]>
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 | * |
e3101897 | 7 | * Licensed under Apache License v2.0. See the file "LICENSE" for more information. |
ef416fc2 | 8 | */ |
9 | ||
10 | /* | |
11 | * Include necessary headers... | |
12 | */ | |
13 | ||
71e16022 | 14 | #include "cups-private.h" |
e53920b9 | 15 | #include <limits.h> |
ef416fc2 | 16 | #include <time.h> |
cc754834 MS |
17 | #ifdef HAVE_ICONV_H |
18 | # include <iconv.h> | |
19 | #endif /* HAVE_ICONV_H */ | |
ef416fc2 | 20 | |
21 | ||
d6ae789d | 22 | /* |
23 | * Local globals... | |
24 | */ | |
25 | ||
cc754834 | 26 | #ifdef HAVE_ICONV_H |
6d2f911b | 27 | static _cups_mutex_t map_mutex = _CUPS_MUTEX_INITIALIZER; |
d6ae789d | 28 | /* Mutex to control access to maps */ |
cc754834 MS |
29 | static iconv_t map_from_utf8 = (iconv_t)-1; |
30 | /* Convert from UTF-8 to charset */ | |
31 | static iconv_t map_to_utf8 = (iconv_t)-1; | |
32 | /* Convert from charset to UTF-8 */ | |
33 | static cups_encoding_t map_encoding = CUPS_AUTO_ENCODING; | |
34 | /* Which charset is cached */ | |
35 | #endif /* HAVE_ICONV_H */ | |
e1d6a774 | 36 | |
ef416fc2 | 37 | |
38 | /* | |
e1d6a774 | 39 | * '_cupsCharmapFlush()' - Flush all character set maps out of cache. |
ef416fc2 | 40 | */ |
41 | ||
e1d6a774 | 42 | void |
d6ae789d | 43 | _cupsCharmapFlush(void) |
ef416fc2 | 44 | { |
cc754834 MS |
45 | #ifdef HAVE_ICONV_H |
46 | if (map_from_utf8 != (iconv_t)-1) | |
ef416fc2 | 47 | { |
cc754834 MS |
48 | iconv_close(map_from_utf8); |
49 | map_from_utf8 = (iconv_t)-1; | |
ef416fc2 | 50 | } |
d6ae789d | 51 | |
cc754834 | 52 | if (map_to_utf8 != (iconv_t)-1) |
ef416fc2 | 53 | { |
cc754834 MS |
54 | iconv_close(map_to_utf8); |
55 | map_to_utf8 = (iconv_t)-1; | |
ef416fc2 | 56 | } |
ef416fc2 | 57 | |
cc754834 MS |
58 | map_encoding = CUPS_AUTO_ENCODING; |
59 | #endif /* HAVE_ICONV_H */ | |
ef416fc2 | 60 | } |
61 | ||
e1d6a774 | 62 | |
ef416fc2 | 63 | /* |
e1d6a774 | 64 | * 'cupsCharsetToUTF8()' - Convert legacy character set to UTF-8. |
ef416fc2 | 65 | */ |
e1d6a774 | 66 | |
67 | int /* O - Count or -1 on error */ | |
68 | cupsCharsetToUTF8( | |
cc754834 MS |
69 | cups_utf8_t *dest, /* O - Target string */ |
70 | const char *src, /* I - Source string */ | |
71 | const int maxout, /* I - Max output */ | |
e1d6a774 | 72 | const cups_encoding_t encoding) /* I - Encoding */ |
ef416fc2 | 73 | { |
cc754834 | 74 | cups_utf8_t *destptr; /* Pointer into UTF-8 buffer */ |
84315f46 | 75 | #ifdef HAVE_ICONV_H |
cc754834 MS |
76 | size_t srclen, /* Length of source string */ |
77 | outBytesLeft; /* Bytes remaining in output buffer */ | |
7cf5915e | 78 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 79 | |
80 | ||
ef416fc2 | 81 | /* |
82 | * Check for valid arguments... | |
83 | */ | |
84 | ||
807315e6 | 85 | DEBUG_printf(("2cupsCharsetToUTF8(dest=%p, src=\"%s\", maxout=%d, encoding=%d)", (void *)dest, src, maxout, encoding)); |
e1d6a774 | 86 | |
cc754834 | 87 | if (!dest || !src || maxout < 1) |
e1d6a774 | 88 | { |
cc754834 MS |
89 | if (dest) |
90 | *dest = '\0'; | |
91 | ||
f11a948a | 92 | DEBUG_puts("3cupsCharsetToUTF8: Bad arguments, returning -1"); |
ef416fc2 | 93 | return (-1); |
e1d6a774 | 94 | } |
ef416fc2 | 95 | |
96 | /* | |
97 | * Handle identity conversions... | |
98 | */ | |
99 | ||
cc754834 MS |
100 | if (encoding == CUPS_UTF8 || encoding <= CUPS_US_ASCII || |
101 | encoding >= CUPS_ENCODING_VBCS_END) | |
ef416fc2 | 102 | { |
07623986 | 103 | strlcpy((char *)dest, src, (size_t)maxout); |
b86bc4cf | 104 | return ((int)strlen((char *)dest)); |
ef416fc2 | 105 | } |
106 | ||
411affcf | 107 | /* |
108 | * Handle ISO-8859-1 to UTF-8 directly... | |
109 | */ | |
110 | ||
cc754834 MS |
111 | destptr = dest; |
112 | ||
411affcf | 113 | if (encoding == CUPS_ISO8859_1) |
114 | { | |
115 | int ch; /* Character from string */ | |
cc754834 | 116 | cups_utf8_t *destend; /* End of UTF-8 buffer */ |
411affcf | 117 | |
118 | ||
411affcf | 119 | destend = dest + maxout - 2; |
120 | ||
121 | while (*src && destptr < destend) | |
122 | { | |
123 | ch = *src++ & 255; | |
124 | ||
125 | if (ch & 128) | |
126 | { | |
7e86f2f6 MS |
127 | *destptr++ = (cups_utf8_t)(0xc0 | (ch >> 6)); |
128 | *destptr++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
411affcf | 129 | } |
130 | else | |
7e86f2f6 | 131 | *destptr++ = (cups_utf8_t)ch; |
411affcf | 132 | } |
133 | ||
134 | *destptr = '\0'; | |
135 | ||
b86bc4cf | 136 | return ((int)(destptr - dest)); |
411affcf | 137 | } |
138 | ||
ef416fc2 | 139 | /* |
e1d6a774 | 140 | * Convert input legacy charset to UTF-8... |
ef416fc2 | 141 | */ |
e1d6a774 | 142 | |
cc754834 | 143 | #ifdef HAVE_ICONV_H |
6d2f911b | 144 | _cupsMutexLock(&map_mutex); |
d6ae789d | 145 | |
cc754834 MS |
146 | if (map_encoding != encoding) |
147 | { | |
52958fdb MS |
148 | char toset[1024]; /* Destination character set */ |
149 | ||
cc754834 MS |
150 | _cupsCharmapFlush(); |
151 | ||
52958fdb MS |
152 | snprintf(toset, sizeof(toset), "%s//IGNORE", _cupsEncodingName(encoding)); |
153 | ||
154 | map_encoding = encoding; | |
cc754834 | 155 | map_from_utf8 = iconv_open(_cupsEncodingName(encoding), "UTF-8"); |
52958fdb | 156 | map_to_utf8 = iconv_open("UTF-8", toset); |
cc754834 MS |
157 | } |
158 | ||
159 | if (map_to_utf8 != (iconv_t)-1) | |
160 | { | |
f99f3698 MS |
161 | char *altdestptr = (char *)dest; /* Silence bogus GCC type-punned */ |
162 | ||
cc754834 | 163 | srclen = strlen(src); |
7e86f2f6 | 164 | outBytesLeft = (size_t)maxout - 1; |
4220952d | 165 | |
f99f3698 MS |
166 | iconv(map_to_utf8, (char **)&src, &srclen, &altdestptr, &outBytesLeft); |
167 | *altdestptr = '\0'; | |
cc754834 MS |
168 | |
169 | _cupsMutexUnlock(&map_mutex); | |
170 | ||
f99f3698 | 171 | return ((int)(altdestptr - (char *)dest)); |
cc754834 | 172 | } |
d6ae789d | 173 | |
6d2f911b | 174 | _cupsMutexUnlock(&map_mutex); |
cc754834 | 175 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 176 | |
cc754834 MS |
177 | /* |
178 | * No iconv() support, so error out... | |
179 | */ | |
180 | ||
181 | *destptr = '\0'; | |
182 | ||
183 | return (-1); | |
ef416fc2 | 184 | } |
185 | ||
e1d6a774 | 186 | |
ef416fc2 | 187 | /* |
e1d6a774 | 188 | * 'cupsUTF8ToCharset()' - Convert UTF-8 to legacy character set. |
ef416fc2 | 189 | */ |
e1d6a774 | 190 | |
191 | int /* O - Count or -1 on error */ | |
192 | cupsUTF8ToCharset( | |
193 | char *dest, /* O - Target string */ | |
194 | const cups_utf8_t *src, /* I - Source string */ | |
195 | const int maxout, /* I - Max output */ | |
196 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 197 | { |
cc754834 | 198 | char *destptr; /* Pointer into destination */ |
84315f46 | 199 | #ifdef HAVE_ICONV_H |
cc754834 MS |
200 | size_t srclen, /* Length of source string */ |
201 | outBytesLeft; /* Bytes remaining in output buffer */ | |
7cf5915e | 202 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 203 | |
204 | ||
ef416fc2 | 205 | /* |
206 | * Check for valid arguments... | |
207 | */ | |
208 | ||
cc754834 | 209 | if (!dest || !src || maxout < 1) |
e1d6a774 | 210 | { |
211 | if (dest) | |
212 | *dest = '\0'; | |
213 | ||
ef416fc2 | 214 | return (-1); |
e1d6a774 | 215 | } |
ef416fc2 | 216 | |
217 | /* | |
218 | * Handle identity conversions... | |
219 | */ | |
220 | ||
22c9029b | 221 | if (encoding == CUPS_UTF8 || |
cc754834 | 222 | encoding >= CUPS_ENCODING_VBCS_END) |
ef416fc2 | 223 | { |
07623986 | 224 | strlcpy(dest, (char *)src, (size_t)maxout); |
b86bc4cf | 225 | return ((int)strlen(dest)); |
ef416fc2 | 226 | } |
227 | ||
411affcf | 228 | /* |
229 | * Handle UTF-8 to ISO-8859-1 directly... | |
230 | */ | |
231 | ||
cc754834 MS |
232 | destptr = dest; |
233 | ||
22c9029b | 234 | if (encoding == CUPS_ISO8859_1 || encoding <= CUPS_US_ASCII) |
411affcf | 235 | { |
22c9029b MS |
236 | int ch, /* Character from string */ |
237 | maxch; /* Maximum character for charset */ | |
cc754834 | 238 | char *destend; /* End of ISO-8859-1 buffer */ |
411affcf | 239 | |
22c9029b | 240 | maxch = encoding == CUPS_ISO8859_1 ? 256 : 128; |
411affcf | 241 | destend = dest + maxout - 1; |
242 | ||
243 | while (*src && destptr < destend) | |
244 | { | |
245 | ch = *src++; | |
246 | ||
247 | if ((ch & 0xe0) == 0xc0) | |
248 | { | |
249 | ch = ((ch & 0x1f) << 6) | (*src++ & 0x3f); | |
250 | ||
22c9029b | 251 | if (ch < maxch) |
7e86f2f6 | 252 | *destptr++ = (char)ch; |
411affcf | 253 | else |
254 | *destptr++ = '?'; | |
255 | } | |
256 | else if ((ch & 0xf0) == 0xe0 || | |
257 | (ch & 0xf8) == 0xf0) | |
258 | *destptr++ = '?'; | |
259 | else if (!(ch & 0x80)) | |
7e86f2f6 | 260 | *destptr++ = (char)ch; |
411affcf | 261 | } |
262 | ||
263 | *destptr = '\0'; | |
264 | ||
b86bc4cf | 265 | return ((int)(destptr - dest)); |
411affcf | 266 | } |
267 | ||
cc754834 | 268 | #ifdef HAVE_ICONV_H |
ef416fc2 | 269 | /* |
e1d6a774 | 270 | * Convert input UTF-8 to legacy charset... |
ef416fc2 | 271 | */ |
e1d6a774 | 272 | |
6d2f911b | 273 | _cupsMutexLock(&map_mutex); |
d6ae789d | 274 | |
cc754834 MS |
275 | if (map_encoding != encoding) |
276 | { | |
52958fdb MS |
277 | char toset[1024]; /* Destination character set */ |
278 | ||
cc754834 MS |
279 | _cupsCharmapFlush(); |
280 | ||
52958fdb MS |
281 | snprintf(toset, sizeof(toset), "%s//IGNORE", _cupsEncodingName(encoding)); |
282 | ||
cc754834 | 283 | map_encoding = encoding; |
52958fdb MS |
284 | map_from_utf8 = iconv_open(_cupsEncodingName(encoding), "UTF-8"); |
285 | map_to_utf8 = iconv_open("UTF-8", toset); | |
cc754834 MS |
286 | } |
287 | ||
288 | if (map_from_utf8 != (iconv_t)-1) | |
289 | { | |
f99f3698 MS |
290 | char *altsrc = (char *)src; /* Silence bogus GCC type-punned */ |
291 | ||
cc754834 | 292 | srclen = strlen((char *)src); |
7e86f2f6 | 293 | outBytesLeft = (size_t)maxout - 1; |
4220952d | 294 | |
f99f3698 | 295 | iconv(map_from_utf8, &altsrc, &srclen, &destptr, &outBytesLeft); |
4220952d | 296 | *destptr = '\0'; |
cc754834 MS |
297 | |
298 | _cupsMutexUnlock(&map_mutex); | |
299 | ||
300 | return ((int)(destptr - dest)); | |
301 | } | |
d6ae789d | 302 | |
6d2f911b | 303 | _cupsMutexUnlock(&map_mutex); |
cc754834 MS |
304 | #endif /* HAVE_ICONV_H */ |
305 | ||
306 | /* | |
307 | * No iconv() support, so error out... | |
308 | */ | |
309 | ||
310 | *destptr = '\0'; | |
d6ae789d | 311 | |
cc754834 | 312 | return (-1); |
ef416fc2 | 313 | } |
314 | ||
ef416fc2 | 315 | |
316 | /* | |
317 | * 'cupsUTF8ToUTF32()' - Convert UTF-8 to UTF-32. | |
318 | * | |
319 | * 32-bit UTF-32 (actually 21-bit) maps to UTF-8 as follows... | |
320 | * | |
321 | * UTF-32 char UTF-8 char(s) | |
322 | * -------------------------------------------------- | |
e1d6a774 | 323 | * 0 to 127 = 0xxxxxxx (US-ASCII) |
ef416fc2 | 324 | * 128 to 2047 = 110xxxxx 10yyyyyy |
325 | * 2048 to 65535 = 1110xxxx 10yyyyyy 10zzzzzz | |
e1d6a774 | 326 | * > 65535 = 11110xxx 10yyyyyy 10zzzzzz 10xxxxxx |
ef416fc2 | 327 | * |
328 | * UTF-32 prohibits chars beyond Plane 16 (> 0x10ffff) in UCS-4, | |
329 | * which would convert to five- or six-octet UTF-8 sequences... | |
ef416fc2 | 330 | */ |
e1d6a774 | 331 | |
332 | int /* O - Count or -1 on error */ | |
333 | cupsUTF8ToUTF32( | |
334 | cups_utf32_t *dest, /* O - Target string */ | |
335 | const cups_utf8_t *src, /* I - Source string */ | |
336 | const int maxout) /* I - Max output */ | |
ef416fc2 | 337 | { |
e1d6a774 | 338 | int i; /* Looping variable */ |
339 | cups_utf8_t ch; /* Character value */ | |
340 | cups_utf8_t next; /* Next character value */ | |
341 | cups_utf32_t ch32; /* UTF-32 character value */ | |
342 | ||
ef416fc2 | 343 | |
344 | /* | |
345 | * Check for valid arguments and clear output... | |
346 | */ | |
e1d6a774 | 347 | |
807315e6 | 348 | DEBUG_printf(("2cupsUTF8ToUTF32(dest=%p, src=\"%s\", maxout=%d)", (void *)dest, src, maxout)); |
c9fc04c6 | 349 | |
e1d6a774 | 350 | if (dest) |
351 | *dest = 0; | |
352 | ||
353 | if (!dest || !src || maxout < 1 || maxout > CUPS_MAX_USTRING) | |
c9fc04c6 | 354 | { |
e07d4801 | 355 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad arguments)"); |
c9fc04c6 | 356 | |
ef416fc2 | 357 | return (-1); |
c9fc04c6 | 358 | } |
ef416fc2 | 359 | |
360 | /* | |
cda47a96 | 361 | * Convert input UTF-8 to output UTF-32... |
ef416fc2 | 362 | */ |
e1d6a774 | 363 | |
e1d6a774 | 364 | for (i = maxout - 1; *src && i > 0; i --) |
ef416fc2 | 365 | { |
e1d6a774 | 366 | ch = *src++; |
ef416fc2 | 367 | |
368 | /* | |
369 | * Convert UTF-8 character(s) to UTF-32 character... | |
370 | */ | |
e1d6a774 | 371 | |
372 | if (!(ch & 0x80)) | |
ef416fc2 | 373 | { |
374 | /* | |
375 | * One-octet UTF-8 <= 127 (US-ASCII)... | |
376 | */ | |
e1d6a774 | 377 | |
378 | *dest++ = ch; | |
c9fc04c6 | 379 | |
e07d4801 | 380 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x => %08X", src[-1], ch)); |
2abf387c | 381 | continue; |
ef416fc2 | 382 | } |
383 | else if ((ch & 0xe0) == 0xc0) | |
384 | { | |
385 | /* | |
386 | * Two-octet UTF-8 <= 2047 (Latin-x)... | |
387 | */ | |
e1d6a774 | 388 | |
389 | next = *src++; | |
c9fc04c6 MS |
390 | if ((next & 0xc0) != 0x80) |
391 | { | |
e07d4801 | 392 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 393 | |
ef416fc2 | 394 | return (-1); |
c9fc04c6 | 395 | } |
e1d6a774 | 396 | |
7e86f2f6 | 397 | ch32 = (cups_utf32_t)((ch & 0x1f) << 6) | (cups_utf32_t)(next & 0x3f); |
ef416fc2 | 398 | |
399 | /* | |
400 | * Check for non-shortest form (invalid UTF-8)... | |
401 | */ | |
e1d6a774 | 402 | |
403 | if (ch32 < 0x80) | |
c9fc04c6 | 404 | { |
e07d4801 | 405 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 406 | |
ef416fc2 | 407 | return (-1); |
c9fc04c6 | 408 | } |
e1d6a774 | 409 | |
410 | *dest++ = ch32; | |
c9fc04c6 | 411 | |
e07d4801 | 412 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x %02x => %08X", |
c9fc04c6 | 413 | src[-2], src[-1], (unsigned)ch32)); |
ef416fc2 | 414 | } |
415 | else if ((ch & 0xf0) == 0xe0) | |
416 | { | |
417 | /* | |
418 | * Three-octet UTF-8 <= 65535 (Plane 0 - BMP)... | |
419 | */ | |
e1d6a774 | 420 | |
421 | next = *src++; | |
c9fc04c6 MS |
422 | if ((next & 0xc0) != 0x80) |
423 | { | |
e07d4801 | 424 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 425 | |
ef416fc2 | 426 | return (-1); |
c9fc04c6 | 427 | } |
e1d6a774 | 428 | |
7e86f2f6 | 429 | ch32 = (cups_utf32_t)((ch & 0x0f) << 6) | (cups_utf32_t)(next & 0x3f); |
e1d6a774 | 430 | |
431 | next = *src++; | |
c9fc04c6 MS |
432 | if ((next & 0xc0) != 0x80) |
433 | { | |
e07d4801 | 434 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 435 | |
ef416fc2 | 436 | return (-1); |
c9fc04c6 | 437 | } |
e1d6a774 | 438 | |
7e86f2f6 | 439 | ch32 = (ch32 << 6) | (cups_utf32_t)(next & 0x3f); |
ef416fc2 | 440 | |
441 | /* | |
442 | * Check for non-shortest form (invalid UTF-8)... | |
443 | */ | |
e1d6a774 | 444 | |
445 | if (ch32 < 0x800) | |
c9fc04c6 | 446 | { |
e07d4801 | 447 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 448 | |
ef416fc2 | 449 | return (-1); |
c9fc04c6 | 450 | } |
e1d6a774 | 451 | |
452 | *dest++ = ch32; | |
c9fc04c6 | 453 | |
e07d4801 | 454 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x %02x %02x => %08X", |
c9fc04c6 | 455 | src[-3], src[-2], src[-1], (unsigned)ch32)); |
ef416fc2 | 456 | } |
457 | else if ((ch & 0xf8) == 0xf0) | |
458 | { | |
459 | /* | |
e1d6a774 | 460 | * Four-octet UTF-8... |
ef416fc2 | 461 | */ |
e1d6a774 | 462 | |
463 | next = *src++; | |
c9fc04c6 MS |
464 | if ((next & 0xc0) != 0x80) |
465 | { | |
e07d4801 | 466 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 467 | |
ef416fc2 | 468 | return (-1); |
c9fc04c6 | 469 | } |
e1d6a774 | 470 | |
7e86f2f6 | 471 | ch32 = (cups_utf32_t)((ch & 0x07) << 6) | (cups_utf32_t)(next & 0x3f); |
e1d6a774 | 472 | |
473 | next = *src++; | |
c9fc04c6 MS |
474 | if ((next & 0xc0) != 0x80) |
475 | { | |
e07d4801 | 476 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 477 | |
e1d6a774 | 478 | return (-1); |
c9fc04c6 | 479 | } |
e1d6a774 | 480 | |
7e86f2f6 | 481 | ch32 = (ch32 << 6) | (cups_utf32_t)(next & 0x3f); |
e1d6a774 | 482 | |
483 | next = *src++; | |
c9fc04c6 MS |
484 | if ((next & 0xc0) != 0x80) |
485 | { | |
e07d4801 | 486 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 487 | |
e1d6a774 | 488 | return (-1); |
c9fc04c6 | 489 | } |
e1d6a774 | 490 | |
7e86f2f6 | 491 | ch32 = (ch32 << 6) | (cups_utf32_t)(next & 0x3f); |
e1d6a774 | 492 | |
ef416fc2 | 493 | /* |
e1d6a774 | 494 | * Check for non-shortest form (invalid UTF-8)... |
ef416fc2 | 495 | */ |
e1d6a774 | 496 | |
497 | if (ch32 < 0x10000) | |
c9fc04c6 | 498 | { |
e07d4801 | 499 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 500 | |
e1d6a774 | 501 | return (-1); |
c9fc04c6 | 502 | } |
e1d6a774 | 503 | |
504 | *dest++ = ch32; | |
c9fc04c6 | 505 | |
e07d4801 | 506 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x %02x %02x %02x => %08X", |
c9fc04c6 | 507 | src[-4], src[-3], src[-2], src[-1], (unsigned)ch32)); |
ef416fc2 | 508 | } |
509 | else | |
510 | { | |
511 | /* | |
e1d6a774 | 512 | * More than 4-octet (invalid UTF-8 sequence)... |
ef416fc2 | 513 | */ |
e1d6a774 | 514 | |
e07d4801 | 515 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 516 | |
ef416fc2 | 517 | return (-1); |
518 | } | |
519 | ||
520 | /* | |
521 | * Check for UTF-16 surrogate (illegal UTF-8)... | |
522 | */ | |
ef416fc2 | 523 | |
2abf387c | 524 | if (ch32 >= 0xd800 && ch32 <= 0xdfff) |
ef416fc2 | 525 | return (-1); |
526 | } | |
e1d6a774 | 527 | |
ef416fc2 | 528 | *dest = 0; |
e1d6a774 | 529 | |
e07d4801 | 530 | DEBUG_printf(("3cupsUTF8ToUTF32: Returning %d characters", maxout - 1 - i)); |
c9fc04c6 MS |
531 | |
532 | return (maxout - 1 - i); | |
ef416fc2 | 533 | } |
534 | ||
e1d6a774 | 535 | |
ef416fc2 | 536 | /* |
537 | * 'cupsUTF32ToUTF8()' - Convert UTF-32 to UTF-8. | |
538 | * | |
539 | * 32-bit UTF-32 (actually 21-bit) maps to UTF-8 as follows... | |
540 | * | |
541 | * UTF-32 char UTF-8 char(s) | |
542 | * -------------------------------------------------- | |
e1d6a774 | 543 | * 0 to 127 = 0xxxxxxx (US-ASCII) |
ef416fc2 | 544 | * 128 to 2047 = 110xxxxx 10yyyyyy |
545 | * 2048 to 65535 = 1110xxxx 10yyyyyy 10zzzzzz | |
e1d6a774 | 546 | * > 65535 = 11110xxx 10yyyyyy 10zzzzzz 10xxxxxx |
ef416fc2 | 547 | * |
548 | * UTF-32 prohibits chars beyond Plane 16 (> 0x10ffff) in UCS-4, | |
549 | * which would convert to five- or six-octet UTF-8 sequences... | |
ef416fc2 | 550 | */ |
e1d6a774 | 551 | |
552 | int /* O - Count or -1 on error */ | |
553 | cupsUTF32ToUTF8( | |
554 | cups_utf8_t *dest, /* O - Target string */ | |
555 | const cups_utf32_t *src, /* I - Source string */ | |
556 | const int maxout) /* I - Max output */ | |
ef416fc2 | 557 | { |
e1d6a774 | 558 | cups_utf8_t *start; /* Start of destination string */ |
559 | int i; /* Looping variable */ | |
560 | int swap; /* Byte-swap input to output */ | |
561 | cups_utf32_t ch; /* Character value */ | |
562 | ||
ef416fc2 | 563 | |
564 | /* | |
565 | * Check for valid arguments and clear output... | |
566 | */ | |
e1d6a774 | 567 | |
807315e6 | 568 | DEBUG_printf(("2cupsUTF32ToUTF8(dest=%p, src=%p, maxout=%d)", (void *)dest, (void *)src, maxout)); |
c9fc04c6 | 569 | |
e1d6a774 | 570 | if (dest) |
571 | *dest = '\0'; | |
572 | ||
573 | if (!dest || !src || maxout < 1) | |
c9fc04c6 | 574 | { |
e07d4801 | 575 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (bad args)"); |
c9fc04c6 | 576 | |
ef416fc2 | 577 | return (-1); |
c9fc04c6 | 578 | } |
ef416fc2 | 579 | |
580 | /* | |
581 | * Check for leading BOM in UTF-32 and inverted BOM... | |
582 | */ | |
e1d6a774 | 583 | |
584 | start = dest; | |
585 | swap = *src == 0xfffe0000; | |
586 | ||
e07d4801 | 587 | DEBUG_printf(("4cupsUTF32ToUTF8: swap=%d", swap)); |
c9fc04c6 | 588 | |
e1d6a774 | 589 | if (*src == 0xfffe0000 || *src == 0xfeff) |
590 | src ++; | |
ef416fc2 | 591 | |
592 | /* | |
593 | * Convert input UTF-32 to output UTF-8... | |
594 | */ | |
e1d6a774 | 595 | |
596 | for (i = maxout - 1; *src && i > 0;) | |
ef416fc2 | 597 | { |
e1d6a774 | 598 | ch = *src++; |
ef416fc2 | 599 | |
600 | /* | |
601 | * Byte swap input UTF-32, if necessary... | |
e1d6a774 | 602 | * (only byte-swapping 24 of 32 bits) |
ef416fc2 | 603 | */ |
e1d6a774 | 604 | |
ef416fc2 | 605 | if (swap) |
606 | ch = ((ch >> 24) | ((ch >> 8) & 0xff00) | ((ch << 8) & 0xff0000)); | |
607 | ||
608 | /* | |
e1d6a774 | 609 | * Check for beyond Plane 16 (invalid UTF-32)... |
ef416fc2 | 610 | */ |
ef416fc2 | 611 | |
ef416fc2 | 612 | if (ch > 0x10ffff) |
c9fc04c6 | 613 | { |
e07d4801 | 614 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (character out of range)"); |
c9fc04c6 | 615 | |
ef416fc2 | 616 | return (-1); |
c9fc04c6 | 617 | } |
ef416fc2 | 618 | |
ef416fc2 | 619 | /* |
620 | * Convert UTF-32 character to UTF-8 character(s)... | |
621 | */ | |
e1d6a774 | 622 | |
623 | if (ch < 0x80) | |
ef416fc2 | 624 | { |
625 | /* | |
626 | * One-octet UTF-8 <= 127 (US-ASCII)... | |
627 | */ | |
e1d6a774 | 628 | |
629 | *dest++ = (cups_utf8_t)ch; | |
630 | i --; | |
c9fc04c6 | 631 | |
e07d4801 | 632 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x", (unsigned)ch, dest[-1])); |
ef416fc2 | 633 | } |
e1d6a774 | 634 | else if (ch < 0x800) |
ef416fc2 | 635 | { |
636 | /* | |
637 | * Two-octet UTF-8 <= 2047 (Latin-x)... | |
638 | */ | |
e1d6a774 | 639 | |
640 | if (i < 2) | |
c9fc04c6 | 641 | { |
e07d4801 | 642 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 2)"); |
c9fc04c6 | 643 | |
e1d6a774 | 644 | return (-1); |
c9fc04c6 | 645 | } |
e1d6a774 | 646 | |
647 | *dest++ = (cups_utf8_t)(0xc0 | ((ch >> 6) & 0x1f)); | |
648 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
649 | i -= 2; | |
c9fc04c6 | 650 | |
e07d4801 | 651 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x", (unsigned)ch, |
c9fc04c6 | 652 | dest[-2], dest[-1])); |
ef416fc2 | 653 | } |
e1d6a774 | 654 | else if (ch < 0x10000) |
ef416fc2 | 655 | { |
656 | /* | |
657 | * Three-octet UTF-8 <= 65535 (Plane 0 - BMP)... | |
658 | */ | |
e1d6a774 | 659 | |
660 | if (i < 3) | |
c9fc04c6 | 661 | { |
e07d4801 | 662 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 3)"); |
c9fc04c6 | 663 | |
e1d6a774 | 664 | return (-1); |
c9fc04c6 | 665 | } |
e1d6a774 | 666 | |
667 | *dest++ = (cups_utf8_t)(0xe0 | ((ch >> 12) & 0x0f)); | |
668 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 6) & 0x3f)); | |
669 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
670 | i -= 3; | |
c9fc04c6 | 671 | |
e07d4801 | 672 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x %02x", (unsigned)ch, |
c9fc04c6 | 673 | dest[-3], dest[-2], dest[-1])); |
e1d6a774 | 674 | } |
675 | else | |
676 | { | |
677 | /* | |
678 | * Four-octet UTF-8... | |
679 | */ | |
680 | ||
681 | if (i < 4) | |
e07d4801 MS |
682 | { |
683 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 4)"); | |
684 | ||
e1d6a774 | 685 | return (-1); |
e07d4801 | 686 | } |
e1d6a774 | 687 | |
688 | *dest++ = (cups_utf8_t)(0xf0 | ((ch >> 18) & 0x07)); | |
689 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 12) & 0x3f)); | |
690 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 6) & 0x3f)); | |
691 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
692 | i -= 4; | |
c9fc04c6 | 693 | |
e07d4801 | 694 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x %02x %02x", |
c9fc04c6 | 695 | (unsigned)ch, dest[-4], dest[-3], dest[-2], dest[-1])); |
ef416fc2 | 696 | } |
697 | } | |
e1d6a774 | 698 | |
ef416fc2 | 699 | *dest = '\0'; |
e1d6a774 | 700 | |
e07d4801 | 701 | DEBUG_printf(("3cupsUTF32ToUTF8: Returning %d", (int)(dest - start))); |
c9fc04c6 | 702 | |
e1d6a774 | 703 | return ((int)(dest - start)); |
ef416fc2 | 704 | } |