * Check for valid arguments... | |
99 | */ | |
100 | ||
f11a948a | 101 | DEBUG_printf(("2cupsCharsetToUTF8(dest=%p, src=\"%s\", maxout=%d, encoding=%d)", |
e1d6a774 | 102 | dest, src, maxout, encoding)); |
103 | ||
cc754834 | 104 | if (!dest || !src || maxout < 1) |
e1d6a774 | 105 | { |
cc754834 MS |
106 | if (dest) |
107 | *dest = '\0'; | |
108 | ||
f11a948a | 109 | DEBUG_puts("3cupsCharsetToUTF8: Bad arguments, returning -1"); |
ef416fc2 | 110 | return (-1); |
e1d6a774 | 111 | } |
ef416fc2 | 112 | |
113 | /* | |
114 | * Handle identity conversions... | |
115 | */ | |
116 | ||
cc754834 MS |
117 | if (encoding == CUPS_UTF8 || encoding <= CUPS_US_ASCII || |
118 | encoding >= CUPS_ENCODING_VBCS_END) | |
ef416fc2 | 119 | { |
e1d6a774 | 120 | strlcpy((char *)dest, src, maxout); |
b86bc4cf | 121 | return ((int)strlen((char *)dest)); |
ef416fc2 | 122 | } |
123 | ||
411affcf | 124 | /* |
125 | * Handle ISO-8859-1 to UTF-8 directly... | |
126 | */ | |
127 | ||
cc754834 MS |
128 | destptr = dest; |
129 | ||
411affcf | 130 | if (encoding == CUPS_ISO8859_1) |
131 | { | |
132 | int ch; /* Character from string */ | |
cc754834 | 133 | cups_utf8_t *destend; /* End of UTF-8 buffer */ |
411affcf | 134 | |
135 | ||
411affcf | 136 | destend = dest + maxout - 2; |
137 | ||
138 | while (*src && destptr < destend) | |
139 | { | |
140 | ch = *src++ & 255; | |
141 | ||
142 | if (ch & 128) | |
143 | { | |
144 | *destptr++ = 0xc0 | (ch >> 6); | |
145 | *destptr++ = 0x80 | (ch & 0x3f); | |
146 | } | |
147 | else | |
148 | *destptr++ = ch; | |
149 | } | |
150 | ||
151 | *destptr = '\0'; | |
152 | ||
b86bc4cf | 153 | return ((int)(destptr - dest)); |
411affcf | 154 | } |
155 | ||
ef416fc2 | 156 | /* |
e1d6a774 | 157 | * Convert input legacy charset to UTF-8... |
ef416fc2 | 158 | */ |
e1d6a774 | 159 | |
cc754834 | 160 | #ifdef HAVE_ICONV_H |
6d2f911b | 161 | _cupsMutexLock(&map_mutex); |
d6ae789d | 162 | |
cc754834 MS |
163 | if (map_encoding != encoding) |
164 | { | |
165 | _cupsCharmapFlush(); | |
166 | ||
167 | map_from_utf8 = iconv_open(_cupsEncodingName(encoding), "UTF-8"); | |
168 | map_to_utf8 = iconv_open("UTF-8", _cupsEncodingName(encoding)); | |
169 | map_encoding = encoding; | |
170 | } | |
171 | ||
172 | if (map_to_utf8 != (iconv_t)-1) | |
173 | { | |
174 | srclen = strlen(src); | |
175 | outBytesLeft = maxout - 1; | |
4220952d MS |
176 | |
177 | iconv(map_to_utf8, (char **)&src, &srclen, (char **)&destptr, | |
178 | &outBytesLeft); | |
179 | *destptr = '\0'; | |
cc754834 MS |
180 | |
181 | _cupsMutexUnlock(&map_mutex); | |
182 | ||
183 | return ((int)(destptr - dest)); | |
184 | } | |
d6ae789d | 185 | |
6d2f911b | 186 | _cupsMutexUnlock(&map_mutex); |
cc754834 | 187 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 188 | |
cc754834 MS |
189 | /* |
190 | * No iconv() support, so error out... | |
191 | */ | |
192 | ||
193 | *destptr = '\0'; | |
194 | ||
195 | return (-1); | |
ef416fc2 | 196 | } |
197 | ||
e1d6a774 | 198 | |
ef416fc2 | 199 | /* |
e1d6a774 | 200 | * 'cupsUTF8ToCharset()' - Convert UTF-8 to legacy character set. |
ef416fc2 | 201 | */ |
e1d6a774 | 202 | |
203 | int /* O - Count or -1 on error */ | |
204 | cupsUTF8ToCharset( | |
205 | char *dest, /* O - Target string */ | |
206 | const cups_utf8_t *src, /* I - Source string */ | |
207 | const int maxout, /* I - Max output */ | |
208 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 209 | { |
cc754834 | 210 | char *destptr; /* Pointer into destination */ |
7cf5915e | 211 | #ifdef HAVE_ICONV_H |
cc754834 MS |
212 | size_t srclen, /* Length of source string */ |
213 | outBytesLeft; /* Bytes remaining in output buffer */ | |
7cf5915e | 214 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 215 | |
216 | ||
ef416fc2 | 217 | /* |
218 | * Check for valid arguments... | |
219 | */ | |
220 | ||
cc754834 | 221 | if (!dest || !src || maxout < 1) |
e1d6a774 | 222 | { |
223 | if (dest) | |
224 | *dest = '\0'; | |
225 | ||
ef416fc2 | 226 | return (-1); |
e1d6a774 | 227 | } |
ef416fc2 | 228 | |
229 | /* | |
230 | * Handle identity conversions... | |
231 | */ | |
232 | ||
cc754834 MS |
233 | if (encoding == CUPS_UTF8 || encoding <= CUPS_US_ASCII || |
234 | encoding >= CUPS_ENCODING_VBCS_END) | |
ef416fc2 | 235 | { |
e1d6a774 | 236 | strlcpy(dest, (char *)src, maxout); |
b86bc4cf | 237 | return ((int)strlen(dest)); |
ef416fc2 | 238 | } |
239 | ||
411affcf | 240 | /* |
241 | * Handle UTF-8 to ISO-8859-1 directly... | |
242 | */ | |
243 | ||
cc754834 MS |
244 | destptr = dest; |
245 | ||
411affcf | 246 | if (encoding == CUPS_ISO8859_1) |
247 | { | |
248 | int ch; /* Character from string */ | |
cc754834 | 249 | char *destend; /* End of ISO-8859-1 buffer */ |
411affcf | 250 | |
251 | ||
411affcf | 252 | destend = dest + maxout - 1; |
253 | ||
254 | while (*src && destptr < destend) | |
255 | { | |
256 | ch = *src++; | |
257 | ||
258 | if ((ch & 0xe0) == 0xc0) | |
259 | { | |
260 | ch = ((ch & 0x1f) << 6) | (*src++ & 0x3f); | |
261 | ||
262 | if (ch < 256) | |
263 | *destptr++ = ch; | |
264 | else | |
265 | *destptr++ = '?'; | |
266 | } | |
267 | else if ((ch & 0xf0) == 0xe0 || | |
268 | (ch & 0xf8) == 0xf0) | |
269 | *destptr++ = '?'; | |
270 | else if (!(ch & 0x80)) | |
271 | *destptr++ = ch; | |
272 | } | |
273 | ||
274 | *destptr = '\0'; | |
275 | ||
b86bc4cf | 276 | return ((int)(destptr - dest)); |
411affcf | 277 | } |
278 | ||
cc754834 | 279 | #ifdef HAVE_ICONV_H |
ef416fc2 | 280 | /* |
e1d6a774 | 281 | * Convert input UTF-8 to legacy charset... |
ef416fc2 | 282 | */ |
e1d6a774 | 283 | |
6d2f911b | 284 | _cupsMutexLock(&map_mutex); |
d6ae789d | 285 | |
cc754834 MS |
286 | if (map_encoding != encoding) |
287 | { | |
288 | _cupsCharmapFlush(); | |
289 | ||
290 | map_from_utf8 = iconv_open(_cupsEncodingName(encoding), "UTF-8"); | |
291 | map_to_utf8 = iconv_open("UTF-8", _cupsEncodingName(encoding)); | |
292 | map_encoding = encoding; | |
293 | } | |
294 | ||
295 | if (map_from_utf8 != (iconv_t)-1) | |
296 | { | |
297 | srclen = strlen((char *)src); | |
298 | outBytesLeft = maxout - 1; | |
4220952d MS |
299 | |
300 | iconv(map_from_utf8, (char **)&src, &srclen, &destptr, &outBytesLeft); | |
301 | *destptr = '\0'; | |
cc754834 MS |
302 | |
303 | _cupsMutexUnlock(&map_mutex); | |
304 | ||
305 | return ((int)(destptr - dest)); | |
306 | } | |
d6ae789d | 307 | |
6d2f911b | 308 | _cupsMutexUnlock(&map_mutex); |
cc754834 MS |
309 | #endif /* HAVE_ICONV_H */ |
310 | ||
311 | /* | |
312 | * No iconv() support, so error out... | |
313 | */ | |
314 | ||
315 | *destptr = '\0'; | |
d6ae789d | 316 | |
cc754834 | 317 | return (-1); |
ef416fc2 | 318 | } |
319 | ||
ef416fc2 | 320 | |
321 | /* | |
322 | * 'cupsUTF8ToUTF32()' - Convert UTF-8 to UTF-32. | |
323 | * | |
324 | * 32-bit UTF-32 (actually 21-bit) maps to UTF-8 as follows... | |
325 | * | |
326 | * UTF-32 char UTF-8 char(s) | |
327 | * -------------------------------------------------- | |
e1d6a774 | 328 | * 0 to 127 = 0xxxxxxx (US-ASCII) |
ef416fc2 | 329 | * 128 to 2047 = 110xxxxx 10yyyyyy |
330 | * 2048 to 65535 = 1110xxxx 10yyyyyy 10zzzzzz | |
e1d6a774 | 331 | * > 65535 = 11110xxx 10yyyyyy 10zzzzzz 10xxxxxx |
ef416fc2 | 332 | * |
333 | * UTF-32 prohibits chars beyond Plane 16 (> 0x10ffff) in UCS-4, | |
334 | * which would convert to five- or six-octet UTF-8 sequences... | |
ef416fc2 | 335 | */ |
e1d6a774 | 336 | |
337 | int /* O - Count or -1 on error */ | |
338 | cupsUTF8ToUTF32( | |
339 | cups_utf32_t *dest, /* O - Target string */ | |
340 | const cups_utf8_t *src, /* I - Source string */ | |
341 | const int maxout) /* I - Max output */ | |
ef416fc2 | 342 | { |
e1d6a774 | 343 | int i; /* Looping variable */ |
344 | cups_utf8_t ch; /* Character value */ | |
345 | cups_utf8_t next; /* Next character value */ | |
346 | cups_utf32_t ch32; /* UTF-32 character value */ | |
347 | ||
ef416fc2 | 348 | |
349 | /* | |
350 | * Check for valid arguments and clear output... | |
351 | */ | |
e1d6a774 | 352 | |
e07d4801 MS |
353 | DEBUG_printf(("2cupsUTF8ToUTF32(dest=%p, src=\"%s\", maxout=%d)", dest, |
354 | src, maxout)); | |
c9fc04c6 | 355 | |
e1d6a774 | 356 | if (dest) |
357 | *dest = 0; | |
358 | ||
359 | if (!dest || !src || maxout < 1 || maxout > CUPS_MAX_USTRING) | |
c9fc04c6 | 360 | { |
e07d4801 | 361 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad arguments)"); |
c9fc04c6 | 362 | |
ef416fc2 | 363 | return (-1); |
c9fc04c6 | 364 | } |
ef416fc2 | 365 | |
366 | /* | |
cda47a96 | 367 | * Convert input UTF-8 to output UTF-32... |
ef416fc2 | 368 | */ |
e1d6a774 | 369 | |
e1d6a774 | 370 | for (i = maxout - 1; *src && i > 0; i --) |
ef416fc2 | 371 | { |
e1d6a774 | 372 | ch = *src++; |
ef416fc2 | 373 | |
374 | /* | |
375 | * Convert UTF-8 character(s) to UTF-32 character... | |
376 | */ | |
e1d6a774 | 377 | |
378 | if (!(ch & 0x80)) | |
ef416fc2 | 379 | { |
380 | /* | |
381 | * One-octet UTF-8 <= 127 (US-ASCII)... | |
382 | */ | |
e1d6a774 | 383 | |
384 | *dest++ = ch; | |
c9fc04c6 | 385 | |
e07d4801 | 386 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x => %08X", src[-1], ch)); |
2abf387c | 387 | continue; |
ef416fc2 | 388 | } |
389 | else if ((ch & 0xe0) == 0xc0) | |
390 | { | |
391 | /* | |
392 | * Two-octet UTF-8 <= 2047 (Latin-x)... | |
393 | */ | |
e1d6a774 | 394 | |
395 | next = *src++; | |
c9fc04c6 MS |
396 | if ((next & 0xc0) != 0x80) |
397 | { | |
e07d4801 | 398 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 399 | |
ef416fc2 | 400 | return (-1); |
c9fc04c6 | 401 | } |
e1d6a774 | 402 | |
ef416fc2 | 403 | ch32 = ((ch & 0x1f) << 6) | (next & 0x3f); |
404 | ||
405 | /* | |
406 | * Check for non-shortest form (invalid UTF-8)... | |
407 | */ | |
e1d6a774 | 408 | |
409 | if (ch32 < 0x80) | |
c9fc04c6 | 410 | { |
e07d4801 | 411 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 412 | |
ef416fc2 | 413 | return (-1); |
c9fc04c6 | 414 | } |
e1d6a774 | 415 | |
416 | *dest++ = ch32; | |
c9fc04c6 | 417 | |
e07d4801 | 418 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x %02x => %08X", |
c9fc04c6 | 419 | src[-2], src[-1], (unsigned)ch32)); |
ef416fc2 | 420 | } |
421 | else if ((ch & 0xf0) == 0xe0) | |
422 | { | |
423 | /* | |
424 | * Three-octet UTF-8 <= 65535 (Plane 0 - BMP)... | |
425 | */ | |
e1d6a774 | 426 | |
427 | next = *src++; | |
c9fc04c6 MS |
428 | if ((next & 0xc0) != 0x80) |
429 | { | |
e07d4801 | 430 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 431 | |
ef416fc2 | 432 | return (-1); |
c9fc04c6 | 433 | } |
e1d6a774 | 434 | |
435 | ch32 = ((ch & 0x0f) << 6) | (next & 0x3f); | |
436 | ||
437 | next = *src++; | |
c9fc04c6 MS |
438 | if ((next & 0xc0) != 0x80) |
439 | { | |
e07d4801 | 440 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 441 | |
ef416fc2 | 442 | return (-1); |
c9fc04c6 | 443 | } |
e1d6a774 | 444 | |
445 | ch32 = (ch32 << 6) | (next & 0x3f); | |
ef416fc2 | 446 | |
447 | /* | |
448 | * Check for non-shortest form (invalid UTF-8)... | |
449 | */ | |
e1d6a774 | 450 | |
451 | if (ch32 < 0x800) | |
c9fc04c6 | 452 | { |
e07d4801 | 453 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 454 | |
ef416fc2 | 455 | return (-1); |
c9fc04c6 | 456 | } |
e1d6a774 | 457 | |
458 | *dest++ = ch32; | |
c9fc04c6 | 459 | |
e07d4801 | 460 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x %02x %02x => %08X", |
c9fc04c6 | 461 | src[-3], src[-2], src[-1], (unsigned)ch32)); |
ef416fc2 | 462 | } |
463 | else if ((ch & 0xf8) == 0xf0) | |
464 | { | |
465 | /* | |
e1d6a774 | 466 | * Four-octet UTF-8... |
ef416fc2 | 467 | */ |
e1d6a774 | 468 | |
469 | next = *src++; | |
c9fc04c6 MS |
470 | if ((next & 0xc0) != 0x80) |
471 | { | |
e07d4801 | 472 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 473 | |
ef416fc2 | 474 | return (-1); |
c9fc04c6 | 475 | } |
e1d6a774 | 476 | |
477 | ch32 = ((ch & 0x07) << 6) | (next & 0x3f); | |
478 | ||
479 | next = *src++; | |
c9fc04c6 MS |
480 | if ((next & 0xc0) != 0x80) |
481 | { | |
e07d4801 | 482 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 483 | |
e1d6a774 | 484 | return (-1); |
c9fc04c6 | 485 | } |
e1d6a774 | 486 | |
487 | ch32 = (ch32 << 6) | (next & 0x3f); | |
488 | ||
489 | next = *src++; | |
c9fc04c6 MS |
490 | if ((next & 0xc0) != 0x80) |
491 | { | |
e07d4801 | 492 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 493 | |
e1d6a774 | 494 | return (-1); |
c9fc04c6 | 495 | } |
e1d6a774 | 496 | |
497 | ch32 = (ch32 << 6) | (next & 0x3f); | |
498 | ||
ef416fc2 | 499 | /* |
e1d6a774 | 500 | * Check for non-shortest form (invalid UTF-8)... |
ef416fc2 | 501 | */ |
e1d6a774 | 502 | |
503 | if (ch32 < 0x10000) | |
c9fc04c6 | 504 | { |
e07d4801 | 505 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 506 | |
e1d6a774 | 507 | return (-1); |
c9fc04c6 | 508 | } |
e1d6a774 | 509 | |
510 | *dest++ = ch32; | |
c9fc04c6 | 511 | |
e07d4801 | 512 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x %02x %02x %02x => %08X", |
c9fc04c6 | 513 | src[-4], src[-3], src[-2], src[-1], (unsigned)ch32)); |
ef416fc2 | 514 | } |
515 | else | |
516 | { | |
517 | /* | |
e1d6a774 | 518 | * More than 4-octet (invalid UTF-8 sequence)... |
ef416fc2 | 519 | */ |
e1d6a774 | 520 | |
e07d4801 | 521 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 522 | |
ef416fc2 | 523 | return (-1); |
524 | } | |
525 | ||
526 | /* | |
527 | * Check for UTF-16 surrogate (illegal UTF-8)... | |
528 | */ | |
ef416fc2 | 529 | |
2abf387c | 530 | if (ch32 >= 0xd800 && ch32 <= 0xdfff) |
ef416fc2 | 531 | return (-1); |
532 | } | |
e1d6a774 | 533 | |
ef416fc2 | 534 | *dest = 0; |
e1d6a774 | 535 | |
e07d4801 | 536 | DEBUG_printf(("3cupsUTF8ToUTF32: Returning %d characters", maxout - 1 - i)); |
c9fc04c6 MS |
537 | |
538 | return (maxout - 1 - i); | |
ef416fc2 | 539 | } |
540 | ||
e1d6a774 | 541 | |
ef416fc2 | 542 | /* |
543 | * 'cupsUTF32ToUTF8()' - Convert UTF-32 to UTF-8. | |
544 | * | |
545 | * 32-bit UTF-32 (actually 21-bit) maps to UTF-8 as follows... | |
546 | * | |
547 | * UTF-32 char UTF-8 char(s) | |
548 | * -------------------------------------------------- | |
e1d6a774 | 549 | * 0 to 127 = 0xxxxxxx (US-ASCII) |
ef416fc2 | 550 | * 128 to 2047 = 110xxxxx 10yyyyyy |
551 | * 2048 to 65535 = 1110xxxx 10yyyyyy 10zzzzzz | |
e1d6a774 | 552 | * > 65535 = 11110xxx 10yyyyyy 10zzzzzz 10xxxxxx |
ef416fc2 | 553 | * |
554 | * UTF-32 prohibits chars beyond Plane 16 (> 0x10ffff) in UCS-4, | |
555 | * which would convert to five- or six-octet UTF-8 sequences... | |
ef416fc2 | 556 | */ |
e1d6a774 | 557 | |
558 | int /* O - Count or -1 on error */ | |
559 | cupsUTF32ToUTF8( | |
560 | cups_utf8_t *dest, /* O - Target string */ | |
561 | const cups_utf32_t *src, /* I - Source string */ | |
562 | const int maxout) /* I - Max output */ | |
ef416fc2 | 563 | { |
e1d6a774 | 564 | cups_utf8_t *start; /* Start of destination string */ |
565 | int i; /* Looping variable */ | |
566 | int swap; /* Byte-swap input to output */ | |
567 | cups_utf32_t ch; /* Character value */ | |
568 | ||
ef416fc2 | 569 | |
570 | /* | |
571 | * Check for valid arguments and clear output... | |
572 | */ | |
e1d6a774 | 573 | |
e07d4801 | 574 | DEBUG_printf(("2cupsUTF32ToUTF8(dest=%p, src=%p, maxout=%d)", dest, src, |
c9fc04c6 MS |
575 | maxout)); |
576 | ||
e1d6a774 | 577 | if (dest) |
578 | *dest = '\0'; | |
579 | ||
580 | if (!dest || !src || maxout < 1) | |
c9fc04c6 | 581 | { |
e07d4801 | 582 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (bad args)"); |
c9fc04c6 | 583 | |
ef416fc2 | 584 | return (-1); |
c9fc04c6 | 585 | } |
ef416fc2 | 586 | |
587 | /* | |
588 | * Check for leading BOM in UTF-32 and inverted BOM... | |
589 | */ | |
e1d6a774 | 590 | |
591 | start = dest; | |
592 | swap = *src == 0xfffe0000; | |
593 | ||
e07d4801 | 594 | DEBUG_printf(("4cupsUTF32ToUTF8: swap=%d", swap)); |
c9fc04c6 | 595 | |
e1d6a774 | 596 | if (*src == 0xfffe0000 || *src == 0xfeff) |
597 | src ++; | |
ef416fc2 | 598 | |
599 | /* | |
600 | * Convert input UTF-32 to output UTF-8... | |
601 | */ | |
e1d6a774 | 602 | |
603 | for (i = maxout - 1; *src && i > 0;) | |
ef416fc2 | 604 | { |
e1d6a774 | 605 | ch = *src++; |
ef416fc2 | 606 | |
607 | /* | |
608 | * Byte swap input UTF-32, if necessary... | |
e1d6a774 | 609 | * (only byte-swapping 24 of 32 bits) |
ef416fc2 | 610 | */ |
e1d6a774 | 611 | |
ef416fc2 | 612 | if (swap) |
613 | ch = ((ch >> 24) | ((ch >> 8) & 0xff00) | ((ch << 8) & 0xff0000)); | |
614 | ||
615 | /* | |
e1d6a774 | 616 | * Check for beyond Plane 16 (invalid UTF-32)... |
ef416fc2 | 617 | */ |
ef416fc2 | 618 | |
ef416fc2 | 619 | if (ch > 0x10ffff) |
c9fc04c6 | 620 | { |
e07d4801 | 621 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (character out of range)"); |
c9fc04c6 | 622 | |
ef416fc2 | 623 | return (-1); |
c9fc04c6 | 624 | } |
ef416fc2 | 625 | |
ef416fc2 | 626 | /* |
627 | * Convert UTF-32 character to UTF-8 character(s)... | |
628 | */ | |
e1d6a774 | 629 | |
630 | if (ch < 0x80) | |
ef416fc2 | 631 | { |
632 | /* | |
633 | * One-octet UTF-8 <= 127 (US-ASCII)... | |
634 | */ | |
e1d6a774 | 635 | |
636 | *dest++ = (cups_utf8_t)ch; | |
637 | i --; | |
c9fc04c6 | 638 | |
e07d4801 | 639 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x", (unsigned)ch, dest[-1])); |
ef416fc2 | 640 | } |
e1d6a774 | 641 | else if (ch < 0x800) |
ef416fc2 | 642 | { |
643 | /* | |
644 | * Two-octet UTF-8 <= 2047 (Latin-x)... | |
645 | */ | |
e1d6a774 | 646 | |
647 | if (i < 2) | |
c9fc04c6 | 648 | { |
e07d4801 | 649 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 2)"); |
c9fc04c6 | 650 | |
e1d6a774 | 651 | return (-1); |
c9fc04c6 | 652 | } |
e1d6a774 | 653 | |
654 | *dest++ = (cups_utf8_t)(0xc0 | ((ch >> 6) & 0x1f)); | |
655 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
656 | i -= 2; | |
c9fc04c6 | 657 | |
e07d4801 | 658 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x", (unsigned)ch, |
c9fc04c6 | 659 | dest[-2], dest[-1])); |
ef416fc2 | 660 | } |
e1d6a774 | 661 | else if (ch < 0x10000) |
ef416fc2 | 662 | { |
663 | /* | |
664 | * Three-octet UTF-8 <= 65535 (Plane 0 - BMP)... | |
665 | */ | |
e1d6a774 | 666 | |
667 | if (i < 3) | |
c9fc04c6 | 668 | { |
e07d4801 | 669 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 3)"); |
c9fc04c6 | 670 | |
e1d6a774 | 671 | return (-1); |
c9fc04c6 | 672 | } |
e1d6a774 | 673 | |
674 | *dest++ = (cups_utf8_t)(0xe0 | ((ch >> 12) & 0x0f)); | |
675 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 6) & 0x3f)); | |
676 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
677 | i -= 3; | |
c9fc04c6 | 678 | |
e07d4801 | 679 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x %02x", (unsigned)ch, |
c9fc04c6 | 680 | dest[-3], dest[-2], dest[-1])); |
e1d6a774 | 681 | } |
682 | else | |
683 | { | |
684 | /* | |
685 | * Four-octet UTF-8... | |
686 | */ | |
687 | ||
688 | if (i < 4) | |
e07d4801 MS |
689 | { |
690 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 4)"); | |
691 | ||
e1d6a774 | 692 | return (-1); |
e07d4801 | 693 | } |
e1d6a774 | 694 | |
695 | *dest++ = (cups_utf8_t)(0xf0 | ((ch >> 18) & 0x07)); | |
696 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 12) & 0x3f)); | |
697 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 6) & 0x3f)); | |
698 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
699 | i -= 4; | |
c9fc04c6 | 700 | |
e07d4801 | 701 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x %02x %02x", |
c9fc04c6 | 702 | (unsigned)ch, dest[-4], dest[-3], dest[-2], dest[-1])); |
ef416fc2 | 703 | } |
704 | } | |
e1d6a774 | 705 | |
ef416fc2 | 706 | *dest = '\0'; |
e1d6a774 | 707 | |
e07d4801 | 708 | DEBUG_printf(("3cupsUTF32ToUTF8: Returning %d", (int)(dest - start))); |
c9fc04c6 | 709 | |
e1d6a774 | 710 | return ((int)(dest - start)); |
ef416fc2 | 711 | } |
712 | ||
e1d6a774 | 713 | |
ef416fc2 | 714 | /* |
31db8ded | 715 | * End of "$Id: transcode.c 9306 2010-09-16 21:43:57Z mike $" |
ef416fc2 | 716 | */ |