]>
Commit | Line | Data |
---|---|---|
ef416fc2 | 1 | /* |
7e86f2f6 | 2 | * Transcoding support for CUPS. |
ef416fc2 | 3 | * |
7e86f2f6 MS |
4 | * Copyright 2007-2014 by Apple Inc. |
5 | * Copyright 1997-2007 by Easy Software Products. | |
ef416fc2 | 6 | * |
7e86f2f6 MS |
7 | * These coded instructions, statements, and computer programs are the |
8 | * property of Apple Inc. and are protected by Federal copyright | |
9 | * law. Distribution and use rights are outlined in the file "LICENSE.txt" | |
10 | * which should have been included with this file. If this file is | |
11 | * file is missing or damaged, see the license at "http://www.cups.org/". | |
ef416fc2 | 12 | * |
7e86f2f6 | 13 | * This file is subject to the Apple OS-Developed Software exception. |
ef416fc2 | 14 | */ |
15 | ||
16 | /* | |
17 | * Include necessary headers... | |
18 | */ | |
19 | ||
71e16022 | 20 | #include "cups-private.h" |
e53920b9 | 21 | #include <limits.h> |
ef416fc2 | 22 | #include <time.h> |
cc754834 MS |
23 | #ifdef HAVE_ICONV_H |
24 | # include <iconv.h> | |
25 | #endif /* HAVE_ICONV_H */ | |
ef416fc2 | 26 | |
27 | ||
d6ae789d | 28 | /* |
29 | * Local globals... | |
30 | */ | |
31 | ||
cc754834 | 32 | #ifdef HAVE_ICONV_H |
6d2f911b | 33 | static _cups_mutex_t map_mutex = _CUPS_MUTEX_INITIALIZER; |
d6ae789d | 34 | /* Mutex to control access to maps */ |
cc754834 MS |
35 | static iconv_t map_from_utf8 = (iconv_t)-1; |
36 | /* Convert from UTF-8 to charset */ | |
37 | static iconv_t map_to_utf8 = (iconv_t)-1; | |
38 | /* Convert from charset to UTF-8 */ | |
39 | static cups_encoding_t map_encoding = CUPS_AUTO_ENCODING; | |
40 | /* Which charset is cached */ | |
41 | #endif /* HAVE_ICONV_H */ | |
e1d6a774 | 42 | |
ef416fc2 | 43 | |
44 | /* | |
e1d6a774 | 45 | * '_cupsCharmapFlush()' - Flush all character set maps out of cache. |
ef416fc2 | 46 | */ |
47 | ||
e1d6a774 | 48 | void |
d6ae789d | 49 | _cupsCharmapFlush(void) |
ef416fc2 | 50 | { |
cc754834 MS |
51 | #ifdef HAVE_ICONV_H |
52 | if (map_from_utf8 != (iconv_t)-1) | |
ef416fc2 | 53 | { |
cc754834 MS |
54 | iconv_close(map_from_utf8); |
55 | map_from_utf8 = (iconv_t)-1; | |
ef416fc2 | 56 | } |
d6ae789d | 57 | |
cc754834 | 58 | if (map_to_utf8 != (iconv_t)-1) |
ef416fc2 | 59 | { |
cc754834 MS |
60 | iconv_close(map_to_utf8); |
61 | map_to_utf8 = (iconv_t)-1; | |
ef416fc2 | 62 | } |
ef416fc2 | 63 | |
cc754834 MS |
64 | map_encoding = CUPS_AUTO_ENCODING; |
65 | #endif /* HAVE_ICONV_H */ | |
ef416fc2 | 66 | } |
67 | ||
e1d6a774 | 68 | |
ef416fc2 | 69 | /* |
e1d6a774 | 70 | * 'cupsCharsetToUTF8()' - Convert legacy character set to UTF-8. |
ef416fc2 | 71 | */ |
e1d6a774 | 72 | |
73 | int /* O - Count or -1 on error */ | |
74 | cupsCharsetToUTF8( | |
cc754834 MS |
75 | cups_utf8_t *dest, /* O - Target string */ |
76 | const char *src, /* I - Source string */ | |
77 | const int maxout, /* I - Max output */ | |
e1d6a774 | 78 | const cups_encoding_t encoding) /* I - Encoding */ |
ef416fc2 | 79 | { |
cc754834 | 80 | cups_utf8_t *destptr; /* Pointer into UTF-8 buffer */ |
84315f46 | 81 | #ifdef HAVE_ICONV_H |
cc754834 MS |
82 | size_t srclen, /* Length of source string */ |
83 | outBytesLeft; /* Bytes remaining in output buffer */ | |
7cf5915e | 84 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 85 | |
86 | ||
ef416fc2 | 87 | /* |
88 | * Check for valid arguments... | |
89 | */ | |
90 | ||
807315e6 | 91 | DEBUG_printf(("2cupsCharsetToUTF8(dest=%p, src=\"%s\", maxout=%d, encoding=%d)", (void *)dest, src, maxout, encoding)); |
e1d6a774 | 92 | |
cc754834 | 93 | if (!dest || !src || maxout < 1) |
e1d6a774 | 94 | { |
cc754834 MS |
95 | if (dest) |
96 | *dest = '\0'; | |
97 | ||
f11a948a | 98 | DEBUG_puts("3cupsCharsetToUTF8: Bad arguments, returning -1"); |
ef416fc2 | 99 | return (-1); |
e1d6a774 | 100 | } |
ef416fc2 | 101 | |
102 | /* | |
103 | * Handle identity conversions... | |
104 | */ | |
105 | ||
cc754834 MS |
106 | if (encoding == CUPS_UTF8 || encoding <= CUPS_US_ASCII || |
107 | encoding >= CUPS_ENCODING_VBCS_END) | |
ef416fc2 | 108 | { |
07623986 | 109 | strlcpy((char *)dest, src, (size_t)maxout); |
b86bc4cf | 110 | return ((int)strlen((char *)dest)); |
ef416fc2 | 111 | } |
112 | ||
411affcf | 113 | /* |
114 | * Handle ISO-8859-1 to UTF-8 directly... | |
115 | */ | |
116 | ||
cc754834 MS |
117 | destptr = dest; |
118 | ||
411affcf | 119 | if (encoding == CUPS_ISO8859_1) |
120 | { | |
121 | int ch; /* Character from string */ | |
cc754834 | 122 | cups_utf8_t *destend; /* End of UTF-8 buffer */ |
411affcf | 123 | |
124 | ||
411affcf | 125 | destend = dest + maxout - 2; |
126 | ||
127 | while (*src && destptr < destend) | |
128 | { | |
129 | ch = *src++ & 255; | |
130 | ||
131 | if (ch & 128) | |
132 | { | |
7e86f2f6 MS |
133 | *destptr++ = (cups_utf8_t)(0xc0 | (ch >> 6)); |
134 | *destptr++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
411affcf | 135 | } |
136 | else | |
7e86f2f6 | 137 | *destptr++ = (cups_utf8_t)ch; |
411affcf | 138 | } |
139 | ||
140 | *destptr = '\0'; | |
141 | ||
b86bc4cf | 142 | return ((int)(destptr - dest)); |
411affcf | 143 | } |
144 | ||
ef416fc2 | 145 | /* |
e1d6a774 | 146 | * Convert input legacy charset to UTF-8... |
ef416fc2 | 147 | */ |
e1d6a774 | 148 | |
cc754834 | 149 | #ifdef HAVE_ICONV_H |
6d2f911b | 150 | _cupsMutexLock(&map_mutex); |
d6ae789d | 151 | |
cc754834 MS |
152 | if (map_encoding != encoding) |
153 | { | |
52958fdb MS |
154 | char toset[1024]; /* Destination character set */ |
155 | ||
cc754834 MS |
156 | _cupsCharmapFlush(); |
157 | ||
52958fdb MS |
158 | snprintf(toset, sizeof(toset), "%s//IGNORE", _cupsEncodingName(encoding)); |
159 | ||
160 | map_encoding = encoding; | |
cc754834 | 161 | map_from_utf8 = iconv_open(_cupsEncodingName(encoding), "UTF-8"); |
52958fdb | 162 | map_to_utf8 = iconv_open("UTF-8", toset); |
cc754834 MS |
163 | } |
164 | ||
165 | if (map_to_utf8 != (iconv_t)-1) | |
166 | { | |
f99f3698 MS |
167 | char *altdestptr = (char *)dest; /* Silence bogus GCC type-punned */ |
168 | ||
cc754834 | 169 | srclen = strlen(src); |
7e86f2f6 | 170 | outBytesLeft = (size_t)maxout - 1; |
4220952d | 171 | |
f99f3698 MS |
172 | iconv(map_to_utf8, (char **)&src, &srclen, &altdestptr, &outBytesLeft); |
173 | *altdestptr = '\0'; | |
cc754834 MS |
174 | |
175 | _cupsMutexUnlock(&map_mutex); | |
176 | ||
f99f3698 | 177 | return ((int)(altdestptr - (char *)dest)); |
cc754834 | 178 | } |
d6ae789d | 179 | |
6d2f911b | 180 | _cupsMutexUnlock(&map_mutex); |
cc754834 | 181 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 182 | |
cc754834 MS |
183 | /* |
184 | * No iconv() support, so error out... | |
185 | */ | |
186 | ||
187 | *destptr = '\0'; | |
188 | ||
189 | return (-1); | |
ef416fc2 | 190 | } |
191 | ||
e1d6a774 | 192 | |
ef416fc2 | 193 | /* |
e1d6a774 | 194 | * 'cupsUTF8ToCharset()' - Convert UTF-8 to legacy character set. |
ef416fc2 | 195 | */ |
e1d6a774 | 196 | |
197 | int /* O - Count or -1 on error */ | |
198 | cupsUTF8ToCharset( | |
199 | char *dest, /* O - Target string */ | |
200 | const cups_utf8_t *src, /* I - Source string */ | |
201 | const int maxout, /* I - Max output */ | |
202 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 203 | { |
cc754834 | 204 | char *destptr; /* Pointer into destination */ |
84315f46 | 205 | #ifdef HAVE_ICONV_H |
cc754834 MS |
206 | size_t srclen, /* Length of source string */ |
207 | outBytesLeft; /* Bytes remaining in output buffer */ | |
7cf5915e | 208 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 209 | |
210 | ||
ef416fc2 | 211 | /* |
212 | * Check for valid arguments... | |
213 | */ | |
214 | ||
cc754834 | 215 | if (!dest || !src || maxout < 1) |
e1d6a774 | 216 | { |
217 | if (dest) | |
218 | *dest = '\0'; | |
219 | ||
ef416fc2 | 220 | return (-1); |
e1d6a774 | 221 | } |
ef416fc2 | 222 | |
223 | /* | |
224 | * Handle identity conversions... | |
225 | */ | |
226 | ||
22c9029b | 227 | if (encoding == CUPS_UTF8 || |
cc754834 | 228 | encoding >= CUPS_ENCODING_VBCS_END) |
ef416fc2 | 229 | { |
07623986 | 230 | strlcpy(dest, (char *)src, (size_t)maxout); |
b86bc4cf | 231 | return ((int)strlen(dest)); |
ef416fc2 | 232 | } |
233 | ||
411affcf | 234 | /* |
235 | * Handle UTF-8 to ISO-8859-1 directly... | |
236 | */ | |
237 | ||
cc754834 MS |
238 | destptr = dest; |
239 | ||
22c9029b | 240 | if (encoding == CUPS_ISO8859_1 || encoding <= CUPS_US_ASCII) |
411affcf | 241 | { |
22c9029b MS |
242 | int ch, /* Character from string */ |
243 | maxch; /* Maximum character for charset */ | |
cc754834 | 244 | char *destend; /* End of ISO-8859-1 buffer */ |
411affcf | 245 | |
22c9029b | 246 | maxch = encoding == CUPS_ISO8859_1 ? 256 : 128; |
411affcf | 247 | destend = dest + maxout - 1; |
248 | ||
249 | while (*src && destptr < destend) | |
250 | { | |
251 | ch = *src++; | |
252 | ||
253 | if ((ch & 0xe0) == 0xc0) | |
254 | { | |
255 | ch = ((ch & 0x1f) << 6) | (*src++ & 0x3f); | |
256 | ||
22c9029b | 257 | if (ch < maxch) |
7e86f2f6 | 258 | *destptr++ = (char)ch; |
411affcf | 259 | else |
260 | *destptr++ = '?'; | |
261 | } | |
262 | else if ((ch & 0xf0) == 0xe0 || | |
263 | (ch & 0xf8) == 0xf0) | |
264 | *destptr++ = '?'; | |
265 | else if (!(ch & 0x80)) | |
7e86f2f6 | 266 | *destptr++ = (char)ch; |
411affcf | 267 | } |
268 | ||
269 | *destptr = '\0'; | |
270 | ||
b86bc4cf | 271 | return ((int)(destptr - dest)); |
411affcf | 272 | } |
273 | ||
cc754834 | 274 | #ifdef HAVE_ICONV_H |
ef416fc2 | 275 | /* |
e1d6a774 | 276 | * Convert input UTF-8 to legacy charset... |
ef416fc2 | 277 | */ |
e1d6a774 | 278 | |
6d2f911b | 279 | _cupsMutexLock(&map_mutex); |
d6ae789d | 280 | |
cc754834 MS |
281 | if (map_encoding != encoding) |
282 | { | |
52958fdb MS |
283 | char toset[1024]; /* Destination character set */ |
284 | ||
cc754834 MS |
285 | _cupsCharmapFlush(); |
286 | ||
52958fdb MS |
287 | snprintf(toset, sizeof(toset), "%s//IGNORE", _cupsEncodingName(encoding)); |
288 | ||
cc754834 | 289 | map_encoding = encoding; |
52958fdb MS |
290 | map_from_utf8 = iconv_open(_cupsEncodingName(encoding), "UTF-8"); |
291 | map_to_utf8 = iconv_open("UTF-8", toset); | |
cc754834 MS |
292 | } |
293 | ||
294 | if (map_from_utf8 != (iconv_t)-1) | |
295 | { | |
f99f3698 MS |
296 | char *altsrc = (char *)src; /* Silence bogus GCC type-punned */ |
297 | ||
cc754834 | 298 | srclen = strlen((char *)src); |
7e86f2f6 | 299 | outBytesLeft = (size_t)maxout - 1; |
4220952d | 300 | |
f99f3698 | 301 | iconv(map_from_utf8, &altsrc, &srclen, &destptr, &outBytesLeft); |
4220952d | 302 | *destptr = '\0'; |
cc754834 MS |
303 | |
304 | _cupsMutexUnlock(&map_mutex); | |
305 | ||
306 | return ((int)(destptr - dest)); | |
307 | } | |
d6ae789d | 308 | |
6d2f911b | 309 | _cupsMutexUnlock(&map_mutex); |
cc754834 MS |
310 | #endif /* HAVE_ICONV_H */ |
311 | ||
312 | /* | |
313 | * No iconv() support, so error out... | |
314 | */ | |
315 | ||
316 | *destptr = '\0'; | |
d6ae789d | 317 | |
cc754834 | 318 | return (-1); |
ef416fc2 | 319 | } |
320 | ||
ef416fc2 | 321 | |
322 | /* | |
323 | * 'cupsUTF8ToUTF32()' - Convert UTF-8 to UTF-32. | |
324 | * | |
325 | * 32-bit UTF-32 (actually 21-bit) maps to UTF-8 as follows... | |
326 | * | |
327 | * UTF-32 char UTF-8 char(s) | |
328 | * -------------------------------------------------- | |
e1d6a774 | 329 | * 0 to 127 = 0xxxxxxx (US-ASCII) |
ef416fc2 | 330 | * 128 to 2047 = 110xxxxx 10yyyyyy |
331 | * 2048 to 65535 = 1110xxxx 10yyyyyy 10zzzzzz | |
e1d6a774 | 332 | * > 65535 = 11110xxx 10yyyyyy 10zzzzzz 10xxxxxx |
ef416fc2 | 333 | * |
334 | * UTF-32 prohibits chars beyond Plane 16 (> 0x10ffff) in UCS-4, | |
335 | * which would convert to five- or six-octet UTF-8 sequences... | |
ef416fc2 | 336 | */ |
e1d6a774 | 337 | |
338 | int /* O - Count or -1 on error */ | |
339 | cupsUTF8ToUTF32( | |
340 | cups_utf32_t *dest, /* O - Target string */ | |
341 | const cups_utf8_t *src, /* I - Source string */ | |
342 | const int maxout) /* I - Max output */ | |
ef416fc2 | 343 | { |
e1d6a774 | 344 | int i; /* Looping variable */ |
345 | cups_utf8_t ch; /* Character value */ | |
346 | cups_utf8_t next; /* Next character value */ | |
347 | cups_utf32_t ch32; /* UTF-32 character value */ | |
348 | ||
ef416fc2 | 349 | |
350 | /* | |
351 | * Check for valid arguments and clear output... | |
352 | */ | |
e1d6a774 | 353 | |
807315e6 | 354 | DEBUG_printf(("2cupsUTF8ToUTF32(dest=%p, src=\"%s\", maxout=%d)", (void *)dest, 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 | |
7e86f2f6 | 403 | ch32 = (cups_utf32_t)((ch & 0x1f) << 6) | (cups_utf32_t)(next & 0x3f); |
ef416fc2 | 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 | |
7e86f2f6 | 435 | ch32 = (cups_utf32_t)((ch & 0x0f) << 6) | (cups_utf32_t)(next & 0x3f); |
e1d6a774 | 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 | |
7e86f2f6 | 445 | ch32 = (ch32 << 6) | (cups_utf32_t)(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 | |
7e86f2f6 | 477 | ch32 = (cups_utf32_t)((ch & 0x07) << 6) | (cups_utf32_t)(next & 0x3f); |
e1d6a774 | 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 | |
7e86f2f6 | 487 | ch32 = (ch32 << 6) | (cups_utf32_t)(next & 0x3f); |
e1d6a774 | 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 | |
7e86f2f6 | 497 | ch32 = (ch32 << 6) | (cups_utf32_t)(next & 0x3f); |
e1d6a774 | 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 | |
807315e6 | 574 | DEBUG_printf(("2cupsUTF32ToUTF8(dest=%p, src=%p, maxout=%d)", (void *)dest, (void *)src, maxout)); |
c9fc04c6 | 575 | |
e1d6a774 | 576 | if (dest) |
577 | *dest = '\0'; | |
578 | ||
579 | if (!dest || !src || maxout < 1) | |
c9fc04c6 | 580 | { |
e07d4801 | 581 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (bad args)"); |
c9fc04c6 | 582 | |
ef416fc2 | 583 | return (-1); |
c9fc04c6 | 584 | } |
ef416fc2 | 585 | |
586 | /* | |
587 | * Check for leading BOM in UTF-32 and inverted BOM... | |
588 | */ | |
e1d6a774 | 589 | |
590 | start = dest; | |
591 | swap = *src == 0xfffe0000; | |
592 | ||
e07d4801 | 593 | DEBUG_printf(("4cupsUTF32ToUTF8: swap=%d", swap)); |
c9fc04c6 | 594 | |
e1d6a774 | 595 | if (*src == 0xfffe0000 || *src == 0xfeff) |
596 | src ++; | |
ef416fc2 | 597 | |
598 | /* | |
599 | * Convert input UTF-32 to output UTF-8... | |
600 | */ | |
e1d6a774 | 601 | |
602 | for (i = maxout - 1; *src && i > 0;) | |
ef416fc2 | 603 | { |
e1d6a774 | 604 | ch = *src++; |
ef416fc2 | 605 | |
606 | /* | |
607 | * Byte swap input UTF-32, if necessary... | |
e1d6a774 | 608 | * (only byte-swapping 24 of 32 bits) |
ef416fc2 | 609 | */ |
e1d6a774 | 610 | |
ef416fc2 | 611 | if (swap) |
612 | ch = ((ch >> 24) | ((ch >> 8) & 0xff00) | ((ch << 8) & 0xff0000)); | |
613 | ||
614 | /* | |
e1d6a774 | 615 | * Check for beyond Plane 16 (invalid UTF-32)... |
ef416fc2 | 616 | */ |
ef416fc2 | 617 | |
ef416fc2 | 618 | if (ch > 0x10ffff) |
c9fc04c6 | 619 | { |
e07d4801 | 620 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (character out of range)"); |
c9fc04c6 | 621 | |
ef416fc2 | 622 | return (-1); |
c9fc04c6 | 623 | } |
ef416fc2 | 624 | |
ef416fc2 | 625 | /* |
626 | * Convert UTF-32 character to UTF-8 character(s)... | |
627 | */ | |
e1d6a774 | 628 | |
629 | if (ch < 0x80) | |
ef416fc2 | 630 | { |
631 | /* | |
632 | * One-octet UTF-8 <= 127 (US-ASCII)... | |
633 | */ | |
e1d6a774 | 634 | |
635 | *dest++ = (cups_utf8_t)ch; | |
636 | i --; | |
c9fc04c6 | 637 | |
e07d4801 | 638 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x", (unsigned)ch, dest[-1])); |
ef416fc2 | 639 | } |
e1d6a774 | 640 | else if (ch < 0x800) |
ef416fc2 | 641 | { |
642 | /* | |
643 | * Two-octet UTF-8 <= 2047 (Latin-x)... | |
644 | */ | |
e1d6a774 | 645 | |
646 | if (i < 2) | |
c9fc04c6 | 647 | { |
e07d4801 | 648 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 2)"); |
c9fc04c6 | 649 | |
e1d6a774 | 650 | return (-1); |
c9fc04c6 | 651 | } |
e1d6a774 | 652 | |
653 | *dest++ = (cups_utf8_t)(0xc0 | ((ch >> 6) & 0x1f)); | |
654 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
655 | i -= 2; | |
c9fc04c6 | 656 | |
e07d4801 | 657 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x", (unsigned)ch, |
c9fc04c6 | 658 | dest[-2], dest[-1])); |
ef416fc2 | 659 | } |
e1d6a774 | 660 | else if (ch < 0x10000) |
ef416fc2 | 661 | { |
662 | /* | |
663 | * Three-octet UTF-8 <= 65535 (Plane 0 - BMP)... | |
664 | */ | |
e1d6a774 | 665 | |
666 | if (i < 3) | |
c9fc04c6 | 667 | { |
e07d4801 | 668 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 3)"); |
c9fc04c6 | 669 | |
e1d6a774 | 670 | return (-1); |
c9fc04c6 | 671 | } |
e1d6a774 | 672 | |
673 | *dest++ = (cups_utf8_t)(0xe0 | ((ch >> 12) & 0x0f)); | |
674 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 6) & 0x3f)); | |
675 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
676 | i -= 3; | |
c9fc04c6 | 677 | |
e07d4801 | 678 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x %02x", (unsigned)ch, |
c9fc04c6 | 679 | dest[-3], dest[-2], dest[-1])); |
e1d6a774 | 680 | } |
681 | else | |
682 | { | |
683 | /* | |
684 | * Four-octet UTF-8... | |
685 | */ | |
686 | ||
687 | if (i < 4) | |
e07d4801 MS |
688 | { |
689 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 4)"); | |
690 | ||
e1d6a774 | 691 | return (-1); |
e07d4801 | 692 | } |
e1d6a774 | 693 | |
694 | *dest++ = (cups_utf8_t)(0xf0 | ((ch >> 18) & 0x07)); | |
695 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 12) & 0x3f)); | |
696 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 6) & 0x3f)); | |
697 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
698 | i -= 4; | |
c9fc04c6 | 699 | |
e07d4801 | 700 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x %02x %02x", |
c9fc04c6 | 701 | (unsigned)ch, dest[-4], dest[-3], dest[-2], dest[-1])); |
ef416fc2 | 702 | } |
703 | } | |
e1d6a774 | 704 | |
ef416fc2 | 705 | *dest = '\0'; |
e1d6a774 | 706 | |
e07d4801 | 707 | DEBUG_printf(("3cupsUTF32ToUTF8: Returning %d", (int)(dest - start))); |
c9fc04c6 | 708 | |
e1d6a774 | 709 | return ((int)(dest - start)); |
ef416fc2 | 710 | } |