]>
Commit | Line | Data |
---|---|---|
ef416fc2 | 1 | /* |
75bd9771 | 2 | * "$Id: transcode.c 7560 2008-05-13 06:34:04Z mike $" |
ef416fc2 | 3 | * |
4 | * Transcoding support for the Common UNIX Printing System (CUPS). | |
5 | * | |
91c84a35 | 6 | * Copyright 2007-2008 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. |
e1d6a774 | 20 | * _cupsCharmapFree() - Free a character set map. |
21 | * _cupsCharmapGet() - Get a character set map. | |
ef416fc2 | 22 | * cupsCharsetToUTF8() - Convert legacy character set to UTF-8. |
e1d6a774 | 23 | * cupsUTF8ToCharset() - Convert UTF-8 to legacy character set. |
ef416fc2 | 24 | * cupsUTF8ToUTF32() - Convert UTF-8 to UTF-32. |
25 | * cupsUTF32ToUTF8() - Convert UTF-32 to UTF-8. | |
e1d6a774 | 26 | * compare_wide() - Compare key for wide (VBCS) match. |
27 | * conv_sbcs_to_utf8() - Convert legacy SBCS to UTF-8. | |
ef416fc2 | 28 | * conv_utf8_to_sbcs() - Convert UTF-8 to legacy SBCS. |
29 | * conv_utf8_to_vbcs() - Convert UTF-8 to legacy DBCS/VBCS. | |
ef416fc2 | 30 | * conv_vbcs_to_utf8() - Convert legacy DBCS/VBCS to UTF-8. |
e1d6a774 | 31 | * free_sbcs_charmap() - Free memory used by a single byte character set. |
32 | * free_vbcs_charmap() - Free memory used by a variable byte character set. | |
d6ae789d | 33 | * get_charmap() - Lookup or get a character set map (private). |
e1d6a774 | 34 | * get_charmap_count() - Count lines in a charmap file. |
35 | * get_sbcs_charmap() - Get SBCS Charmap. | |
36 | * get_vbcs_charmap() - Get DBCS/VBCS Charmap. | |
ef416fc2 | 37 | */ |
38 | ||
39 | /* | |
40 | * Include necessary headers... | |
41 | */ | |
42 | ||
43 | #include "globals.h" | |
e1d6a774 | 44 | #include "debug.h" |
e53920b9 | 45 | #include <limits.h> |
ef416fc2 | 46 | #include <stdlib.h> |
47 | #include <errno.h> | |
48 | #include <time.h> | |
49 | ||
50 | ||
d6ae789d | 51 | /* |
52 | * Local globals... | |
53 | */ | |
54 | ||
55 | #ifdef HAVE_PTHREAD_H | |
56 | static pthread_mutex_t map_mutex = PTHREAD_MUTEX_INITIALIZER; | |
57 | /* Mutex to control access to maps */ | |
58 | #endif /* HAVE_PTHREAD_H */ | |
59 | static _cups_cmap_t *cmap_cache = NULL; | |
60 | /* SBCS Charmap Cache */ | |
61 | static _cups_vmap_t *vmap_cache = NULL; | |
62 | /* VBCS Charmap Cache */ | |
63 | ||
64 | ||
ef416fc2 | 65 | /* |
e1d6a774 | 66 | * Local functions... |
ef416fc2 | 67 | */ |
68 | ||
e1d6a774 | 69 | static int compare_wide(const void *k1, const void *k2); |
70 | static int conv_sbcs_to_utf8(cups_utf8_t *dest, | |
71 | const cups_sbcs_t *src, | |
72 | int maxout, | |
73 | const cups_encoding_t encoding); | |
74 | static int conv_utf8_to_sbcs(cups_sbcs_t *dest, | |
75 | const cups_utf8_t *src, | |
76 | int maxout, | |
77 | const cups_encoding_t encoding); | |
78 | static int conv_utf8_to_vbcs(cups_sbcs_t *dest, | |
79 | const cups_utf8_t *src, | |
80 | int maxout, | |
81 | const cups_encoding_t encoding); | |
82 | static int conv_vbcs_to_utf8(cups_utf8_t *dest, | |
83 | const cups_sbcs_t *src, | |
84 | int maxout, | |
85 | const cups_encoding_t encoding); | |
86 | static void free_sbcs_charmap(_cups_cmap_t *sbcs); | |
87 | static void free_vbcs_charmap(_cups_vmap_t *vbcs); | |
d6ae789d | 88 | static void *get_charmap(const cups_encoding_t encoding); |
e1d6a774 | 89 | static int get_charmap_count(cups_file_t *fp); |
90 | static _cups_cmap_t *get_sbcs_charmap(const cups_encoding_t encoding, | |
91 | const char *filename); | |
92 | static _cups_vmap_t *get_vbcs_charmap(const cups_encoding_t encoding, | |
93 | const char *filename); | |
94 | ||
ef416fc2 | 95 | |
96 | /* | |
e1d6a774 | 97 | * '_cupsCharmapFlush()' - Flush all character set maps out of cache. |
ef416fc2 | 98 | */ |
99 | ||
e1d6a774 | 100 | void |
d6ae789d | 101 | _cupsCharmapFlush(void) |
ef416fc2 | 102 | { |
e1d6a774 | 103 | _cups_cmap_t *cmap, /* Legacy SBCS / Unicode Charset Map */ |
104 | *cnext; /* Next Legacy SBCS Charset Map */ | |
105 | _cups_vmap_t *vmap, /* Legacy VBCS / Unicode Charset Map */ | |
106 | *vnext; /* Next Legacy VBCS Charset Map */ | |
ef416fc2 | 107 | |
108 | ||
d6ae789d | 109 | #ifdef HAVE_PTHREAD_H |
110 | pthread_mutex_lock(&map_mutex); | |
111 | #endif /* HAVE_PTHREAD_H */ | |
112 | ||
ef416fc2 | 113 | /* |
e1d6a774 | 114 | * Loop through SBCS charset map cache, free all memory... |
ef416fc2 | 115 | */ |
116 | ||
d6ae789d | 117 | for (cmap = cmap_cache; cmap; cmap = cnext) |
e1d6a774 | 118 | { |
119 | cnext = cmap->next; | |
ef416fc2 | 120 | |
e1d6a774 | 121 | free_sbcs_charmap(cmap); |
122 | } | |
ef416fc2 | 123 | |
d6ae789d | 124 | cmap_cache = NULL; |
ef416fc2 | 125 | |
126 | /* | |
e1d6a774 | 127 | * Loop through DBCS/VBCS charset map cache, free all memory... |
ef416fc2 | 128 | */ |
129 | ||
d6ae789d | 130 | for (vmap = vmap_cache; vmap; vmap = vnext) |
e1d6a774 | 131 | { |
132 | vnext = vmap->next; | |
133 | ||
134 | free_vbcs_charmap(vmap); | |
e1d6a774 | 135 | } |
136 | ||
d6ae789d | 137 | vmap_cache = NULL; |
138 | ||
139 | #ifdef HAVE_PTHREAD_H | |
140 | pthread_mutex_unlock(&map_mutex); | |
141 | #endif /* HAVE_PTHREAD_H */ | |
ef416fc2 | 142 | } |
143 | ||
e1d6a774 | 144 | |
ef416fc2 | 145 | /* |
e1d6a774 | 146 | * '_cupsCharmapFree()' - Free a character set map. |
ef416fc2 | 147 | * |
e1d6a774 | 148 | * This does not actually free; use '_cupsCharmapFlush()' for that. |
ef416fc2 | 149 | */ |
e1d6a774 | 150 | |
ef416fc2 | 151 | void |
e1d6a774 | 152 | _cupsCharmapFree( |
153 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 154 | { |
e1d6a774 | 155 | _cups_cmap_t *cmap; /* Legacy SBCS / Unicode Charset Map */ |
156 | _cups_vmap_t *vmap; /* Legacy VBCS / Unicode Charset Map */ | |
e1d6a774 | 157 | |
ef416fc2 | 158 | |
159 | /* | |
160 | * See if we already have this SBCS charset map loaded... | |
161 | */ | |
e1d6a774 | 162 | |
d6ae789d | 163 | #ifdef HAVE_PTHREAD_H |
164 | pthread_mutex_lock(&map_mutex); | |
165 | #endif /* HAVE_PTHREAD_H */ | |
166 | ||
167 | for (cmap = cmap_cache; cmap; cmap = cmap->next) | |
ef416fc2 | 168 | { |
169 | if (cmap->encoding == encoding) | |
170 | { | |
171 | if (cmap->used > 0) | |
172 | cmap->used --; | |
d6ae789d | 173 | break; |
ef416fc2 | 174 | } |
175 | } | |
176 | ||
177 | /* | |
178 | * See if we already have this DBCS/VBCS charset map loaded... | |
179 | */ | |
e1d6a774 | 180 | |
d6ae789d | 181 | for (vmap = vmap_cache; vmap; vmap = vmap->next) |
ef416fc2 | 182 | { |
183 | if (vmap->encoding == encoding) | |
184 | { | |
185 | if (vmap->used > 0) | |
186 | vmap->used --; | |
d6ae789d | 187 | break; |
ef416fc2 | 188 | } |
189 | } | |
d6ae789d | 190 | |
191 | #ifdef HAVE_PTHREAD_H | |
192 | pthread_mutex_unlock(&map_mutex); | |
193 | #endif /* HAVE_PTHREAD_H */ | |
fa73b229 | 194 | } |
195 | ||
196 | ||
197 | /* | |
e1d6a774 | 198 | * '_cupsCharmapGet()' - Get a character set map. |
199 | * | |
200 | * This code handles single-byte (SBCS), double-byte (DBCS), and | |
201 | * variable-byte (VBCS) character sets _without_ charset escapes... | |
202 | * This code does not handle multiple-byte character sets (MBCS) | |
203 | * (such as ISO-2022-JP) with charset switching via escapes... | |
fa73b229 | 204 | */ |
205 | ||
e1d6a774 | 206 | void * /* O - Charset map pointer */ |
207 | _cupsCharmapGet( | |
208 | const cups_encoding_t encoding) /* I - Encoding */ | |
fa73b229 | 209 | { |
d6ae789d | 210 | void *charmap; /* Charset map pointer */ |
e1d6a774 | 211 | |
fa73b229 | 212 | |
e1d6a774 | 213 | DEBUG_printf(("_cupsCharmapGet(encoding=%d)\n", encoding)); |
ef416fc2 | 214 | |
215 | /* | |
e1d6a774 | 216 | * Check for valid arguments... |
ef416fc2 | 217 | */ |
e1d6a774 | 218 | |
219 | if (encoding < 0 || encoding >= CUPS_ENCODING_VBCS_END) | |
ef416fc2 | 220 | { |
e1d6a774 | 221 | DEBUG_puts(" Bad encoding, returning NULL!"); |
222 | return (NULL); | |
ef416fc2 | 223 | } |
ef416fc2 | 224 | |
225 | /* | |
d6ae789d | 226 | * Lookup or get the charset map pointer and return... |
ef416fc2 | 227 | */ |
e1d6a774 | 228 | |
d6ae789d | 229 | #ifdef HAVE_PTHREAD_H |
230 | pthread_mutex_lock(&map_mutex); | |
231 | #endif /* HAVE_PTHREAD_H */ | |
e1d6a774 | 232 | |
d6ae789d | 233 | charmap = get_charmap(encoding); |
e1d6a774 | 234 | |
d6ae789d | 235 | #ifdef HAVE_PTHREAD_H |
236 | pthread_mutex_unlock(&map_mutex); | |
237 | #endif /* HAVE_PTHREAD_H */ | |
e1d6a774 | 238 | |
d6ae789d | 239 | return (charmap); |
ef416fc2 | 240 | } |
241 | ||
e1d6a774 | 242 | |
ef416fc2 | 243 | /* |
e1d6a774 | 244 | * 'cupsCharsetToUTF8()' - Convert legacy character set to UTF-8. |
ef416fc2 | 245 | * |
246 | * This code handles single-byte (SBCS), double-byte (DBCS), and | |
247 | * variable-byte (VBCS) character sets _without_ charset escapes... | |
248 | * This code does not handle multiple-byte character sets (MBCS) | |
249 | * (such as ISO-2022-JP) with charset switching via escapes... | |
250 | */ | |
e1d6a774 | 251 | |
252 | int /* O - Count or -1 on error */ | |
253 | cupsCharsetToUTF8( | |
254 | cups_utf8_t *dest, /* O - Target string */ | |
255 | const char *src, /* I - Source string */ | |
256 | const int maxout, /* I - Max output */ | |
257 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 258 | { |
d6ae789d | 259 | int bytes; /* Number of bytes converted */ |
260 | ||
261 | ||
ef416fc2 | 262 | /* |
263 | * Check for valid arguments... | |
264 | */ | |
265 | ||
e1d6a774 | 266 | DEBUG_printf(("cupsCharsetToUTF8(dest=%p, src=\"%s\", maxout=%d, encoding=%d)\n", |
267 | dest, src, maxout, encoding)); | |
268 | ||
269 | if (dest) | |
270 | *dest = '\0'; | |
271 | ||
ef416fc2 | 272 | if (!dest || !src || maxout < 1 || maxout > CUPS_MAX_USTRING) |
e1d6a774 | 273 | { |
274 | DEBUG_puts(" Bad arguments, returning -1"); | |
ef416fc2 | 275 | return (-1); |
e1d6a774 | 276 | } |
ef416fc2 | 277 | |
278 | /* | |
279 | * Handle identity conversions... | |
280 | */ | |
281 | ||
282 | if (encoding == CUPS_UTF8 || | |
283 | encoding < 0 || encoding >= CUPS_ENCODING_VBCS_END) | |
284 | { | |
e1d6a774 | 285 | strlcpy((char *)dest, src, maxout); |
b86bc4cf | 286 | return ((int)strlen((char *)dest)); |
ef416fc2 | 287 | } |
288 | ||
411affcf | 289 | /* |
290 | * Handle ISO-8859-1 to UTF-8 directly... | |
291 | */ | |
292 | ||
293 | if (encoding == CUPS_ISO8859_1) | |
294 | { | |
295 | int ch; /* Character from string */ | |
296 | cups_utf8_t *destptr, /* Pointer into UTF-8 buffer */ | |
297 | *destend; /* End of UTF-8 buffer */ | |
298 | ||
299 | ||
300 | destptr = dest; | |
301 | destend = dest + maxout - 2; | |
302 | ||
303 | while (*src && destptr < destend) | |
304 | { | |
305 | ch = *src++ & 255; | |
306 | ||
307 | if (ch & 128) | |
308 | { | |
309 | *destptr++ = 0xc0 | (ch >> 6); | |
310 | *destptr++ = 0x80 | (ch & 0x3f); | |
311 | } | |
312 | else | |
313 | *destptr++ = ch; | |
314 | } | |
315 | ||
316 | *destptr = '\0'; | |
317 | ||
b86bc4cf | 318 | return ((int)(destptr - dest)); |
411affcf | 319 | } |
320 | ||
ef416fc2 | 321 | /* |
e1d6a774 | 322 | * Convert input legacy charset to UTF-8... |
ef416fc2 | 323 | */ |
e1d6a774 | 324 | |
d6ae789d | 325 | #ifdef HAVE_PTHREAD_H |
326 | pthread_mutex_lock(&map_mutex); | |
327 | #endif /* HAVE_PTHREAD_H */ | |
328 | ||
ef416fc2 | 329 | if (encoding < CUPS_ENCODING_SBCS_END) |
d6ae789d | 330 | bytes = conv_sbcs_to_utf8(dest, (cups_sbcs_t *)src, maxout, encoding); |
ef416fc2 | 331 | else |
91c84a35 | 332 | bytes = conv_vbcs_to_utf8(dest, (cups_sbcs_t *)src, maxout, encoding); |
d6ae789d | 333 | |
334 | #ifdef HAVE_PTHREAD_H | |
335 | pthread_mutex_unlock(&map_mutex); | |
336 | #endif /* HAVE_PTHREAD_H */ | |
337 | ||
338 | return (bytes); | |
ef416fc2 | 339 | } |
340 | ||
e1d6a774 | 341 | |
ef416fc2 | 342 | /* |
e1d6a774 | 343 | * 'cupsUTF8ToCharset()' - Convert UTF-8 to legacy character set. |
ef416fc2 | 344 | * |
345 | * This code handles single-byte (SBCS), double-byte (DBCS), and | |
346 | * variable-byte (VBCS) character sets _without_ charset escapes... | |
347 | * This code does not handle multiple-byte character sets (MBCS) | |
348 | * (such as ISO-2022-JP) with charset switching via escapes... | |
349 | */ | |
e1d6a774 | 350 | |
351 | int /* O - Count or -1 on error */ | |
352 | cupsUTF8ToCharset( | |
353 | char *dest, /* O - Target string */ | |
354 | const cups_utf8_t *src, /* I - Source string */ | |
355 | const int maxout, /* I - Max output */ | |
356 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 357 | { |
d6ae789d | 358 | int bytes; /* Number of bytes converted */ |
359 | ||
360 | ||
ef416fc2 | 361 | /* |
362 | * Check for valid arguments... | |
363 | */ | |
364 | ||
365 | if (!dest || !src || maxout < 1 || maxout > CUPS_MAX_USTRING) | |
e1d6a774 | 366 | { |
367 | if (dest) | |
368 | *dest = '\0'; | |
369 | ||
ef416fc2 | 370 | return (-1); |
e1d6a774 | 371 | } |
ef416fc2 | 372 | |
373 | /* | |
374 | * Handle identity conversions... | |
375 | */ | |
376 | ||
377 | if (encoding == CUPS_UTF8 || | |
378 | encoding < 0 || encoding >= CUPS_ENCODING_VBCS_END) | |
379 | { | |
e1d6a774 | 380 | strlcpy(dest, (char *)src, maxout); |
b86bc4cf | 381 | return ((int)strlen(dest)); |
ef416fc2 | 382 | } |
383 | ||
411affcf | 384 | /* |
385 | * Handle UTF-8 to ISO-8859-1 directly... | |
386 | */ | |
387 | ||
388 | if (encoding == CUPS_ISO8859_1) | |
389 | { | |
390 | int ch; /* Character from string */ | |
391 | char *destptr, /* Pointer into ISO-8859-1 buffer */ | |
392 | *destend; /* End of ISO-8859-1 buffer */ | |
393 | ||
394 | ||
395 | destptr = dest; | |
396 | destend = dest + maxout - 1; | |
397 | ||
398 | while (*src && destptr < destend) | |
399 | { | |
400 | ch = *src++; | |
401 | ||
402 | if ((ch & 0xe0) == 0xc0) | |
403 | { | |
404 | ch = ((ch & 0x1f) << 6) | (*src++ & 0x3f); | |
405 | ||
406 | if (ch < 256) | |
407 | *destptr++ = ch; | |
408 | else | |
409 | *destptr++ = '?'; | |
410 | } | |
411 | else if ((ch & 0xf0) == 0xe0 || | |
412 | (ch & 0xf8) == 0xf0) | |
413 | *destptr++ = '?'; | |
414 | else if (!(ch & 0x80)) | |
415 | *destptr++ = ch; | |
416 | } | |
417 | ||
418 | *destptr = '\0'; | |
419 | ||
b86bc4cf | 420 | return ((int)(destptr - dest)); |
411affcf | 421 | } |
422 | ||
ef416fc2 | 423 | /* |
e1d6a774 | 424 | * Convert input UTF-8 to legacy charset... |
ef416fc2 | 425 | */ |
e1d6a774 | 426 | |
d6ae789d | 427 | #ifdef HAVE_PTHREAD_H |
428 | pthread_mutex_lock(&map_mutex); | |
429 | #endif /* HAVE_PTHREAD_H */ | |
430 | ||
ef416fc2 | 431 | if (encoding < CUPS_ENCODING_SBCS_END) |
d6ae789d | 432 | bytes = conv_utf8_to_sbcs((cups_sbcs_t *)dest, src, maxout, encoding); |
ef416fc2 | 433 | else |
91c84a35 | 434 | bytes = conv_utf8_to_vbcs((cups_sbcs_t *)dest, src, maxout, encoding); |
d6ae789d | 435 | |
436 | #ifdef HAVE_PTHREAD_H | |
437 | pthread_mutex_unlock(&map_mutex); | |
438 | #endif /* HAVE_PTHREAD_H */ | |
439 | ||
440 | return (bytes); | |
ef416fc2 | 441 | } |
442 | ||
ef416fc2 | 443 | |
444 | /* | |
445 | * 'cupsUTF8ToUTF32()' - Convert UTF-8 to UTF-32. | |
446 | * | |
447 | * 32-bit UTF-32 (actually 21-bit) maps to UTF-8 as follows... | |
448 | * | |
449 | * UTF-32 char UTF-8 char(s) | |
450 | * -------------------------------------------------- | |
e1d6a774 | 451 | * 0 to 127 = 0xxxxxxx (US-ASCII) |
ef416fc2 | 452 | * 128 to 2047 = 110xxxxx 10yyyyyy |
453 | * 2048 to 65535 = 1110xxxx 10yyyyyy 10zzzzzz | |
e1d6a774 | 454 | * > 65535 = 11110xxx 10yyyyyy 10zzzzzz 10xxxxxx |
ef416fc2 | 455 | * |
456 | * UTF-32 prohibits chars beyond Plane 16 (> 0x10ffff) in UCS-4, | |
457 | * which would convert to five- or six-octet UTF-8 sequences... | |
ef416fc2 | 458 | */ |
e1d6a774 | 459 | |
460 | int /* O - Count or -1 on error */ | |
461 | cupsUTF8ToUTF32( | |
462 | cups_utf32_t *dest, /* O - Target string */ | |
463 | const cups_utf8_t *src, /* I - Source string */ | |
464 | const int maxout) /* I - Max output */ | |
ef416fc2 | 465 | { |
e1d6a774 | 466 | int i; /* Looping variable */ |
467 | cups_utf8_t ch; /* Character value */ | |
468 | cups_utf8_t next; /* Next character value */ | |
469 | cups_utf32_t ch32; /* UTF-32 character value */ | |
470 | ||
ef416fc2 | 471 | |
472 | /* | |
473 | * Check for valid arguments and clear output... | |
474 | */ | |
e1d6a774 | 475 | |
c9fc04c6 MS |
476 | DEBUG_printf(("cupsUTF8ToUTF32(dest=%p, src=\"%s\", maxout=%d)\n", dest, |
477 | src ? (const char *)src : "(null)", maxout)); | |
478 | ||
e1d6a774 | 479 | if (dest) |
480 | *dest = 0; | |
481 | ||
482 | if (!dest || !src || maxout < 1 || maxout > CUPS_MAX_USTRING) | |
c9fc04c6 MS |
483 | { |
484 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad arguments)"); | |
485 | ||
ef416fc2 | 486 | return (-1); |
c9fc04c6 | 487 | } |
ef416fc2 | 488 | |
489 | /* | |
cda47a96 | 490 | * Convert input UTF-8 to output UTF-32... |
ef416fc2 | 491 | */ |
e1d6a774 | 492 | |
e1d6a774 | 493 | for (i = maxout - 1; *src && i > 0; i --) |
ef416fc2 | 494 | { |
e1d6a774 | 495 | ch = *src++; |
ef416fc2 | 496 | |
497 | /* | |
498 | * Convert UTF-8 character(s) to UTF-32 character... | |
499 | */ | |
e1d6a774 | 500 | |
501 | if (!(ch & 0x80)) | |
ef416fc2 | 502 | { |
503 | /* | |
504 | * One-octet UTF-8 <= 127 (US-ASCII)... | |
505 | */ | |
e1d6a774 | 506 | |
507 | *dest++ = ch; | |
c9fc04c6 MS |
508 | |
509 | DEBUG_printf(("cupsUTF8ToUTF32: %02x => %08X\n", src[-1], ch)); | |
2abf387c | 510 | continue; |
ef416fc2 | 511 | } |
512 | else if ((ch & 0xe0) == 0xc0) | |
513 | { | |
514 | /* | |
515 | * Two-octet UTF-8 <= 2047 (Latin-x)... | |
516 | */ | |
e1d6a774 | 517 | |
518 | next = *src++; | |
c9fc04c6 MS |
519 | if ((next & 0xc0) != 0x80) |
520 | { | |
521 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); | |
522 | ||
ef416fc2 | 523 | return (-1); |
c9fc04c6 | 524 | } |
e1d6a774 | 525 | |
ef416fc2 | 526 | ch32 = ((ch & 0x1f) << 6) | (next & 0x3f); |
527 | ||
528 | /* | |
529 | * Check for non-shortest form (invalid UTF-8)... | |
530 | */ | |
e1d6a774 | 531 | |
532 | if (ch32 < 0x80) | |
c9fc04c6 MS |
533 | { |
534 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); | |
535 | ||
ef416fc2 | 536 | return (-1); |
c9fc04c6 | 537 | } |
e1d6a774 | 538 | |
539 | *dest++ = ch32; | |
c9fc04c6 MS |
540 | |
541 | DEBUG_printf(("cupsUTF8ToUTF32: %02x %02x => %08X\n", | |
542 | src[-2], src[-1], (unsigned)ch32)); | |
ef416fc2 | 543 | } |
544 | else if ((ch & 0xf0) == 0xe0) | |
545 | { | |
546 | /* | |
547 | * Three-octet UTF-8 <= 65535 (Plane 0 - BMP)... | |
548 | */ | |
e1d6a774 | 549 | |
550 | next = *src++; | |
c9fc04c6 MS |
551 | if ((next & 0xc0) != 0x80) |
552 | { | |
553 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); | |
554 | ||
ef416fc2 | 555 | return (-1); |
c9fc04c6 | 556 | } |
e1d6a774 | 557 | |
558 | ch32 = ((ch & 0x0f) << 6) | (next & 0x3f); | |
559 | ||
560 | next = *src++; | |
c9fc04c6 MS |
561 | if ((next & 0xc0) != 0x80) |
562 | { | |
563 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); | |
564 | ||
ef416fc2 | 565 | return (-1); |
c9fc04c6 | 566 | } |
e1d6a774 | 567 | |
568 | ch32 = (ch32 << 6) | (next & 0x3f); | |
ef416fc2 | 569 | |
570 | /* | |
571 | * Check for non-shortest form (invalid UTF-8)... | |
572 | */ | |
e1d6a774 | 573 | |
574 | if (ch32 < 0x800) | |
c9fc04c6 MS |
575 | { |
576 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); | |
577 | ||
ef416fc2 | 578 | return (-1); |
c9fc04c6 | 579 | } |
e1d6a774 | 580 | |
581 | *dest++ = ch32; | |
c9fc04c6 MS |
582 | |
583 | DEBUG_printf(("cupsUTF8ToUTF32: %02x %02x %02x => %08X\n", | |
584 | src[-3], src[-2], src[-1], (unsigned)ch32)); | |
ef416fc2 | 585 | } |
586 | else if ((ch & 0xf8) == 0xf0) | |
587 | { | |
588 | /* | |
e1d6a774 | 589 | * Four-octet UTF-8... |
ef416fc2 | 590 | */ |
e1d6a774 | 591 | |
592 | next = *src++; | |
c9fc04c6 MS |
593 | if ((next & 0xc0) != 0x80) |
594 | { | |
595 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); | |
596 | ||
ef416fc2 | 597 | return (-1); |
c9fc04c6 | 598 | } |
e1d6a774 | 599 | |
600 | ch32 = ((ch & 0x07) << 6) | (next & 0x3f); | |
601 | ||
602 | next = *src++; | |
c9fc04c6 MS |
603 | if ((next & 0xc0) != 0x80) |
604 | { | |
605 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); | |
606 | ||
e1d6a774 | 607 | return (-1); |
c9fc04c6 | 608 | } |
e1d6a774 | 609 | |
610 | ch32 = (ch32 << 6) | (next & 0x3f); | |
611 | ||
612 | next = *src++; | |
c9fc04c6 MS |
613 | if ((next & 0xc0) != 0x80) |
614 | { | |
615 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); | |
616 | ||
e1d6a774 | 617 | return (-1); |
c9fc04c6 | 618 | } |
e1d6a774 | 619 | |
620 | ch32 = (ch32 << 6) | (next & 0x3f); | |
621 | ||
ef416fc2 | 622 | /* |
e1d6a774 | 623 | * Check for non-shortest form (invalid UTF-8)... |
ef416fc2 | 624 | */ |
e1d6a774 | 625 | |
626 | if (ch32 < 0x10000) | |
c9fc04c6 MS |
627 | { |
628 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); | |
629 | ||
e1d6a774 | 630 | return (-1); |
c9fc04c6 | 631 | } |
e1d6a774 | 632 | |
633 | *dest++ = ch32; | |
c9fc04c6 MS |
634 | |
635 | DEBUG_printf(("cupsUTF8ToUTF32: %02x %02x %02x %02x => %08X\n", | |
636 | src[-4], src[-3], src[-2], src[-1], (unsigned)ch32)); | |
ef416fc2 | 637 | } |
638 | else | |
639 | { | |
640 | /* | |
e1d6a774 | 641 | * More than 4-octet (invalid UTF-8 sequence)... |
ef416fc2 | 642 | */ |
e1d6a774 | 643 | |
c9fc04c6 MS |
644 | DEBUG_puts("cupsUTF8ToUTF32: Returning -1 (bad UTF-8 sequence)"); |
645 | ||
ef416fc2 | 646 | return (-1); |
647 | } | |
648 | ||
649 | /* | |
650 | * Check for UTF-16 surrogate (illegal UTF-8)... | |
651 | */ | |
ef416fc2 | 652 | |
2abf387c | 653 | if (ch32 >= 0xd800 && ch32 <= 0xdfff) |
ef416fc2 | 654 | return (-1); |
655 | } | |
e1d6a774 | 656 | |
ef416fc2 | 657 | *dest = 0; |
e1d6a774 | 658 | |
c9fc04c6 MS |
659 | DEBUG_printf(("cupsUTF8ToUTF32: Returning %d characters\n", maxout - 1 - i)); |
660 | ||
661 | return (maxout - 1 - i); | |
ef416fc2 | 662 | } |
663 | ||
e1d6a774 | 664 | |
ef416fc2 | 665 | /* |
666 | * 'cupsUTF32ToUTF8()' - Convert UTF-32 to UTF-8. | |
667 | * | |
668 | * 32-bit UTF-32 (actually 21-bit) maps to UTF-8 as follows... | |
669 | * | |
670 | * UTF-32 char UTF-8 char(s) | |
671 | * -------------------------------------------------- | |
e1d6a774 | 672 | * 0 to 127 = 0xxxxxxx (US-ASCII) |
ef416fc2 | 673 | * 128 to 2047 = 110xxxxx 10yyyyyy |
674 | * 2048 to 65535 = 1110xxxx 10yyyyyy 10zzzzzz | |
e1d6a774 | 675 | * > 65535 = 11110xxx 10yyyyyy 10zzzzzz 10xxxxxx |
ef416fc2 | 676 | * |
677 | * UTF-32 prohibits chars beyond Plane 16 (> 0x10ffff) in UCS-4, | |
678 | * which would convert to five- or six-octet UTF-8 sequences... | |
ef416fc2 | 679 | */ |
e1d6a774 | 680 | |
681 | int /* O - Count or -1 on error */ | |
682 | cupsUTF32ToUTF8( | |
683 | cups_utf8_t *dest, /* O - Target string */ | |
684 | const cups_utf32_t *src, /* I - Source string */ | |
685 | const int maxout) /* I - Max output */ | |
ef416fc2 | 686 | { |
e1d6a774 | 687 | cups_utf8_t *start; /* Start of destination string */ |
688 | int i; /* Looping variable */ | |
689 | int swap; /* Byte-swap input to output */ | |
690 | cups_utf32_t ch; /* Character value */ | |
691 | ||
ef416fc2 | 692 | |
693 | /* | |
694 | * Check for valid arguments and clear output... | |
695 | */ | |
e1d6a774 | 696 | |
c9fc04c6 MS |
697 | DEBUG_printf(("cupsUTF32ToUTF8(dest=%p, src=%p, maxout=%d)\n", dest, src, |
698 | maxout)); | |
699 | ||
e1d6a774 | 700 | if (dest) |
701 | *dest = '\0'; | |
702 | ||
703 | if (!dest || !src || maxout < 1) | |
c9fc04c6 MS |
704 | { |
705 | DEBUG_puts("cupsUTF32ToUTF8: Returning -1 (bad args)"); | |
706 | ||
ef416fc2 | 707 | return (-1); |
c9fc04c6 | 708 | } |
ef416fc2 | 709 | |
710 | /* | |
711 | * Check for leading BOM in UTF-32 and inverted BOM... | |
712 | */ | |
e1d6a774 | 713 | |
714 | start = dest; | |
715 | swap = *src == 0xfffe0000; | |
716 | ||
c9fc04c6 MS |
717 | DEBUG_printf(("cupsUTF32ToUTF8: swap=%d\n", swap)); |
718 | ||
e1d6a774 | 719 | if (*src == 0xfffe0000 || *src == 0xfeff) |
720 | src ++; | |
ef416fc2 | 721 | |
722 | /* | |
723 | * Convert input UTF-32 to output UTF-8... | |
724 | */ | |
e1d6a774 | 725 | |
726 | for (i = maxout - 1; *src && i > 0;) | |
ef416fc2 | 727 | { |
e1d6a774 | 728 | ch = *src++; |
ef416fc2 | 729 | |
730 | /* | |
731 | * Byte swap input UTF-32, if necessary... | |
e1d6a774 | 732 | * (only byte-swapping 24 of 32 bits) |
ef416fc2 | 733 | */ |
e1d6a774 | 734 | |
ef416fc2 | 735 | if (swap) |
736 | ch = ((ch >> 24) | ((ch >> 8) & 0xff00) | ((ch << 8) & 0xff0000)); | |
737 | ||
738 | /* | |
e1d6a774 | 739 | * Check for beyond Plane 16 (invalid UTF-32)... |
ef416fc2 | 740 | */ |
ef416fc2 | 741 | |
ef416fc2 | 742 | if (ch > 0x10ffff) |
c9fc04c6 MS |
743 | { |
744 | DEBUG_puts("cupsUTF32ToUTF8: Returning -1 (character out of range)"); | |
745 | ||
ef416fc2 | 746 | return (-1); |
c9fc04c6 | 747 | } |
ef416fc2 | 748 | |
ef416fc2 | 749 | /* |
750 | * Convert UTF-32 character to UTF-8 character(s)... | |
751 | */ | |
e1d6a774 | 752 | |
753 | if (ch < 0x80) | |
ef416fc2 | 754 | { |
755 | /* | |
756 | * One-octet UTF-8 <= 127 (US-ASCII)... | |
757 | */ | |
e1d6a774 | 758 | |
759 | *dest++ = (cups_utf8_t)ch; | |
760 | i --; | |
c9fc04c6 MS |
761 | |
762 | DEBUG_printf(("cupsUTF32ToUTF8: %08x => %02x\n", (unsigned)ch, dest[-1])); | |
ef416fc2 | 763 | } |
e1d6a774 | 764 | else if (ch < 0x800) |
ef416fc2 | 765 | { |
766 | /* | |
767 | * Two-octet UTF-8 <= 2047 (Latin-x)... | |
768 | */ | |
e1d6a774 | 769 | |
770 | if (i < 2) | |
c9fc04c6 MS |
771 | { |
772 | DEBUG_puts("cupsUTF32ToUTF8: Returning -1 (too long 2)"); | |
773 | ||
e1d6a774 | 774 | return (-1); |
c9fc04c6 | 775 | } |
e1d6a774 | 776 | |
777 | *dest++ = (cups_utf8_t)(0xc0 | ((ch >> 6) & 0x1f)); | |
778 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
779 | i -= 2; | |
c9fc04c6 MS |
780 | |
781 | DEBUG_printf(("cupsUTF32ToUTF8: %08x => %02x %02x\n", (unsigned)ch, | |
782 | dest[-2], dest[-1])); | |
ef416fc2 | 783 | } |
e1d6a774 | 784 | else if (ch < 0x10000) |
ef416fc2 | 785 | { |
786 | /* | |
787 | * Three-octet UTF-8 <= 65535 (Plane 0 - BMP)... | |
788 | */ | |
e1d6a774 | 789 | |
790 | if (i < 3) | |
c9fc04c6 MS |
791 | { |
792 | DEBUG_puts("cupsUTF32ToUTF8: Returning -1 (too long 3)"); | |
793 | ||
e1d6a774 | 794 | return (-1); |
c9fc04c6 | 795 | } |
e1d6a774 | 796 | |
797 | *dest++ = (cups_utf8_t)(0xe0 | ((ch >> 12) & 0x0f)); | |
798 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 6) & 0x3f)); | |
799 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
800 | i -= 3; | |
c9fc04c6 MS |
801 | |
802 | DEBUG_printf(("cupsUTF32ToUTF8: %08x => %02x %02x %02x\n", (unsigned)ch, | |
803 | dest[-3], dest[-2], dest[-1])); | |
e1d6a774 | 804 | } |
805 | else | |
806 | { | |
807 | /* | |
808 | * Four-octet UTF-8... | |
809 | */ | |
810 | ||
811 | if (i < 4) | |
812 | return (-1); | |
813 | ||
814 | *dest++ = (cups_utf8_t)(0xf0 | ((ch >> 18) & 0x07)); | |
815 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 12) & 0x3f)); | |
816 | *dest++ = (cups_utf8_t)(0x80 | ((ch >> 6) & 0x3f)); | |
817 | *dest++ = (cups_utf8_t)(0x80 | (ch & 0x3f)); | |
818 | i -= 4; | |
c9fc04c6 MS |
819 | |
820 | DEBUG_printf(("cupsUTF32ToUTF8: %08x => %02x %02x %02x %02x\n", | |
821 | (unsigned)ch, dest[-4], dest[-3], dest[-2], dest[-1])); | |
ef416fc2 | 822 | } |
823 | } | |
e1d6a774 | 824 | |
ef416fc2 | 825 | *dest = '\0'; |
e1d6a774 | 826 | |
c9fc04c6 MS |
827 | DEBUG_printf(("cupsUTF32ToUTF8: Returning %d\n", (int)(dest - start))); |
828 | ||
e1d6a774 | 829 | return ((int)(dest - start)); |
ef416fc2 | 830 | } |
831 | ||
e1d6a774 | 832 | |
ef416fc2 | 833 | /* |
e1d6a774 | 834 | * 'compare_wide()' - Compare key for wide (VBCS) match. |
835 | */ | |
836 | ||
837 | static int | |
838 | compare_wide(const void *k1, /* I - Key char */ | |
839 | const void *k2) /* I - Map char */ | |
840 | { | |
841 | cups_vbcs_t key; /* Legacy key character */ | |
842 | cups_vbcs_t map; /* Legacy map character */ | |
843 | ||
844 | ||
845 | key = *((cups_vbcs_t *)k1); | |
846 | map = ((_cups_wide2uni_t *)k2)->widechar; | |
847 | ||
848 | return ((int)(key - map)); | |
849 | } | |
850 | ||
851 | ||
852 | /* | |
853 | * 'conv_sbcs_to_utf8()' - Convert legacy SBCS to UTF-8. | |
ef416fc2 | 854 | */ |
e1d6a774 | 855 | |
856 | static int /* O - Count or -1 on error */ | |
857 | conv_sbcs_to_utf8( | |
858 | cups_utf8_t *dest, /* O - Target string */ | |
859 | const cups_sbcs_t *src, /* I - Source string */ | |
860 | int maxout, /* I - Max output */ | |
861 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 862 | { |
e1d6a774 | 863 | _cups_cmap_t *cmap; /* Legacy SBCS / Unicode Charset Map */ |
864 | cups_ucs2_t *crow; /* Pointer to UCS-2 row in 'char2uni' */ | |
865 | cups_sbcs_t legchar; /* Legacy character value */ | |
866 | cups_utf32_t work[CUPS_MAX_USTRING], /* Internal UCS-4 string */ | |
867 | *workptr; /* Pointer into string */ | |
868 | ||
ef416fc2 | 869 | |
870 | /* | |
e1d6a774 | 871 | * Find legacy charset map in cache... |
ef416fc2 | 872 | */ |
e1d6a774 | 873 | |
d6ae789d | 874 | if ((cmap = (_cups_cmap_t *)get_charmap(encoding)) == NULL) |
ef416fc2 | 875 | return (-1); |
ef416fc2 | 876 | |
877 | /* | |
e1d6a774 | 878 | * Convert input legacy charset to internal UCS-4 (and insert BOM)... |
ef416fc2 | 879 | */ |
ef416fc2 | 880 | |
e1d6a774 | 881 | work[0] = 0xfeff; |
882 | for (workptr = work + 1; *src && workptr < (work + CUPS_MAX_USTRING - 1);) | |
ef416fc2 | 883 | { |
e1d6a774 | 884 | legchar = *src++; |
ef416fc2 | 885 | |
886 | /* | |
e1d6a774 | 887 | * Convert ASCII verbatim (optimization)... |
ef416fc2 | 888 | */ |
ef416fc2 | 889 | |
e1d6a774 | 890 | if (legchar < 0x80) |
891 | *workptr++ = (cups_utf32_t)legchar; | |
892 | else | |
ef416fc2 | 893 | { |
e1d6a774 | 894 | /* |
895 | * Convert unknown character to Replacement Character... | |
896 | */ | |
ef416fc2 | 897 | |
e1d6a774 | 898 | crow = cmap->char2uni + legchar; |
899 | ||
900 | if (!*crow) | |
901 | *workptr++ = 0xfffd; | |
902 | else | |
903 | *workptr++ = (cups_utf32_t)*crow; | |
ef416fc2 | 904 | } |
ef416fc2 | 905 | } |
e1d6a774 | 906 | |
907 | *workptr = 0; | |
908 | ||
909 | /* | |
910 | * Convert internal UCS-4 to output UTF-8 (and delete BOM)... | |
911 | */ | |
912 | ||
d6ae789d | 913 | cmap->used --; |
e1d6a774 | 914 | |
915 | return (cupsUTF32ToUTF8(dest, work, maxout)); | |
ef416fc2 | 916 | } |
917 | ||
e1d6a774 | 918 | |
ef416fc2 | 919 | /* |
e1d6a774 | 920 | * 'conv_utf8_to_sbcs()' - Convert UTF-8 to legacy SBCS. |
ef416fc2 | 921 | */ |
e1d6a774 | 922 | |
923 | static int /* O - Count or -1 on error */ | |
924 | conv_utf8_to_sbcs( | |
925 | cups_sbcs_t *dest, /* O - Target string */ | |
926 | const cups_utf8_t *src, /* I - Source string */ | |
927 | int maxout, /* I - Max output */ | |
928 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 929 | { |
e1d6a774 | 930 | cups_sbcs_t *start; /* Start of destination string */ |
931 | _cups_cmap_t *cmap; /* Legacy SBCS / Unicode Charset Map */ | |
932 | cups_sbcs_t *srow; /* Pointer to SBCS row in 'uni2char' */ | |
933 | cups_utf32_t unichar; /* Character value */ | |
934 | cups_utf32_t work[CUPS_MAX_USTRING], /* Internal UCS-4 string */ | |
935 | *workptr; /* Pointer into string */ | |
936 | ||
ef416fc2 | 937 | |
938 | /* | |
e1d6a774 | 939 | * Find legacy charset map in cache... |
ef416fc2 | 940 | */ |
e1d6a774 | 941 | |
d6ae789d | 942 | if ((cmap = (_cups_cmap_t *)get_charmap(encoding)) == NULL) |
ef416fc2 | 943 | return (-1); |
ef416fc2 | 944 | |
945 | /* | |
e1d6a774 | 946 | * Convert input UTF-8 to internal UCS-4 (and insert BOM)... |
ef416fc2 | 947 | */ |
e1d6a774 | 948 | |
949 | if (cupsUTF8ToUTF32(work, src, CUPS_MAX_USTRING) < 0) | |
950 | return (-1); | |
ef416fc2 | 951 | |
952 | /* | |
e1d6a774 | 953 | * Convert internal UCS-4 to SBCS legacy charset (and delete BOM)... |
ef416fc2 | 954 | */ |
e1d6a774 | 955 | |
58dc1933 | 956 | for (workptr = work, start = dest; *workptr && maxout > 0; maxout --) |
ef416fc2 | 957 | { |
e1d6a774 | 958 | unichar = *workptr++; |
959 | if (!unichar) | |
ef416fc2 | 960 | break; |
ef416fc2 | 961 | |
962 | /* | |
e1d6a774 | 963 | * Convert ASCII verbatim (optimization)... |
ef416fc2 | 964 | */ |
ef416fc2 | 965 | |
e1d6a774 | 966 | if (unichar < 0x80) |
967 | { | |
968 | *dest++ = (cups_sbcs_t)unichar; | |
969 | continue; | |
970 | } | |
ef416fc2 | 971 | |
972 | /* | |
e1d6a774 | 973 | * Convert unknown character to visible replacement... |
ef416fc2 | 974 | */ |
ef416fc2 | 975 | |
e1d6a774 | 976 | srow = cmap->uni2char[(int)((unichar >> 8) & 0xff)]; |
ef416fc2 | 977 | |
e1d6a774 | 978 | if (srow) |
979 | srow += (int)(unichar & 0xff); | |
ef416fc2 | 980 | |
e1d6a774 | 981 | if (!srow || !*srow) |
982 | *dest++ = '?'; | |
983 | else | |
984 | *dest++ = *srow; | |
ef416fc2 | 985 | } |
ef416fc2 | 986 | |
e1d6a774 | 987 | *dest = '\0'; |
988 | ||
d6ae789d | 989 | cmap->used --; |
e1d6a774 | 990 | |
991 | return ((int)(dest - start)); | |
ef416fc2 | 992 | } |
993 | ||
e1d6a774 | 994 | |
ef416fc2 | 995 | /* |
e1d6a774 | 996 | * 'conv_utf8_to_vbcs()' - Convert UTF-8 to legacy DBCS/VBCS. |
ef416fc2 | 997 | */ |
e1d6a774 | 998 | |
999 | static int /* O - Count or -1 on error */ | |
1000 | conv_utf8_to_vbcs( | |
1001 | cups_sbcs_t *dest, /* O - Target string */ | |
1002 | const cups_utf8_t *src, /* I - Source string */ | |
1003 | int maxout, /* I - Max output */ | |
1004 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 1005 | { |
e1d6a774 | 1006 | cups_sbcs_t *start; /* Start of destination string */ |
1007 | _cups_vmap_t *vmap; /* Legacy DBCS / Unicode Charset Map */ | |
1008 | cups_vbcs_t *vrow; /* Pointer to VBCS row in 'uni2char' */ | |
1009 | cups_utf32_t unichar; /* Character value */ | |
1010 | cups_vbcs_t legchar; /* Legacy character value */ | |
1011 | cups_utf32_t work[CUPS_MAX_USTRING], /* Internal UCS-4 string */ | |
1012 | *workptr; /* Pointer into string */ | |
ef416fc2 | 1013 | |
ef416fc2 | 1014 | |
c9fc04c6 MS |
1015 | DEBUG_printf(("conv_utf8_to_vbcs(dest=%p, src=\"%s\", maxout=%d, " |
1016 | "encoding=%d)\n", dest, src, maxout, encoding)); | |
1017 | ||
ef416fc2 | 1018 | /* |
e1d6a774 | 1019 | * Find legacy charset map in cache... |
ef416fc2 | 1020 | */ |
ef416fc2 | 1021 | |
d6ae789d | 1022 | if ((vmap = (_cups_vmap_t *)get_charmap(encoding)) == NULL) |
c9fc04c6 MS |
1023 | { |
1024 | DEBUG_puts("conv_utf8_to_vbcs: Returning -1 (no charmap)"); | |
1025 | ||
e1d6a774 | 1026 | return (-1); |
c9fc04c6 | 1027 | } |
ef416fc2 | 1028 | |
1029 | /* | |
e1d6a774 | 1030 | * Convert input UTF-8 to internal UCS-4 (and insert BOM)... |
ef416fc2 | 1031 | */ |
e1d6a774 | 1032 | |
1033 | if (cupsUTF8ToUTF32(work, src, CUPS_MAX_USTRING) < 0) | |
c9fc04c6 MS |
1034 | { |
1035 | DEBUG_puts("conv_utf8_to_vbcs: Returning -1 (Unable to convert to UTF-32)"); | |
1036 | ||
e1d6a774 | 1037 | return (-1); |
c9fc04c6 | 1038 | } |
ef416fc2 | 1039 | |
1040 | /* | |
e1d6a774 | 1041 | * Convert internal UCS-4 to VBCS legacy charset (and delete BOM)... |
ef416fc2 | 1042 | */ |
e1d6a774 | 1043 | |
58dc1933 | 1044 | for (start = dest, workptr = work; *workptr && maxout > 0; maxout --) |
ef416fc2 | 1045 | { |
e1d6a774 | 1046 | unichar = *workptr++; |
ef416fc2 | 1047 | |
1048 | /* | |
e1d6a774 | 1049 | * Convert ASCII verbatim (optimization)... |
ef416fc2 | 1050 | */ |
e1d6a774 | 1051 | |
1052 | if (unichar < 0x80) | |
1053 | { | |
b86bc4cf | 1054 | *dest++ = (cups_sbcs_t)unichar; |
c9fc04c6 MS |
1055 | |
1056 | DEBUG_printf(("conv_utf8_to_vbcs: %08x => %02X\n", (unsigned)unichar, | |
1057 | dest[-1])); | |
1058 | ||
e1d6a774 | 1059 | continue; |
1060 | } | |
ef416fc2 | 1061 | |
1062 | /* | |
e1d6a774 | 1063 | * Convert unknown character to visible replacement... |
ef416fc2 | 1064 | */ |
e1d6a774 | 1065 | |
1066 | vrow = vmap->uni2char[(int)((unichar >> 8) & 0xff)]; | |
1067 | ||
1068 | if (vrow) | |
1069 | vrow += (int)(unichar & 0xff); | |
1070 | ||
1071 | if (!vrow || !*vrow) | |
1072 | legchar = (cups_vbcs_t)'?'; | |
1073 | else | |
1074 | legchar = (cups_vbcs_t)*vrow; | |
ef416fc2 | 1075 | |
1076 | /* | |
e1d6a774 | 1077 | * Save n-byte legacy character... |
ef416fc2 | 1078 | */ |
e1d6a774 | 1079 | |
1080 | if (legchar > 0xffffff) | |
ef416fc2 | 1081 | { |
e1d6a774 | 1082 | if (maxout < 5) |
c9fc04c6 MS |
1083 | { |
1084 | DEBUG_puts("conv_utf8_to_vbcs: Returning -1 (out of space)"); | |
1085 | ||
e1d6a774 | 1086 | return (-1); |
c9fc04c6 | 1087 | } |
e1d6a774 | 1088 | |
1089 | *dest++ = (cups_sbcs_t)(legchar >> 24); | |
1090 | *dest++ = (cups_sbcs_t)(legchar >> 16); | |
1091 | *dest++ = (cups_sbcs_t)(legchar >> 8); | |
1092 | *dest++ = (cups_sbcs_t)legchar; | |
1093 | ||
1094 | maxout -= 3; | |
c9fc04c6 MS |
1095 | |
1096 | DEBUG_printf(("conv_utf8_to_vbcs: %08x => %02X %02X %02X %02X\n", | |
1097 | (unsigned)unichar, dest[-4], dest[-3], dest[-2], dest[-1])); | |
ef416fc2 | 1098 | } |
e1d6a774 | 1099 | else if (legchar > 0xffff) |
1100 | { | |
1101 | if (maxout < 4) | |
c9fc04c6 MS |
1102 | { |
1103 | DEBUG_puts("conv_utf8_to_vbcs: Returning -1 (out of space)"); | |
1104 | ||
e1d6a774 | 1105 | return (-1); |
c9fc04c6 | 1106 | } |
ef416fc2 | 1107 | |
e1d6a774 | 1108 | *dest++ = (cups_sbcs_t)(legchar >> 16); |
1109 | *dest++ = (cups_sbcs_t)(legchar >> 8); | |
1110 | *dest++ = (cups_sbcs_t)legchar; | |
ef416fc2 | 1111 | |
e1d6a774 | 1112 | maxout -= 2; |
c9fc04c6 MS |
1113 | |
1114 | DEBUG_printf(("conv_utf8_to_vbcs: %08x => %02X %02X %02X\n", | |
1115 | (unsigned)unichar, dest[-3], dest[-2], dest[-1])); | |
e1d6a774 | 1116 | } |
1117 | else if (legchar > 0xff) | |
1118 | { | |
1119 | *dest++ = (cups_sbcs_t)(legchar >> 8); | |
1120 | *dest++ = (cups_sbcs_t)legchar; | |
1121 | ||
1122 | maxout --; | |
c9fc04c6 MS |
1123 | |
1124 | DEBUG_printf(("conv_utf8_to_vbcs: %08x => %02X %02X\n", | |
1125 | (unsigned)unichar, dest[-2], dest[-1])); | |
1126 | } | |
1127 | else | |
1128 | { | |
1129 | *dest++ = legchar; | |
1130 | ||
1131 | DEBUG_printf(("conv_utf8_to_vbcs: %08x => %02X\n", | |
1132 | (unsigned)unichar, dest[-1])); | |
e1d6a774 | 1133 | } |
ef416fc2 | 1134 | } |
e1d6a774 | 1135 | |
1136 | *dest = '\0'; | |
1137 | ||
d6ae789d | 1138 | vmap->used --; |
e1d6a774 | 1139 | |
c9fc04c6 MS |
1140 | DEBUG_printf(("conv_utf8_to_vbcs: Returning %d characters\n", |
1141 | (int)(dest - start))); | |
1142 | ||
e1d6a774 | 1143 | return ((int)(dest - start)); |
ef416fc2 | 1144 | } |
1145 | ||
e1d6a774 | 1146 | |
ef416fc2 | 1147 | /* |
e1d6a774 | 1148 | * 'conv_vbcs_to_utf8()' - Convert legacy DBCS/VBCS to UTF-8. |
ef416fc2 | 1149 | */ |
e1d6a774 | 1150 | |
1151 | static int /* O - Count or -1 on error */ | |
1152 | conv_vbcs_to_utf8( | |
1153 | cups_utf8_t *dest, /* O - Target string */ | |
1154 | const cups_sbcs_t *src, /* I - Source string */ | |
1155 | int maxout, /* I - Max output */ | |
1156 | const cups_encoding_t encoding) /* I - Encoding */ | |
ef416fc2 | 1157 | { |
e1d6a774 | 1158 | _cups_vmap_t *vmap; /* Legacy VBCS / Unicode Charset Map */ |
1159 | cups_ucs2_t *crow; /* Pointer to UCS-2 row in 'char2uni' */ | |
1160 | _cups_wide2uni_t *wide2uni; /* Pointer to row in 'wide2uni' */ | |
1161 | cups_sbcs_t leadchar; /* Lead char of n-byte legacy char */ | |
1162 | cups_vbcs_t legchar; /* Legacy character value */ | |
1163 | cups_utf32_t work[CUPS_MAX_USTRING], /* Internal UCS-4 string */ | |
1164 | *workptr; /* Pointer into string */ | |
ef416fc2 | 1165 | |
ef416fc2 | 1166 | |
1167 | /* | |
e1d6a774 | 1168 | * Find legacy charset map in cache... |
ef416fc2 | 1169 | */ |
ef416fc2 | 1170 | |
c9fc04c6 MS |
1171 | DEBUG_printf(("conv_vbcs_to_utf8(dest=%p, src=%p, maxout=%d, encoding=%d)\n", |
1172 | dest, src, maxout, encoding)); | |
1173 | ||
d6ae789d | 1174 | if ((vmap = (_cups_vmap_t *)get_charmap(encoding)) == NULL) |
c9fc04c6 MS |
1175 | { |
1176 | DEBUG_puts("conv_vbcs_to_utf8: Returning -1 (NULL vmap)"); | |
1177 | ||
e1d6a774 | 1178 | return (-1); |
c9fc04c6 | 1179 | } |
ef416fc2 | 1180 | |
1181 | /* | |
e1d6a774 | 1182 | * Convert input legacy charset to internal UCS-4 (and insert BOM)... |
ef416fc2 | 1183 | */ |
ef416fc2 | 1184 | |
e1d6a774 | 1185 | work[0] = 0xfeff; |
1186 | for (workptr = work + 1; *src && workptr < (work + CUPS_MAX_USTRING - 1);) | |
ef416fc2 | 1187 | { |
e1d6a774 | 1188 | legchar = *src++; |
1189 | leadchar = (cups_sbcs_t)legchar; | |
ef416fc2 | 1190 | |
1191 | /* | |
e1d6a774 | 1192 | * Convert ASCII verbatim (optimization)... |
ef416fc2 | 1193 | */ |
ef416fc2 | 1194 | |
e1d6a774 | 1195 | if (legchar < 0x80) |
ef416fc2 | 1196 | { |
e1d6a774 | 1197 | *workptr++ = (cups_utf32_t)legchar; |
c9fc04c6 MS |
1198 | |
1199 | DEBUG_printf(("conv_vbcs_to_utf8: %02X => %08X\n", src[-1], | |
1200 | (unsigned)legchar)); | |
e1d6a774 | 1201 | continue; |
ef416fc2 | 1202 | } |
1203 | ||
1204 | /* | |
e1d6a774 | 1205 | * Convert 2-byte legacy character... |
ef416fc2 | 1206 | */ |
e1d6a774 | 1207 | |
1208 | if (vmap->lead2char[(int)leadchar] == leadchar) | |
ef416fc2 | 1209 | { |
e1d6a774 | 1210 | if (!*src) |
c9fc04c6 MS |
1211 | { |
1212 | DEBUG_puts("conv_vbcs_to_utf8: Returning -1 (short string)"); | |
1213 | ||
e1d6a774 | 1214 | return (-1); |
c9fc04c6 | 1215 | } |
e1d6a774 | 1216 | |
1217 | legchar = (legchar << 8) | *src++; | |
1218 | ||
ef416fc2 | 1219 | /* |
e1d6a774 | 1220 | * Convert unknown character to Replacement Character... |
ef416fc2 | 1221 | */ |
e1d6a774 | 1222 | |
1223 | crow = vmap->char2uni[(int)((legchar >> 8) & 0xff)]; | |
1224 | if (crow) | |
1225 | crow += (int) (legchar & 0xff); | |
1226 | ||
1227 | if (!crow || !*crow) | |
1228 | *workptr++ = 0xfffd; | |
1229 | else | |
1230 | *workptr++ = (cups_utf32_t)*crow; | |
c9fc04c6 MS |
1231 | |
1232 | DEBUG_printf(("conv_vbcs_to_utf8: %02X %02X => %08X\n", | |
1233 | src[-2], src[-1], (unsigned)workptr[-1])); | |
e1d6a774 | 1234 | continue; |
ef416fc2 | 1235 | } |
1236 | ||
1237 | /* | |
e1d6a774 | 1238 | * Fetch 3-byte or 4-byte legacy character... |
ef416fc2 | 1239 | */ |
e1d6a774 | 1240 | |
1241 | if (vmap->lead3char[(int)leadchar] == leadchar) | |
ef416fc2 | 1242 | { |
e1d6a774 | 1243 | if (!*src || !src[1]) |
c9fc04c6 MS |
1244 | { |
1245 | DEBUG_puts("conv_vbcs_to_utf8: Returning -1 (short string 2)"); | |
1246 | ||
e1d6a774 | 1247 | return (-1); |
c9fc04c6 | 1248 | } |
e1d6a774 | 1249 | |
1250 | legchar = (legchar << 8) | *src++; | |
1251 | legchar = (legchar << 8) | *src++; | |
ef416fc2 | 1252 | } |
e1d6a774 | 1253 | else if (vmap->lead4char[(int)leadchar] == leadchar) |
1254 | { | |
1255 | if (!*src || !src[1] || !src[2]) | |
c9fc04c6 MS |
1256 | { |
1257 | DEBUG_puts("conv_vbcs_to_utf8: Returning -1 (short string 3)"); | |
1258 | ||
e1d6a774 | 1259 | return (-1); |
c9fc04c6 | 1260 | } |
e1d6a774 | 1261 | |
1262 | legchar = (legchar << 8) | *src++; | |
1263 | legchar = (legchar << 8) | *src++; | |
1264 | legchar = (legchar << 8) | *src++; | |
1265 | } | |
1266 | else | |
c9fc04c6 MS |
1267 | { |
1268 | DEBUG_puts("conv_vbcs_to_utf8: Returning -1 (bad character)"); | |
1269 | ||
e1d6a774 | 1270 | return (-1); |
c9fc04c6 | 1271 | } |
ef416fc2 | 1272 | |
1273 | /* | |
e1d6a774 | 1274 | * Find 3-byte or 4-byte legacy character... |
ef416fc2 | 1275 | */ |
e1d6a774 | 1276 | |
1277 | wide2uni = (_cups_wide2uni_t *)bsearch(&legchar, | |
1278 | vmap->wide2uni, | |
1279 | vmap->widecount, | |
1280 | sizeof(_cups_wide2uni_t), | |
1281 | compare_wide); | |
ef416fc2 | 1282 | |
1283 | /* | |
e1d6a774 | 1284 | * Convert unknown character to Replacement Character... |
ef416fc2 | 1285 | */ |
e1d6a774 | 1286 | |
1287 | if (!wide2uni || !wide2uni->unichar) | |
1288 | *workptr++ = 0xfffd; | |
1289 | else | |
1290 | *workptr++ = wide2uni->unichar; | |
c9fc04c6 MS |
1291 | |
1292 | if (vmap->lead3char[(int)leadchar] == leadchar) | |
1293 | DEBUG_printf(("conv_vbcs_to_utf8: %02X %02X %02X => %08X\n", | |
1294 | src[-3], src[-2], src[-1], (unsigned)workptr[-1])); | |
1295 | else | |
1296 | DEBUG_printf(("conv_vbcs_to_utf8: %02X %02X %02X %02X => %08X\n", | |
1297 | src[-4], src[-3], src[-2], src[-1], (unsigned)workptr[-1])); | |
ef416fc2 | 1298 | } |
e1d6a774 | 1299 | |
1300 | *workptr = 0; | |
1301 | ||
d6ae789d | 1302 | vmap->used --; |
e1d6a774 | 1303 | |
c9fc04c6 MS |
1304 | DEBUG_printf(("conv_vbcs_to_utf8: Converting %d UTF-32 characters to UTF-8\n", |
1305 | (int)(workptr - work))); | |
1306 | ||
e1d6a774 | 1307 | /* |
1308 | * Convert internal UCS-4 to output UTF-8 (and delete BOM)... | |
1309 | */ | |
1310 | ||
1311 | return (cupsUTF32ToUTF8(dest, work, maxout)); | |
ef416fc2 | 1312 | } |
1313 | ||
e1d6a774 | 1314 | |
ef416fc2 | 1315 | /* |
e1d6a774 | 1316 | * 'free_sbcs_charmap()' - Free memory used by a single byte character set. |
ef416fc2 | 1317 | */ |
e1d6a774 | 1318 | |
1319 | static void | |
1320 | free_sbcs_charmap(_cups_cmap_t *cmap) /* I - Character set */ | |
ef416fc2 | 1321 | { |
e1d6a774 | 1322 | int i; /* Looping variable */ |
ef416fc2 | 1323 | |
ef416fc2 | 1324 | |
e1d6a774 | 1325 | for (i = 0; i < 256; i ++) |
1326 | if (cmap->uni2char[i]) | |
1327 | free(cmap->uni2char[i]); | |
1328 | ||
1329 | free(cmap); | |
1330 | } | |
1331 | ||
1332 | ||
1333 | /* | |
1334 | * 'free_vbcs_charmap()' - Free memory used by a variable byte character set. | |
1335 | */ | |
1336 | ||
1337 | static void | |
1338 | free_vbcs_charmap(_cups_vmap_t *vmap) /* I - Character set */ | |
1339 | { | |
1340 | int i; /* Looping variable */ | |
1341 | ||
1342 | ||
1343 | for (i = 0; i < 256; i ++) | |
1344 | if (vmap->char2uni[i]) | |
1345 | free(vmap->char2uni[i]); | |
1346 | ||
1347 | for (i = 0; i < 256; i ++) | |
1348 | if (vmap->uni2char[i]) | |
1349 | free(vmap->uni2char[i]); | |
1350 | ||
1351 | if (vmap->wide2uni) | |
1352 | free(vmap->wide2uni); | |
1353 | ||
1354 | free(vmap); | |
1355 | } | |
1356 | ||
1357 | ||
d6ae789d | 1358 | /* |
1359 | * 'get_charmap()' - Lookup or get a character set map (private). | |
1360 | * | |
1361 | * This code handles single-byte (SBCS), double-byte (DBCS), and | |
1362 | * variable-byte (VBCS) character sets _without_ charset escapes... | |
1363 | * This code does not handle multiple-byte character sets (MBCS) | |
1364 | * (such as ISO-2022-JP) with charset switching via escapes... | |
1365 | */ | |
1366 | ||
1367 | ||
d09495fa | 1368 | static void * /* O - Charset map pointer */ |
d6ae789d | 1369 | get_charmap( |
1370 | const cups_encoding_t encoding) /* I - Encoding */ | |
1371 | { | |
1372 | char filename[1024]; /* Filename for charset map file */ | |
1373 | _cups_globals_t *cg = _cupsGlobals(); /* Global data */ | |
1374 | ||
1375 | ||
c9fc04c6 MS |
1376 | DEBUG_printf(("get_charmap(encoding=%d)\n", encoding)); |
1377 | ||
d6ae789d | 1378 | /* |
1379 | * Get the data directory and charset map name... | |
1380 | */ | |
1381 | ||
1382 | snprintf(filename, sizeof(filename), "%s/charmaps/%s.txt", | |
1383 | cg->cups_datadir, _cupsEncodingName(encoding)); | |
1384 | ||
c9fc04c6 | 1385 | DEBUG_printf(("get_charmap: filename=\"%s\"\n", filename)); |
d6ae789d | 1386 | |
1387 | /* | |
1388 | * Read charset map input file into cache... | |
1389 | */ | |
1390 | ||
1391 | if (encoding < CUPS_ENCODING_SBCS_END) | |
1392 | return (get_sbcs_charmap(encoding, filename)); | |
1393 | else if (encoding < CUPS_ENCODING_VBCS_END) | |
1394 | return (get_vbcs_charmap(encoding, filename)); | |
1395 | else | |
1396 | return (NULL); | |
1397 | } | |
1398 | ||
1399 | ||
e1d6a774 | 1400 | /* |
1401 | * 'get_charmap_count()' - Count lines in a charmap file. | |
1402 | */ | |
1403 | ||
1404 | static int /* O - Count or -1 on error */ | |
1405 | get_charmap_count(cups_file_t *fp) /* I - File to read from */ | |
1406 | { | |
1407 | int count; /* Number of lines */ | |
1408 | char line[256]; /* Line from input map file */ | |
ef416fc2 | 1409 | |
ef416fc2 | 1410 | |
1411 | /* | |
e1d6a774 | 1412 | * Count lines in map input file... |
ef416fc2 | 1413 | */ |
ef416fc2 | 1414 | |
e1d6a774 | 1415 | count = 0; |
ef416fc2 | 1416 | |
e1d6a774 | 1417 | while (cupsFileGets(fp, line, sizeof(line))) |
1418 | if (line[0] == '0') | |
1419 | count ++; | |
ef416fc2 | 1420 | |
e1d6a774 | 1421 | /* |
1422 | * Return the number of lines... | |
1423 | */ | |
1424 | ||
1425 | if (count > 0) | |
1426 | return (count); | |
1427 | else | |
1428 | return (-1); | |
ef416fc2 | 1429 | } |
1430 | ||
e1d6a774 | 1431 | |
ef416fc2 | 1432 | /* |
e1d6a774 | 1433 | * 'get_sbcs_charmap()' - Get SBCS Charmap. |
ef416fc2 | 1434 | */ |
e1d6a774 | 1435 | |
1436 | static _cups_cmap_t * /* O - Charmap or 0 on error */ | |
1437 | get_sbcs_charmap( | |
1438 | const cups_encoding_t encoding, /* I - Charmap Encoding */ | |
1439 | const char *filename) /* I - Charmap Filename */ | |
ef416fc2 | 1440 | { |
e1d6a774 | 1441 | unsigned long legchar; /* Legacy character value */ |
1442 | cups_utf32_t unichar; /* Unicode character value */ | |
1443 | _cups_cmap_t *cmap; /* Legacy SBCS / Unicode Charset Map */ | |
1444 | cups_file_t *fp; /* Charset map file pointer */ | |
1445 | char *s; /* Line parsing pointer */ | |
1446 | cups_ucs2_t *crow; /* Pointer to UCS-2 row in 'char2uni' */ | |
1447 | cups_sbcs_t *srow; /* Pointer to SBCS row in 'uni2char' */ | |
1448 | char line[256]; /* Line from charset map file */ | |
e1d6a774 | 1449 | |
ef416fc2 | 1450 | |
1451 | /* | |
e1d6a774 | 1452 | * See if we already have this SBCS charset map loaded... |
ef416fc2 | 1453 | */ |
e1d6a774 | 1454 | |
c9fc04c6 MS |
1455 | DEBUG_printf(("get_sbcs_charmap(encoding=%d, filename=\"%s\")\n", encoding, |
1456 | filename)); | |
1457 | ||
d6ae789d | 1458 | for (cmap = cmap_cache; cmap; cmap = cmap->next) |
e1d6a774 | 1459 | { |
1460 | if (cmap->encoding == encoding) | |
1461 | { | |
1462 | cmap->used ++; | |
c9fc04c6 | 1463 | DEBUG_printf(("get_sbcs_charmap: Returning existing cmap=%p\n", cmap)); |
d6ae789d | 1464 | |
e1d6a774 | 1465 | return ((void *)cmap); |
1466 | } | |
1467 | } | |
ef416fc2 | 1468 | |
1469 | /* | |
e1d6a774 | 1470 | * Open SBCS charset map input file... |
ef416fc2 | 1471 | */ |
e1d6a774 | 1472 | |
1473 | if ((fp = cupsFileOpen(filename, "r")) == NULL) | |
c9fc04c6 MS |
1474 | { |
1475 | DEBUG_printf(("get_sbcs_charmap: Returning NULL (%s)\n", strerror(errno))); | |
1476 | ||
e1d6a774 | 1477 | return (NULL); |
c9fc04c6 | 1478 | } |
ef416fc2 | 1479 | |
1480 | /* | |
e1d6a774 | 1481 | * Allocate memory for SBCS charset map... |
ef416fc2 | 1482 | */ |
e1d6a774 | 1483 | |
1484 | if ((cmap = (_cups_cmap_t *)calloc(1, sizeof(_cups_cmap_t))) == NULL) | |
1485 | { | |
1486 | cupsFileClose(fp); | |
c9fc04c6 | 1487 | DEBUG_puts("get_sbcs_charmap: Returning NULL (Unable to allocate memory)"); |
d6ae789d | 1488 | |
e1d6a774 | 1489 | return (NULL); |
1490 | } | |
1491 | ||
1492 | cmap->used ++; | |
1493 | cmap->encoding = encoding; | |
ef416fc2 | 1494 | |
1495 | /* | |
e1d6a774 | 1496 | * Save SBCS charset map into memory for transcoding... |
ef416fc2 | 1497 | */ |
e1d6a774 | 1498 | |
1499 | while (cupsFileGets(fp, line, sizeof(line))) | |
ef416fc2 | 1500 | { |
e1d6a774 | 1501 | if (line[0] != '0') |
1502 | continue; | |
1503 | ||
1504 | legchar = strtol(line, &s, 16); | |
1505 | if (legchar < 0 || legchar > 0xff) | |
1506 | goto sbcs_error; | |
1507 | ||
1508 | unichar = strtol(s, NULL, 16); | |
1509 | if (unichar < 0 || unichar > 0xffff) | |
1510 | goto sbcs_error; | |
ef416fc2 | 1511 | |
1512 | /* | |
e1d6a774 | 1513 | * Save legacy to Unicode mapping in direct lookup table... |
ef416fc2 | 1514 | */ |
e1d6a774 | 1515 | |
1516 | crow = cmap->char2uni + legchar; | |
1517 | *crow = (cups_ucs2_t)(unichar & 0xffff); | |
ef416fc2 | 1518 | |
1519 | /* | |
e1d6a774 | 1520 | * Save Unicode to legacy mapping in indirect lookup table... |
ef416fc2 | 1521 | */ |
e1d6a774 | 1522 | |
1523 | srow = cmap->uni2char[(unichar >> 8) & 0xff]; | |
1524 | if (!srow) | |
ef416fc2 | 1525 | { |
e1d6a774 | 1526 | srow = (cups_sbcs_t *)calloc(256, sizeof(cups_sbcs_t)); |
1527 | if (!srow) | |
1528 | goto sbcs_error; | |
1529 | ||
1530 | cmap->uni2char[(unichar >> 8) & 0xff] = srow; | |
ef416fc2 | 1531 | } |
1532 | ||
e1d6a774 | 1533 | srow += unichar & 0xff; |
1534 | ||
ef416fc2 | 1535 | /* |
e1d6a774 | 1536 | * Convert Replacement Character to visible replacement... |
ef416fc2 | 1537 | */ |
e1d6a774 | 1538 | |
1539 | if (unichar == 0xfffd) | |
1540 | legchar = (unsigned long)'?'; | |
ef416fc2 | 1541 | |
1542 | /* | |
e1d6a774 | 1543 | * First (oldest) legacy character uses Unicode mapping cell... |
ef416fc2 | 1544 | */ |
ef416fc2 | 1545 | |
e1d6a774 | 1546 | if (!*srow) |
1547 | *srow = (cups_sbcs_t)legchar; | |
1548 | } | |
ef416fc2 | 1549 | |
e1d6a774 | 1550 | cupsFileClose(fp); |
1551 | ||
ef416fc2 | 1552 | /* |
e1d6a774 | 1553 | * Add it to the cache and return... |
ef416fc2 | 1554 | */ |
e1d6a774 | 1555 | |
d6ae789d | 1556 | cmap->next = cmap_cache; |
1557 | cmap_cache = cmap; | |
e1d6a774 | 1558 | |
c9fc04c6 | 1559 | DEBUG_printf(("get_sbcs_charmap: Returning new cmap=%p\n", cmap)); |
e1d6a774 | 1560 | |
1561 | return (cmap); | |
ef416fc2 | 1562 | |
1563 | /* | |
e1d6a774 | 1564 | * If we get here, there was an error in the cmap file... |
ef416fc2 | 1565 | */ |
e1d6a774 | 1566 | |
1567 | sbcs_error: | |
1568 | ||
1569 | free_sbcs_charmap(cmap); | |
1570 | ||
1571 | cupsFileClose(fp); | |
1572 | ||
c9fc04c6 | 1573 | DEBUG_puts("get_sbcs_charmap: Returning NULL (Read/format error)"); |
e1d6a774 | 1574 | |
1575 | return (NULL); | |
1576 | } | |
1577 | ||
1578 | ||
1579 | /* | |
1580 | * 'get_vbcs_charmap()' - Get DBCS/VBCS Charmap. | |
1581 | */ | |
1582 | ||
1583 | static _cups_vmap_t * /* O - Charmap or 0 on error */ | |
1584 | get_vbcs_charmap( | |
1585 | const cups_encoding_t encoding, /* I - Charmap Encoding */ | |
1586 | const char *filename) /* I - Charmap Filename */ | |
1587 | { | |
1588 | _cups_vmap_t *vmap; /* Legacy VBCS / Unicode Charset Map */ | |
1589 | cups_ucs2_t *crow; /* Pointer to UCS-2 row in 'char2uni' */ | |
1590 | cups_vbcs_t *vrow; /* Pointer to VBCS row in 'uni2char' */ | |
1591 | _cups_wide2uni_t *wide2uni; /* Pointer to row in 'wide2uni' */ | |
1592 | cups_sbcs_t leadchar; /* Lead char of 2-byte legacy char */ | |
1593 | unsigned long legchar; /* Legacy character value */ | |
1594 | cups_utf32_t unichar; /* Unicode character value */ | |
1595 | int mapcount; /* Count of lines in charmap file */ | |
1596 | cups_file_t *fp; /* Charset map file pointer */ | |
1597 | char *s; /* Line parsing pointer */ | |
1598 | char line[256]; /* Line from charset map file */ | |
1599 | int i; /* Loop variable */ | |
09a101d6 | 1600 | int legacy; /* 32-bit legacy char */ |
e1d6a774 | 1601 | |
1602 | ||
1603 | DEBUG_printf(("get_vbcs_charmap(encoding=%d, filename=\"%s\")\n", | |
1604 | encoding, filename)); | |
ef416fc2 | 1605 | |
1606 | /* | |
e1d6a774 | 1607 | * See if we already have this DBCS/VBCS charset map loaded... |
ef416fc2 | 1608 | */ |
ef416fc2 | 1609 | |
d6ae789d | 1610 | for (vmap = vmap_cache; vmap; vmap = vmap->next) |
e1d6a774 | 1611 | { |
1612 | if (vmap->encoding == encoding) | |
ef416fc2 | 1613 | { |
e1d6a774 | 1614 | vmap->used ++; |
c9fc04c6 | 1615 | DEBUG_printf(("get_vbcs_charmap: Returning existing vmap=%p\n", vmap)); |
d6ae789d | 1616 | |
e1d6a774 | 1617 | return ((void *)vmap); |
ef416fc2 | 1618 | } |
ef416fc2 | 1619 | } |
ef416fc2 | 1620 | |
1621 | /* | |
e1d6a774 | 1622 | * Open VBCS charset map input file... |
ef416fc2 | 1623 | */ |
ef416fc2 | 1624 | |
e1d6a774 | 1625 | if ((fp = cupsFileOpen(filename, "r")) == NULL) |
1626 | { | |
c9fc04c6 | 1627 | DEBUG_printf(("get_vbcs_charmap: Returning NULL (%s)\n", strerror(errno))); |
d6ae789d | 1628 | |
e1d6a774 | 1629 | return (NULL); |
1630 | } | |
ef416fc2 | 1631 | |
1632 | /* | |
e1d6a774 | 1633 | * Count lines in charmap file... |
ef416fc2 | 1634 | */ |
e1d6a774 | 1635 | |
1636 | if ((mapcount = get_charmap_count(fp)) <= 0) | |
1637 | { | |
c9fc04c6 | 1638 | DEBUG_puts("get_vbcs_charmap: Unable to get charmap count!"); |
d6ae789d | 1639 | |
91c84a35 MS |
1640 | cupsFileClose(fp); |
1641 | ||
e1d6a774 | 1642 | return (NULL); |
1643 | } | |
1644 | ||
c9fc04c6 | 1645 | DEBUG_printf(("get_vbcs_charmap: mapcount=%d\n", mapcount)); |
ef416fc2 | 1646 | |
1647 | /* | |
e1d6a774 | 1648 | * Allocate memory for DBCS/VBCS charset map... |
ef416fc2 | 1649 | */ |
e1d6a774 | 1650 | |
1651 | if ((vmap = (_cups_vmap_t *)calloc(1, sizeof(_cups_vmap_t))) == NULL) | |
1652 | { | |
c9fc04c6 | 1653 | DEBUG_puts("get_vbcs_charmap: Unable to allocate memory!"); |
d6ae789d | 1654 | |
91c84a35 MS |
1655 | cupsFileClose(fp); |
1656 | ||
e1d6a774 | 1657 | return (NULL); |
1658 | } | |
1659 | ||
1660 | vmap->used ++; | |
1661 | vmap->encoding = encoding; | |
ef416fc2 | 1662 | |
1663 | /* | |
e1d6a774 | 1664 | * Save DBCS/VBCS charset map into memory for transcoding... |
ef416fc2 | 1665 | */ |
e1d6a774 | 1666 | |
1667 | leadchar = 0; | |
1668 | wide2uni = NULL; | |
1669 | ||
1670 | cupsFileRewind(fp); | |
1671 | ||
09a101d6 | 1672 | i = 0; |
1673 | legacy = 0; | |
e1d6a774 | 1674 | |
1675 | while (cupsFileGets(fp, line, sizeof(line))) | |
ef416fc2 | 1676 | { |
e1d6a774 | 1677 | if (line[0] != '0') |
1678 | continue; | |
1679 | ||
1680 | legchar = strtoul(line, &s, 16); | |
1681 | if (legchar == ULONG_MAX) | |
1682 | goto vbcs_error; | |
1683 | ||
1684 | unichar = strtol(s, NULL, 16); | |
1685 | if (unichar < 0 || unichar > 0xffff) | |
1686 | goto vbcs_error; | |
1687 | ||
1688 | i ++; | |
1689 | ||
1690 | /* DEBUG_printf((" i=%d, legchar=0x%08lx, unichar=0x%04x\n", i, | |
1691 | legchar, (unsigned)unichar)); */ | |
ef416fc2 | 1692 | |
1693 | /* | |
e1d6a774 | 1694 | * Save lead char of 2/3/4-byte legacy char... |
ef416fc2 | 1695 | */ |
e1d6a774 | 1696 | |
c9fc04c6 | 1697 | if (legchar > 0xffffff) |
ef416fc2 | 1698 | { |
c9fc04c6 MS |
1699 | leadchar = (cups_sbcs_t)(legchar >> 24); |
1700 | vmap->lead4char[leadchar] = leadchar; | |
e1d6a774 | 1701 | } |
c9fc04c6 | 1702 | else if (legchar > 0xffff) |
e1d6a774 | 1703 | { |
1704 | leadchar = (cups_sbcs_t)(legchar >> 16); | |
1705 | vmap->lead3char[leadchar] = leadchar; | |
1706 | } | |
c9fc04c6 | 1707 | else if (legchar > 0xff) |
e1d6a774 | 1708 | { |
c9fc04c6 MS |
1709 | leadchar = (cups_sbcs_t)(legchar >> 8); |
1710 | vmap->lead2char[leadchar] = leadchar; | |
ef416fc2 | 1711 | } |
1712 | ||
1713 | /* | |
e1d6a774 | 1714 | * Save Legacy to Unicode mapping... |
ef416fc2 | 1715 | */ |
e1d6a774 | 1716 | |
1717 | if (legchar <= 0xffff) | |
ef416fc2 | 1718 | { |
ef416fc2 | 1719 | /* |
e1d6a774 | 1720 | * Save DBCS 16-bit to Unicode mapping in indirect lookup table... |
ef416fc2 | 1721 | */ |
e1d6a774 | 1722 | |
1723 | crow = vmap->char2uni[(int)leadchar]; | |
1724 | if (!crow) | |
1725 | { | |
1726 | crow = (cups_ucs2_t *)calloc(256, sizeof(cups_ucs2_t)); | |
1727 | if (!crow) | |
1728 | goto vbcs_error; | |
1729 | ||
1730 | vmap->char2uni[(int)leadchar] = crow; | |
1731 | } | |
1732 | ||
1733 | crow[(int)(legchar & 0xff)] = (cups_ucs2_t)unichar; | |
1734 | } | |
1735 | else | |
1736 | { | |
1737 | /* | |
1738 | * Save VBCS 32-bit to Unicode mapping in sorted list table... | |
1739 | */ | |
1740 | ||
09a101d6 | 1741 | if (!legacy) |
e1d6a774 | 1742 | { |
09a101d6 | 1743 | legacy = 1; |
e1d6a774 | 1744 | vmap->widecount = (mapcount - i + 1); |
1745 | wide2uni = (_cups_wide2uni_t *)calloc(vmap->widecount, | |
1746 | sizeof(_cups_wide2uni_t)); | |
1747 | if (!wide2uni) | |
1748 | goto vbcs_error; | |
1749 | ||
1750 | vmap->wide2uni = wide2uni; | |
1751 | } | |
1752 | ||
1753 | wide2uni->widechar = (cups_vbcs_t)legchar; | |
1754 | wide2uni->unichar = (cups_ucs2_t)unichar; | |
1755 | wide2uni ++; | |
ef416fc2 | 1756 | } |
1757 | ||
1758 | /* | |
e1d6a774 | 1759 | * Save Unicode to legacy mapping in indirect lookup table... |
ef416fc2 | 1760 | */ |
e1d6a774 | 1761 | |
1762 | vrow = vmap->uni2char[(int)((unichar >> 8) & 0xff)]; | |
1763 | if (!vrow) | |
ef416fc2 | 1764 | { |
e1d6a774 | 1765 | vrow = (cups_vbcs_t *)calloc(256, sizeof(cups_vbcs_t)); |
1766 | if (!vrow) | |
1767 | goto vbcs_error; | |
1768 | ||
1769 | vmap->uni2char[(int) ((unichar >> 8) & 0xff)] = vrow; | |
ef416fc2 | 1770 | } |
e1d6a774 | 1771 | |
1772 | vrow += (int)(unichar & 0xff); | |
ef416fc2 | 1773 | |
1774 | /* | |
e1d6a774 | 1775 | * Convert Replacement Character to visible replacement... |
ef416fc2 | 1776 | */ |
e1d6a774 | 1777 | |
1778 | if (unichar == 0xfffd) | |
1779 | legchar = (unsigned long)'?'; | |
ef416fc2 | 1780 | |
1781 | /* | |
e1d6a774 | 1782 | * First (oldest) legacy character uses Unicode mapping cell... |
ef416fc2 | 1783 | */ |
e1d6a774 | 1784 | |
1785 | if (!*vrow) | |
1786 | *vrow = (cups_vbcs_t)legchar; | |
ef416fc2 | 1787 | } |
e1d6a774 | 1788 | |
1789 | vmap->charcount = (i - vmap->widecount); | |
1790 | ||
1791 | cupsFileClose(fp); | |
ef416fc2 | 1792 | |
1793 | /* | |
e1d6a774 | 1794 | * Add it to the cache and return... |
ef416fc2 | 1795 | */ |
ef416fc2 | 1796 | |
c9fc04c6 | 1797 | vmap->next = vmap_cache; |
d6ae789d | 1798 | vmap_cache = vmap; |
e1d6a774 | 1799 | |
c9fc04c6 | 1800 | DEBUG_printf(("get_vbcs_charmap: Returning new vmap=%p\n", vmap)); |
e1d6a774 | 1801 | |
1802 | return (vmap); | |
1803 | ||
1804 | /* | |
1805 | * If we get here, the file contains errors... | |
1806 | */ | |
1807 | ||
1808 | vbcs_error: | |
1809 | ||
1810 | free_vbcs_charmap(vmap); | |
1811 | ||
1812 | cupsFileClose(fp); | |
1813 | ||
c9fc04c6 | 1814 | DEBUG_puts("get_vbcs_charmap: Returning NULL (Read/format error)"); |
e1d6a774 | 1815 | |
1816 | return (NULL); | |
ef416fc2 | 1817 | } |
1818 | ||
1819 | ||
1820 | /* | |
75bd9771 | 1821 | * End of "$Id: transcode.c 7560 2008-05-13 06:34:04Z mike $" |
ef416fc2 | 1822 | */ |