]>
Commit | Line | Data |
---|---|---|
ef416fc2 | 1 | /* |
f2d18633 | 2 | * "$Id$" |
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 */ |
84315f46 | 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 | { | |
f99f3698 MS |
174 | char *altdestptr = (char *)dest; /* Silence bogus GCC type-punned */ |
175 | ||
cc754834 MS |
176 | srclen = strlen(src); |
177 | outBytesLeft = maxout - 1; | |
4220952d | 178 | |
f99f3698 MS |
179 | iconv(map_to_utf8, (char **)&src, &srclen, &altdestptr, &outBytesLeft); |
180 | *altdestptr = '\0'; | |
cc754834 MS |
181 | |
182 | _cupsMutexUnlock(&map_mutex); | |
183 | ||
f99f3698 | 184 | return ((int)(altdestptr - (char *)dest)); |
cc754834 | 185 | } |
d6ae789d | 186 | |
6d2f911b | 187 | _cupsMutexUnlock(&map_mutex); |
cc754834 | 188 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 189 | |
cc754834 MS |
190 | /* |
191 | * No iconv() support, so error out... | |
192 | */ | |
193 | ||
194 | *destptr = '\0'; | |
195 | ||
196 | return (-1); | |
ef416fc2 | 197 | } |
198 | ||
e1d6a774 | 199 | |
ef416fc2 | 200 | /* |
e1d6a774 | 201 | * 'cupsUTF8ToCharset()' - Convert UTF-8 to legacy character set. |
ef416fc2 | 202 | */ |
e1d6a774 | 203 | |
204 | int /* O - Count or -1 on error */ | |
205 | cupsUTF8ToCharset( | |
206 | char *dest, /* O - Target string */ | |
207 | const cups_utf8_t *src, /* I - Source string */ | |
208 | const int maxout, /* I - Max output */ | |
209 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 210 | { |
cc754834 | 211 | char *destptr; /* Pointer into destination */ |
84315f46 | 212 | #ifdef HAVE_ICONV_H |
cc754834 MS |
213 | size_t srclen, /* Length of source string */ |
214 | outBytesLeft; /* Bytes remaining in output buffer */ | |
7cf5915e | 215 | #endif /* HAVE_ICONV_H */ |
d6ae789d | 216 | |
217 | ||
ef416fc2 | 218 | /* |
219 | * Check for valid arguments... | |
220 | */ | |
221 | ||
cc754834 | 222 | if (!dest || !src || maxout < 1) |
e1d6a774 | 223 | { |
224 | if (dest) | |
225 | *dest = '\0'; | |
226 | ||
ef416fc2 | 227 | return (-1); |
e1d6a774 | 228 | } |
ef416fc2 | 229 | |
230 | /* | |
231 | * Handle identity conversions... | |
232 | */ | |
233 | ||
22c9029b | 234 | if (encoding == CUPS_UTF8 || |
cc754834 | 235 | encoding >= CUPS_ENCODING_VBCS_END) |
ef416fc2 | 236 | { |
e1d6a774 | 237 | strlcpy(dest, (char *)src, maxout); |
b86bc4cf | 238 | return ((int)strlen(dest)); |
ef416fc2 | 239 | } |
240 | ||
411affcf | 241 | /* |
242 | * Handle UTF-8 to ISO-8859-1 directly... | |
243 | */ | |
244 | ||
cc754834 MS |
245 | destptr = dest; |
246 | ||
22c9029b | 247 | if (encoding == CUPS_ISO8859_1 || encoding <= CUPS_US_ASCII) |
411affcf | 248 | { |
22c9029b MS |
249 | int ch, /* Character from string */ |
250 | maxch; /* Maximum character for charset */ | |
cc754834 | 251 | char *destend; /* End of ISO-8859-1 buffer */ |
411affcf | 252 | |
22c9029b | 253 | maxch = encoding == CUPS_ISO8859_1 ? 256 : 128; |
411affcf | 254 | destend = dest + maxout - 1; |
255 | ||
256 | while (*src && destptr < destend) | |
257 | { | |
258 | ch = *src++; | |
259 | ||
260 | if ((ch & 0xe0) == 0xc0) | |
261 | { | |
262 | ch = ((ch & 0x1f) << 6) | (*src++ & 0x3f); | |
263 | ||
22c9029b | 264 | if (ch < maxch) |
411affcf | 265 | *destptr++ = ch; |
266 | else | |
267 | *destptr++ = '?'; | |
268 | } | |
269 | else if ((ch & 0xf0) == 0xe0 || | |
270 | (ch & 0xf8) == 0xf0) | |
271 | *destptr++ = '?'; | |
272 | else if (!(ch & 0x80)) | |
273 | *destptr++ = ch; | |
274 | } | |
275 | ||
276 | *destptr = '\0'; | |
277 | ||
b86bc4cf | 278 | return ((int)(destptr - dest)); |
411affcf | 279 | } |
280 | ||
cc754834 | 281 | #ifdef HAVE_ICONV_H |
ef416fc2 | 282 | /* |
e1d6a774 | 283 | * Convert input UTF-8 to legacy charset... |
ef416fc2 | 284 | */ |
e1d6a774 | 285 | |
6d2f911b | 286 | _cupsMutexLock(&map_mutex); |
d6ae789d | 287 | |
cc754834 MS |
288 | if (map_encoding != encoding) |
289 | { | |
290 | _cupsCharmapFlush(); | |
291 | ||
292 | map_from_utf8 = iconv_open(_cupsEncodingName(encoding), "UTF-8"); | |
293 | map_to_utf8 = iconv_open("UTF-8", _cupsEncodingName(encoding)); | |
294 | map_encoding = encoding; | |
295 | } | |
296 | ||
297 | if (map_from_utf8 != (iconv_t)-1) | |
298 | { | |
f99f3698 MS |
299 | char *altsrc = (char *)src; /* Silence bogus GCC type-punned */ |
300 | ||
cc754834 MS |
301 | srclen = strlen((char *)src); |
302 | outBytesLeft = maxout - 1; | |
4220952d | 303 | |
f99f3698 | 304 | iconv(map_from_utf8, &altsrc, &srclen, &destptr, &outBytesLeft); |
4220952d | 305 | *destptr = '\0'; |
cc754834 MS |
306 | |
307 | _cupsMutexUnlock(&map_mutex); | |
308 | ||
309 | return ((int)(destptr - dest)); | |
310 | } | |
d6ae789d | 311 | |
6d2f911b | 312 | _cupsMutexUnlock(&map_mutex); |
cc754834 MS |
313 | #endif /* HAVE_ICONV_H */ |
314 | ||
315 | /* | |
316 | * No iconv() support, so error out... | |
317 | */ | |
318 | ||
319 | *destptr = '\0'; | |
d6ae789d | 320 | |
cc754834 | 321 | return (-1); |
ef416fc2 | 322 | } |
323 | ||
ef416fc2 | 324 | |
325 | /* | |
326 | * 'cupsUTF8ToUTF32()' - Convert UTF-8 to UTF-32. | |
327 | * | |
328 | * 32-bit UTF-32 (actually 21-bit) maps to UTF-8 as follows... | |
329 | * | |
330 | * UTF-32 char UTF-8 char(s) | |
331 | * -------------------------------------------------- | |
e1d6a774 | 332 | * 0 to 127 = 0xxxxxxx (US-ASCII) |
ef416fc2 | 333 | * 128 to 2047 = 110xxxxx 10yyyyyy |
334 | * 2048 to 65535 = 1110xxxx 10yyyyyy 10zzzzzz | |
e1d6a774 | 335 | * > 65535 = 11110xxx 10yyyyyy 10zzzzzz 10xxxxxx |
ef416fc2 | 336 | * |
337 | * UTF-32 prohibits chars beyond Plane 16 (> 0x10ffff) in UCS-4, | |
338 | * which would convert to five- or six-octet UTF-8 sequences... | |
ef416fc2 | 339 | */ |
e1d6a774 | 340 | |
341 | int /* O - Count or -1 on error */ | |
342 | cupsUTF8ToUTF32( | |
343 | cups_utf32_t *dest, /* O - Target string */ | |
344 | const cups_utf8_t *src, /* I - Source string */ | |
345 | const int maxout) /* I - Max output */ | |
ef416fc2 | 346 | { |
e1d6a774 | 347 | int i; /* Looping variable */ |
348 | cups_utf8_t ch; /* Character value */ | |
349 | cups_utf8_t next; /* Next character value */ | |
350 | cups_utf32_t ch32; /* UTF-32 character value */ | |
351 | ||
ef416fc2 | 352 | |
353 | /* | |
354 | * Check for valid arguments and clear output... | |
355 | */ | |
e1d6a774 | 356 | |
e07d4801 MS |
357 | DEBUG_printf(("2cupsUTF8ToUTF32(dest=%p, src=\"%s\", maxout=%d)", dest, |
358 | src, maxout)); | |
c9fc04c6 | 359 | |
e1d6a774 | 360 | if (dest) |
361 | *dest = 0; | |
362 | ||
363 | if (!dest || !src || maxout < 1 || maxout > CUPS_MAX_USTRING) | |
c9fc04c6 | 364 | { |
e07d4801 | 365 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad arguments)"); |
c9fc04c6 | 366 | |
ef416fc2 | 367 | return (-1); |
c9fc04c6 | 368 | } |
ef416fc2 | 369 | |
370 | /* | |
cda47a96 | 371 | * Convert input UTF-8 to output UTF-32... |
ef416fc2 | 372 | */ |
e1d6a774 | 373 | |
e1d6a774 | 374 | for (i = maxout - 1; *src && i > 0; i --) |
ef416fc2 | 375 | { |
e1d6a774 | 376 | ch = *src++; |
ef416fc2 | 377 | |
378 | /* | |
379 | * Convert UTF-8 character(s) to UTF-32 character... | |
380 | */ | |
e1d6a774 | 381 | |
382 | if (!(ch & 0x80)) | |
ef416fc2 | 383 | { |
384 | /* | |
385 | * One-octet UTF-8 <= 127 (US-ASCII)... | |
386 | */ | |
e1d6a774 | 387 | |
388 | *dest++ = ch; | |
c9fc04c6 | 389 | |
e07d4801 | 390 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x => %08X", src[-1], ch)); |
2abf387c | 391 | continue; |
ef416fc2 | 392 | } |
393 | else if ((ch & 0xe0) == 0xc0) | |
394 | { | |
395 | /* | |
396 | * Two-octet UTF-8 <= 2047 (Latin-x)... | |
397 | */ | |
e1d6a774 | 398 | |
399 | next = *src++; | |
c9fc04c6 MS |
400 | if ((next & 0xc0) != 0x80) |
401 | { | |
e07d4801 | 402 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 403 | |
ef416fc2 | 404 | return (-1); |
c9fc04c6 | 405 | } |
e1d6a774 | 406 | |
ef416fc2 | 407 | ch32 = ((ch & 0x1f) << 6) | (next & 0x3f); |
408 | ||
409 | /* | |
410 | * Check for non-shortest form (invalid UTF-8)... | |
411 | */ | |
e1d6a774 | 412 | |
413 | if (ch32 < 0x80) | |
c9fc04c6 | 414 | { |
e07d4801 | 415 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 416 | |
ef416fc2 | 417 | return (-1); |
c9fc04c6 | 418 | } |
e1d6a774 | 419 | |
420 | *dest++ = ch32; | |
c9fc04c6 | 421 | |
e07d4801 | 422 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x %02x => %08X", |
c9fc04c6 | 423 | src[-2], src[-1], (unsigned)ch32)); |
ef416fc2 | 424 | } |
425 | else if ((ch & 0xf0) == 0xe0) | |
426 | { | |
427 | /* | |
428 | * Three-octet UTF-8 <= 65535 (Plane 0 - BMP)... | |
429 | */ | |
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 | |
439 | ch32 = ((ch & 0x0f) << 6) | (next & 0x3f); | |
440 | ||
441 | next = *src++; | |
c9fc04c6 MS |
442 | if ((next & 0xc0) != 0x80) |
443 | { | |
e07d4801 | 444 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 445 | |
ef416fc2 | 446 | return (-1); |
c9fc04c6 | 447 | } |
e1d6a774 | 448 | |
449 | ch32 = (ch32 << 6) | (next & 0x3f); | |
ef416fc2 | 450 | |
451 | /* | |
452 | * Check for non-shortest form (invalid UTF-8)... | |
453 | */ | |
e1d6a774 | 454 | |
455 | if (ch32 < 0x800) | |
c9fc04c6 | 456 | { |
e07d4801 | 457 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 458 | |
ef416fc2 | 459 | return (-1); |
c9fc04c6 | 460 | } |
e1d6a774 | 461 | |
462 | *dest++ = ch32; | |
c9fc04c6 | 463 | |
e07d4801 | 464 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x %02x %02x => %08X", |
c9fc04c6 | 465 | src[-3], src[-2], src[-1], (unsigned)ch32)); |
ef416fc2 | 466 | } |
467 | else if ((ch & 0xf8) == 0xf0) | |
468 | { | |
469 | /* | |
e1d6a774 | 470 | * Four-octet UTF-8... |
ef416fc2 | 471 | */ |
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 | |
ef416fc2 | 478 | return (-1); |
c9fc04c6 | 479 | } |
e1d6a774 | 480 | |
481 | ch32 = ((ch & 0x07) << 6) | (next & 0x3f); | |
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 | |
491 | ch32 = (ch32 << 6) | (next & 0x3f); | |
492 | ||
493 | next = *src++; | |
c9fc04c6 MS |
494 | if ((next & 0xc0) != 0x80) |
495 | { | |
e07d4801 | 496 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 497 | |
e1d6a774 | 498 | return (-1); |
c9fc04c6 | 499 | } |
e1d6a774 | 500 | |
501 | ch32 = (ch32 << 6) | (next & 0x3f); | |
502 | ||
ef416fc2 | 503 | /* |
e1d6a774 | 504 | * Check for non-shortest form (invalid UTF-8)... |
ef416fc2 | 505 | */ |
e1d6a774 | 506 | |
507 | if (ch32 < 0x10000) | |
c9fc04c6 | 508 | { |
e07d4801 | 509 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 510 | |
e1d6a774 | 511 | return (-1); |
c9fc04c6 | 512 | } |
e1d6a774 | 513 | |
514 | *dest++ = ch32; | |
c9fc04c6 | 515 | |
e07d4801 | 516 | DEBUG_printf(("4cupsUTF8ToUTF32: %02x %02x %02x %02x => %08X", |
c9fc04c6 | 517 | src[-4], src[-3], src[-2], src[-1], (unsigned)ch32)); |
ef416fc2 | 518 | } |
519 | else | |
520 | { | |
521 | /* | |
e1d6a774 | 522 | * More than 4-octet (invalid UTF-8 sequence)... |
ef416fc2 | 523 | */ |
e1d6a774 | 524 | |
e07d4801 | 525 | DEBUG_puts("3cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
c9fc04c6 | 526 | |
ef416fc2 | 527 | return (-1); |
528 | } | |
529 | ||
530 | /* | |
531 | * Check for UTF-16 surrogate (illegal UTF-8)... | |
532 | */ | |
ef416fc2 | 533 | |
2abf387c | 534 | if (ch32 >= 0xd800 && ch32 <= 0xdfff) |
ef416fc2 | 535 | return (-1); |
536 | } | |
e1d6a774 | 537 | |
ef416fc2 | 538 | *dest = 0; |
e1d6a774 | 539 | |
e07d4801 | 540 | DEBUG_printf(("3cupsUTF8ToUTF32: Returning %d characters", maxout - 1 - i)); |
c9fc04c6 MS |
541 | |
542 | return (maxout - 1 - i); | |
ef416fc2 | 543 | } |
544 | ||
e1d6a774 | 545 | |
ef416fc2 | 546 | /* |
547 | * 'cupsUTF32ToUTF8()' - Convert UTF-32 to UTF-8. | |
548 | * | |
549 | * 32-bit UTF-32 (actually 21-bit) maps to UTF-8 as follows... | |
550 | * | |
551 | * UTF-32 char UTF-8 char(s) | |
552 | * -------------------------------------------------- | |
e1d6a774 | 553 | * 0 to 127 = 0xxxxxxx (US-ASCII) |
ef416fc2 | 554 | * 128 to 2047 = 110xxxxx 10yyyyyy |
555 | * 2048 to 65535 = 1110xxxx 10yyyyyy 10zzzzzz | |
e1d6a774 | 556 | * > 65535 = 11110xxx 10yyyyyy 10zzzzzz 10xxxxxx |
ef416fc2 | 557 | * |
558 | * UTF-32 prohibits chars beyond Plane 16 (> 0x10ffff) in UCS-4, | |
559 | * which would convert to five- or six-octet UTF-8 sequences... | |
ef416fc2 | 560 | */ |
e1d6a774 | 561 | |
562 | int /* O - Count or -1 on error */ | |
563 | cupsUTF32ToUTF8( | |
564 | cups_utf8_t *dest, /* O - Target string */ | |
565 | const cups_utf32_t *src, /* I - Source string */ | |
566 | const int maxout) /* I - Max output */ | |
ef416fc2 | 567 | { |
e1d6a774 | 568 | cups_utf8_t *start; /* Start of destination string */ |
569 | int i; /* Looping variable */ | |
570 | int swap; /* Byte-swap input to output */ | |
571 | cups_utf32_t ch; /* Character value */ | |
572 | ||
ef416fc2 | 573 | |
574 | /* | |
575 | * Check for valid arguments and clear output... | |
576 | */ | |
e1d6a774 | 577 | |
e07d4801 | 578 | DEBUG_printf(("2cupsUTF32ToUTF8(dest=%p, src=%p, maxout=%d)", dest, src, |
c9fc04c6 MS |
579 | maxout)); |
580 | ||
e1d6a774 | 581 | if (dest) |
582 | *dest = '\0'; | |
583 | ||
584 | if (!dest || !src || maxout < 1) | |
c9fc04c6 | 585 | { |
e07d4801 | 586 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (bad args)"); |
c9fc04c6 | 587 | |
ef416fc2 | 588 | return (-1); |
c9fc04c6 | 589 | } |
ef416fc2 | 590 | |
591 | /* | |
592 | * Check for leading BOM in UTF-32 and inverted BOM... | |
593 | */ | |
e1d6a774 | 594 | |
595 | start = dest; | |
596 | swap = *src == 0xfffe0000; | |
597 | ||
e07d4801 | 598 | DEBUG_printf(("4cupsUTF32ToUTF8: swap=%d", swap)); |
c9fc04c6 | 599 | |
e1d6a774 | 600 | if (*src == 0xfffe0000 || *src == 0xfeff) |
601 | src ++; | |
ef416fc2 | 602 | |
603 | /* | |
604 | * Convert input UTF-32 to output UTF-8... | |
605 | */ | |
e1d6a774 | 606 | |
607 | for (i = maxout - 1; *src && i > 0;) | |
ef416fc2 | 608 | { |
e1d6a774 | 609 | ch = *src++; |
ef416fc2 | 610 | |
611 | /* | |
612 | * Byte swap input UTF-32, if necessary... | |
e1d6a774 | 613 | * (only byte-swapping 24 of 32 bits) |
ef416fc2 | 614 | */ |
e1d6a774 | 615 | |
ef416fc2 | 616 | if (swap) |
617 | ch = ((ch >> 24) | ((ch >> 8) & 0xff00) | ((ch << 8) & 0xff0000)); | |
618 | ||
619 | /* | |
e1d6a774 | 620 | * Check for beyond Plane 16 (invalid UTF-32)... |
ef416fc2 | 621 | */ |
ef416fc2 | 622 | |
ef416fc2 | 623 | if (ch > 0x10ffff) |
c9fc04c6 | 624 | { |
e07d4801 | 625 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (character out of range)"); |
c9fc04c6 | 626 | |
ef416fc2 | 627 | return (-1); |
c9fc04c6 | 628 | } |
ef416fc2 | 629 | |
ef416fc2 | 630 | /* |
631 | * Convert UTF-32 character to UTF-8 character(s)... | |
632 | */ | |
e1d6a774 | 633 | |
634 | if (ch < 0x80) | |
ef416fc2 | 635 | { |
636 | /* | |
637 | * One-octet UTF-8 <= 127 (US-ASCII)... | |
638 | */ | |
e1d6a774 | 639 | |
640 | *dest++ = (cups_utf8_t)ch; | |
641 | i --; | |
c9fc04c6 | 642 | |
e07d4801 | 643 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x", (unsigned)ch, dest[-1])); |
ef416fc2 | 644 | } |
e1d6a774 | 645 | else if (ch < 0x800) |
ef416fc2 | 646 | { |
647 | /* | |
648 | * Two-octet UTF-8 <= 2047 (Latin-x)... | |
649 | */ | |
e1d6a774 | 650 | |
651 | if (i < 2) | |
c9fc04c6 | 652 | { |
e07d4801 | 653 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 2)"); |
c9fc04c6 | 654 | |
e1d6a774 | 655 | return (-1); |
c9fc04c6 | 656 | } |
e1d6a774 | 657 | |
658 | *dest++ = (cups_utf8_t)(0xc0 | ((ch >> 6) & 0x1f)); | |
659 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
660 | i -= 2; | |
c9fc04c6 | 661 | |
e07d4801 | 662 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x", (unsigned)ch, |
c9fc04c6 | 663 | dest[-2], dest[-1])); |
ef416fc2 | 664 | } |
e1d6a774 | 665 | else if (ch < 0x10000) |
ef416fc2 | 666 | { |
667 | /* | |
668 | * Three-octet UTF-8 <= 65535 (Plane 0 - BMP)... | |
669 | */ | |
e1d6a774 | 670 | |
671 | if (i < 3) | |
c9fc04c6 | 672 | { |
e07d4801 | 673 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 3)"); |
c9fc04c6 | 674 | |
e1d6a774 | 675 | return (-1); |
c9fc04c6 | 676 | } |
e1d6a774 | 677 | |
678 | *dest++ = (cups_utf8_t)(0xe0 | ((ch >> 12) & 0x0f)); | |
679 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 6) & 0x3f)); | |
680 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
681 | i -= 3; | |
c9fc04c6 | 682 | |
e07d4801 | 683 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x %02x", (unsigned)ch, |
c9fc04c6 | 684 | dest[-3], dest[-2], dest[-1])); |
e1d6a774 | 685 | } |
686 | else | |
687 | { | |
688 | /* | |
689 | * Four-octet UTF-8... | |
690 | */ | |
691 | ||
692 | if (i < 4) | |
e07d4801 MS |
693 | { |
694 | DEBUG_puts("3cupsUTF32ToUTF8: Returning -1 (too long 4)"); | |
695 | ||
e1d6a774 | 696 | return (-1); |
e07d4801 | 697 | } |
e1d6a774 | 698 | |
699 | *dest++ = (cups_utf8_t)(0xf0 | ((ch >> 18) & 0x07)); | |
700 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 12) & 0x3f)); | |
701 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 6) & 0x3f)); | |
702 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
703 | i -= 4; | |
c9fc04c6 | 704 | |
e07d4801 | 705 | DEBUG_printf(("4cupsUTF32ToUTF8: %08x => %02x %02x %02x %02x", |
c9fc04c6 | 706 | (unsigned)ch, dest[-4], dest[-3], dest[-2], dest[-1])); |
ef416fc2 | 707 | } |
708 | } | |
e1d6a774 | 709 | |
ef416fc2 | 710 | *dest = '\0'; |
e1d6a774 | 711 | |
e07d4801 | 712 | DEBUG_printf(("3cupsUTF32ToUTF8: Returning %d", (int)(dest - start))); |
c9fc04c6 | 713 | |
e1d6a774 | 714 | return ((int)(dest - start)); |
ef416fc2 | 715 | } |
716 | ||
e1d6a774 | 717 | |
ef416fc2 | 718 | /* |
f2d18633 | 719 | * End of "$Id$" |
ef416fc2 | 720 | */ |