1 /*
2 * iconv implementation using Win32 API to convert.
3 *
4 * This file is placed in the public domain.
5 */
6
7 /* for WC_NO_BEST_FIT_CHARS */
8 #ifndef WINVER
9 # define WINVER 0x0500
10 #endif
11
12 #define STRICT
13 #include <windows.h>
14 #include <errno.h>
15 #include <string.h>
16 #include <stdlib.h>
17
18 #ifdef __GNUC__
19 #define UNUSED __attribute__((unused))
20 #else
21 #define UNUSED
22 #endif
23
24 /* WORKAROUND: */
25 #ifndef UNDER_CE
26 #define GetProcAddressA GetProcAddress
27 #endif
28
29 #if 0
30 # define MAKE_EXE
31 # define MAKE_DLL
32 # define USE_LIBICONV_DLL
33 #endif
34
35 #if !defined(DEFAULT_LIBICONV_DLL)
36 # define DEFAULT_LIBICONV_DLL ""
37 #endif
38
39 #define MB_CHAR_MAX 16
40
41 #define UNICODE_MODE_BOM_DONE 1
42 #define UNICODE_MODE_SWAPPED 2
43
44 #define FLAG_USE_BOM 1
45 #define FLAG_TRANSLIT 2 /* //TRANSLIT */
46 #define FLAG_IGNORE 4 /* //IGNORE */
47
48 typedef unsigned char uchar;
49 typedef unsigned short ushort;
50 typedef unsigned int uint;
51
52 typedef void* iconv_t;
53
54 iconv_t iconv_open(const char *tocode, const char *fromcode);
55 int iconv_close(iconv_t cd);
56 size_t iconv(iconv_t cd, const char **inbuf, size_t *inbytesleft, char **outbuf, size_t *outbytesleft);
57
58 /* libiconv interface for vim */
59 #if defined(MAKE_DLL)
60 int
iconvctl(iconv_t cd,int request,void * argument)61 iconvctl (iconv_t cd, int request, void* argument)
62 {
63 /* not supported */
64 return 0;
65 }
66 #endif
67
68 typedef struct compat_t compat_t;
69 typedef struct csconv_t csconv_t;
70 typedef struct rec_iconv_t rec_iconv_t;
71
72 typedef iconv_t (*f_iconv_open)(const char *tocode, const char *fromcode);
73 typedef int (*f_iconv_close)(iconv_t cd);
74 typedef size_t (*f_iconv)(iconv_t cd, const char **inbuf, size_t *inbytesleft, char **outbuf, size_t *outbytesleft);
75 typedef int* (*f_errno)(void);
76 typedef int (*f_mbtowc)(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize);
77 typedef int (*f_wctomb)(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize);
78 typedef int (*f_mblen)(csconv_t *cv, const uchar *buf, int bufsize);
79 typedef int (*f_flush)(csconv_t *cv, uchar *buf, int bufsize);
80
81 #define COMPAT_IN 1
82 #define COMPAT_OUT 2
83
84 /* unicode mapping for compatibility with other conversion table. */
85 struct compat_t {
86 uint in;
87 uint out;
88 uint flag;
89 };
90
91 struct csconv_t {
92 int codepage;
93 int flags;
94 f_mbtowc mbtowc;
95 f_wctomb wctomb;
96 f_mblen mblen;
97 f_flush flush;
98 DWORD mode;
99 compat_t *compat;
100 };
101
102 struct rec_iconv_t {
103 iconv_t cd;
104 f_iconv_close iconv_close;
105 f_iconv iconv;
106 f_errno _errno;
107 csconv_t from;
108 csconv_t to;
109 #if defined(USE_LIBICONV_DLL)
110 HMODULE hlibiconv;
111 #endif
112 };
113
114 static int win_iconv_open(rec_iconv_t *cd, const char *tocode, const char *fromcode);
115 static int win_iconv_close(iconv_t cd);
116 static size_t win_iconv(iconv_t cd, const char **inbuf, size_t *inbytesleft, char **outbuf, size_t *outbytesleft);
117
118 static int load_mlang(void);
119 static int make_csconv(const char *name, csconv_t *cv);
120 static int name_to_codepage(const char *name);
121 static uint utf16_to_ucs4(const ushort *wbuf);
122 static void ucs4_to_utf16(uint wc, ushort *wbuf, int *wbufsize);
123 static int mbtowc_flags(int codepage);
124 static int must_use_null_useddefaultchar(int codepage);
125 static char *strrstr(const char *str, const char *token);
126 static char *xstrndup(const char *s, size_t n);
127 static int seterror(int err);
128
129 #if defined(USE_LIBICONV_DLL)
130 static int libiconv_iconv_open(rec_iconv_t *cd, const char *tocode, const char *fromcode);
131 static PVOID MyImageDirectoryEntryToData(LPVOID Base, BOOLEAN MappedAsImage, USHORT DirectoryEntry, PULONG Size);
132 static FARPROC find_imported_function(HMODULE hModule, const char *funcname);
133
134 static HMODULE hwiniconv;
135 #endif
136
137 static int sbcs_mblen(csconv_t *cv, const uchar *buf, int bufsize);
138 static int dbcs_mblen(csconv_t *cv, const uchar *buf, int bufsize);
139 static int mbcs_mblen(csconv_t *cv, const uchar *buf, int bufsize);
140 static int utf8_mblen(csconv_t *cv, const uchar *buf, int bufsize);
141 static int eucjp_mblen(csconv_t *cv, const uchar *buf, int bufsize);
142
143 static int kernel_mbtowc(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize);
144 static int kernel_wctomb(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize);
145 static int mlang_mbtowc(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize);
146 static int mlang_wctomb(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize);
147 static int utf16_mbtowc(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize);
148 static int utf16_wctomb(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize);
149 static int utf32_mbtowc(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize);
150 static int utf32_wctomb(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize);
151 static int iso2022jp_mbtowc(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize);
152 static int iso2022jp_wctomb(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize);
153 static int iso2022jp_flush(csconv_t *cv, uchar *buf, int bufsize);
154
155 static struct {
156 int codepage;
157 const char *name;
158 } codepage_alias[] = {
159 {65001, "CP65001"},
160 {65001, "UTF8"},
161 {65001, "UTF-8"},
162
163 {1200, "CP1200"},
164 {1200, "UTF16LE"},
165 {1200, "UTF-16LE"},
166 {1200, "UCS2LE"},
167 {1200, "UCS-2LE"},
168
169 {1201, "CP1201"},
170 {1201, "UTF16BE"},
171 {1201, "UTF-16BE"},
172 {1201, "UCS2BE"},
173 {1201, "UCS-2BE"},
174 {1201, "unicodeFFFE"},
175
176 {12000, "CP12000"},
177 {12000, "UTF32LE"},
178 {12000, "UTF-32LE"},
179 {12000, "UCS4LE"},
180 {12000, "UCS-4LE"},
181
182 {12001, "CP12001"},
183 {12001, "UTF32BE"},
184 {12001, "UTF-32BE"},
185 {12001, "UCS4BE"},
186 {12001, "UCS-4BE"},
187
188 #ifndef GLIB_COMPILATION
189 /*
190 * Default is big endian.
191 * See rfc2781 4.3 Interpreting text labelled as UTF-16.
192 */
193 {1201, "UTF16"},
194 {1201, "UTF-16"},
195 {1201, "UCS2"},
196 {1201, "UCS-2"},
197 {12001, "UTF32"},
198 {12001, "UTF-32"},
199 {12001, "UCS-4"},
200 {12001, "UCS4"},
201 #else
202 /* Default is little endian, because the platform is */
203 {1200, "UTF16"},
204 {1200, "UTF-16"},
205 {1200, "UCS2"},
206 {1200, "UCS-2"},
207 {12000, "UTF32"},
208 {12000, "UTF-32"},
209 {12000, "UCS4"},
210 {12000, "UCS-4"},
211 #endif
212
213 /* copy from libiconv `iconv -l` */
214 /* !IsValidCodePage(367) */
215 {20127, "ANSI_X3.4-1968"},
216 {20127, "ANSI_X3.4-1986"},
217 {20127, "ASCII"},
218 {20127, "CP367"},
219 {20127, "IBM367"},
220 {20127, "ISO-IR-6"},
221 {20127, "ISO646-US"},
222 {20127, "ISO_646.IRV:1991"},
223 {20127, "US"},
224 {20127, "US-ASCII"},
225 {20127, "CSASCII"},
226
227 /* !IsValidCodePage(819) */
228 {1252, "CP819"},
229 {1252, "IBM819"},
230 {28591, "ISO-8859-1"},
231 {28591, "ISO-IR-100"},
232 {28591, "ISO8859-1"},
233 {28591, "ISO_8859-1"},
234 {28591, "ISO_8859-1:1987"},
235 {28591, "L1"},
236 {28591, "LATIN1"},
237 {28591, "CSISOLATIN1"},
238
239 {1250, "CP1250"},
240 {1250, "MS-EE"},
241 {1250, "WINDOWS-1250"},
242
243 {1251, "CP1251"},
244 {1251, "MS-CYRL"},
245 {1251, "WINDOWS-1251"},
246
247 {1252, "CP1252"},
248 {1252, "MS-ANSI"},
249 {1252, "WINDOWS-1252"},
250
251 {1253, "CP1253"},
252 {1253, "MS-GREEK"},
253 {1253, "WINDOWS-1253"},
254
255 {1254, "CP1254"},
256 {1254, "MS-TURK"},
257 {1254, "WINDOWS-1254"},
258
259 {1255, "CP1255"},
260 {1255, "MS-HEBR"},
261 {1255, "WINDOWS-1255"},
262
263 {1256, "CP1256"},
264 {1256, "MS-ARAB"},
265 {1256, "WINDOWS-1256"},
266
267 {1257, "CP1257"},
268 {1257, "WINBALTRIM"},
269 {1257, "WINDOWS-1257"},
270
271 {1258, "CP1258"},
272 {1258, "WINDOWS-1258"},
273
274 {850, "850"},
275 {850, "CP850"},
276 {850, "IBM850"},
277 {850, "CSPC850MULTILINGUAL"},
278
279 /* !IsValidCodePage(862) */
280 {862, "862"},
281 {862, "CP862"},
282 {862, "IBM862"},
283 {862, "CSPC862LATINHEBREW"},
284
285 {866, "866"},
286 {866, "CP866"},
287 {866, "IBM866"},
288 {866, "CSIBM866"},
289
290 /* !IsValidCodePage(154) */
291 {154, "CP154"},
292 {154, "CYRILLIC-ASIAN"},
293 {154, "PT154"},
294 {154, "PTCP154"},
295 {154, "CSPTCP154"},
296
297 /* !IsValidCodePage(1133) */
298 {1133, "CP1133"},
299 {1133, "IBM-CP1133"},
300
301 {874, "CP874"},
302 {874, "WINDOWS-874"},
303
304 /* !IsValidCodePage(51932) */
305 {51932, "CP51932"},
306 {51932, "MS51932"},
307 {51932, "WINDOWS-51932"},
308 {51932, "EUC-JP"},
309
310 {932, "CP932"},
311 {932, "MS932"},
312 {932, "SHIFFT_JIS"},
313 {932, "SHIFFT_JIS-MS"},
314 {932, "SJIS"},
315 {932, "SJIS-MS"},
316 {932, "SJIS-OPEN"},
317 {932, "SJIS-WIN"},
318 {932, "WINDOWS-31J"},
319 {932, "WINDOWS-932"},
320 {932, "CSWINDOWS31J"},
321
322 {50221, "CP50221"},
323 {50221, "ISO-2022-JP"},
324 {50221, "ISO-2022-JP-MS"},
325 {50221, "ISO2022-JP"},
326 {50221, "ISO2022-JP-MS"},
327 {50221, "MS50221"},
328 {50221, "WINDOWS-50221"},
329
330 {936, "CP936"},
331 {936, "GBK"},
332 {936, "MS936"},
333 {936, "WINDOWS-936"},
334
335 {950, "CP950"},
336 {950, "BIG5"},
337 {950, "BIG5HKSCS"},
338 {950, "BIG5-HKSCS"},
339
340 {949, "CP949"},
341 {949, "UHC"},
342 {949, "EUC-KR"},
343
344 {1361, "CP1361"},
345 {1361, "JOHAB"},
346
347 {437, "437"},
348 {437, "CP437"},
349 {437, "IBM437"},
350 {437, "CSPC8CODEPAGE437"},
351
352 {737, "CP737"},
353
354 {775, "CP775"},
355 {775, "IBM775"},
356 {775, "CSPC775BALTIC"},
357
358 {852, "852"},
359 {852, "CP852"},
360 {852, "IBM852"},
361 {852, "CSPCP852"},
362
363 /* !IsValidCodePage(853) */
364 {853, "CP853"},
365
366 {855, "855"},
367 {855, "CP855"},
368 {855, "IBM855"},
369 {855, "CSIBM855"},
370
371 {857, "857"},
372 {857, "CP857"},
373 {857, "IBM857"},
374 {857, "CSIBM857"},
375
376 /* !IsValidCodePage(858) */
377 {858, "CP858"},
378
379 {860, "860"},
380 {860, "CP860"},
381 {860, "IBM860"},
382 {860, "CSIBM860"},
383
384 {861, "861"},
385 {861, "CP-IS"},
386 {861, "CP861"},
387 {861, "IBM861"},
388 {861, "CSIBM861"},
389
390 {863, "863"},
391 {863, "CP863"},
392 {863, "IBM863"},
393 {863, "CSIBM863"},
394
395 {864, "CP864"},
396 {864, "IBM864"},
397 {864, "CSIBM864"},
398
399 {865, "865"},
400 {865, "CP865"},
401 {865, "IBM865"},
402 {865, "CSIBM865"},
403
404 {869, "869"},
405 {869, "CP-GR"},
406 {869, "CP869"},
407 {869, "IBM869"},
408 {869, "CSIBM869"},
409
410 /* !IsValidCodePage(1152) */
411 {1125, "CP1125"},
412
413 /*
414 * Code Page Identifiers
415 * http://msdn2.microsoft.com/en-us/library/ms776446.aspx
416 */
417 {37, "IBM037"}, /* IBM EBCDIC US-Canada */
418 {437, "IBM437"}, /* OEM United States */
419 {500, "IBM500"}, /* IBM EBCDIC International */
420 {708, "ASMO-708"}, /* Arabic (ASMO 708) */
421 /* 709 Arabic (ASMO-449+, BCON V4) */
422 /* 710 Arabic - Transparent Arabic */
423 {720, "DOS-720"}, /* Arabic (Transparent ASMO); Arabic (DOS) */
424 {737, "ibm737"}, /* OEM Greek (formerly 437G); Greek (DOS) */
425 {775, "ibm775"}, /* OEM Baltic; Baltic (DOS) */
426 {850, "ibm850"}, /* OEM Multilingual Latin 1; Western European (DOS) */
427 {852, "ibm852"}, /* OEM Latin 2; Central European (DOS) */
428 {855, "IBM855"}, /* OEM Cyrillic (primarily Russian) */
429 {857, "ibm857"}, /* OEM Turkish; Turkish (DOS) */
430 {858, "IBM00858"}, /* OEM Multilingual Latin 1 + Euro symbol */
431 {860, "IBM860"}, /* OEM Portuguese; Portuguese (DOS) */
432 {861, "ibm861"}, /* OEM Icelandic; Icelandic (DOS) */
433 {862, "DOS-862"}, /* OEM Hebrew; Hebrew (DOS) */
434 {863, "IBM863"}, /* OEM French Canadian; French Canadian (DOS) */
435 {864, "IBM864"}, /* OEM Arabic; Arabic (864) */
436 {865, "IBM865"}, /* OEM Nordic; Nordic (DOS) */
437 {866, "cp866"}, /* OEM Russian; Cyrillic (DOS) */
438 {869, "ibm869"}, /* OEM Modern Greek; Greek, Modern (DOS) */
439 {870, "IBM870"}, /* IBM EBCDIC Multilingual/ROECE (Latin 2); IBM EBCDIC Multilingual Latin 2 */
440 {874, "windows-874"}, /* ANSI/OEM Thai (same as 28605, ISO 8859-15); Thai (Windows) */
441 {875, "cp875"}, /* IBM EBCDIC Greek Modern */
442 {932, "shift_jis"}, /* ANSI/OEM Japanese; Japanese (Shift-JIS) */
443 {932, "shift-jis"}, /* alternative name for it */
444 {936, "gb2312"}, /* ANSI/OEM Simplified Chinese (PRC, Singapore); Chinese Simplified (GB2312) */
445 {949, "ks_c_5601-1987"}, /* ANSI/OEM Korean (Unified Hangul Code) */
446 {950, "big5"}, /* ANSI/OEM Traditional Chinese (Taiwan; Hong Kong SAR, PRC); Chinese Traditional (Big5) */
447 {950, "big5hkscs"}, /* ANSI/OEM Traditional Chinese (Hong Kong SAR); Chinese Traditional (Big5-HKSCS) */
448 {950, "big5-hkscs"}, /* alternative name for it */
449 {1026, "IBM1026"}, /* IBM EBCDIC Turkish (Latin 5) */
450 {1047, "IBM01047"}, /* IBM EBCDIC Latin 1/Open System */
451 {1140, "IBM01140"}, /* IBM EBCDIC US-Canada (037 + Euro symbol); IBM EBCDIC (US-Canada-Euro) */
452 {1141, "IBM01141"}, /* IBM EBCDIC Germany (20273 + Euro symbol); IBM EBCDIC (Germany-Euro) */
453 {1142, "IBM01142"}, /* IBM EBCDIC Denmark-Norway (20277 + Euro symbol); IBM EBCDIC (Denmark-Norway-Euro) */
454 {1143, "IBM01143"}, /* IBM EBCDIC Finland-Sweden (20278 + Euro symbol); IBM EBCDIC (Finland-Sweden-Euro) */
455 {1144, "IBM01144"}, /* IBM EBCDIC Italy (20280 + Euro symbol); IBM EBCDIC (Italy-Euro) */
456 {1145, "IBM01145"}, /* IBM EBCDIC Latin America-Spain (20284 + Euro symbol); IBM EBCDIC (Spain-Euro) */
457 {1146, "IBM01146"}, /* IBM EBCDIC United Kingdom (20285 + Euro symbol); IBM EBCDIC (UK-Euro) */
458 {1147, "IBM01147"}, /* IBM EBCDIC France (20297 + Euro symbol); IBM EBCDIC (France-Euro) */
459 {1148, "IBM01148"}, /* IBM EBCDIC International (500 + Euro symbol); IBM EBCDIC (International-Euro) */
460 {1149, "IBM01149"}, /* IBM EBCDIC Icelandic (20871 + Euro symbol); IBM EBCDIC (Icelandic-Euro) */
461 {1250, "windows-1250"}, /* ANSI Central European; Central European (Windows) */
462 {1251, "windows-1251"}, /* ANSI Cyrillic; Cyrillic (Windows) */
463 {1252, "windows-1252"}, /* ANSI Latin 1; Western European (Windows) */
464 {1253, "windows-1253"}, /* ANSI Greek; Greek (Windows) */
465 {1254, "windows-1254"}, /* ANSI Turkish; Turkish (Windows) */
466 {1255, "windows-1255"}, /* ANSI Hebrew; Hebrew (Windows) */
467 {1256, "windows-1256"}, /* ANSI Arabic; Arabic (Windows) */
468 {1257, "windows-1257"}, /* ANSI Baltic; Baltic (Windows) */
469 {1258, "windows-1258"}, /* ANSI/OEM Vietnamese; Vietnamese (Windows) */
470 {1361, "Johab"}, /* Korean (Johab) */
471 {10000, "macintosh"}, /* MAC Roman; Western European (Mac) */
472 {10001, "x-mac-japanese"}, /* Japanese (Mac) */
473 {10002, "x-mac-chinesetrad"}, /* MAC Traditional Chinese (Big5); Chinese Traditional (Mac) */
474 {10003, "x-mac-korean"}, /* Korean (Mac) */
475 {10004, "x-mac-arabic"}, /* Arabic (Mac) */
476 {10005, "x-mac-hebrew"}, /* Hebrew (Mac) */
477 {10006, "x-mac-greek"}, /* Greek (Mac) */
478 {10007, "x-mac-cyrillic"}, /* Cyrillic (Mac) */
479 {10008, "x-mac-chinesesimp"}, /* MAC Simplified Chinese (GB 2312); Chinese Simplified (Mac) */
480 {10010, "x-mac-romanian"}, /* Romanian (Mac) */
481 {10017, "x-mac-ukrainian"}, /* Ukrainian (Mac) */
482 {10021, "x-mac-thai"}, /* Thai (Mac) */
483 {10029, "x-mac-ce"}, /* MAC Latin 2; Central European (Mac) */
484 {10079, "x-mac-icelandic"}, /* Icelandic (Mac) */
485 {10081, "x-mac-turkish"}, /* Turkish (Mac) */
486 {10082, "x-mac-croatian"}, /* Croatian (Mac) */
487 {20000, "x-Chinese_CNS"}, /* CNS Taiwan; Chinese Traditional (CNS) */
488 {20001, "x-cp20001"}, /* TCA Taiwan */
489 {20002, "x_Chinese-Eten"}, /* Eten Taiwan; Chinese Traditional (Eten) */
490 {20003, "x-cp20003"}, /* IBM5550 Taiwan */
491 {20004, "x-cp20004"}, /* TeleText Taiwan */
492 {20005, "x-cp20005"}, /* Wang Taiwan */
493 {20105, "x-IA5"}, /* IA5 (IRV International Alphabet No. 5, 7-bit); Western European (IA5) */
494 {20106, "x-IA5-German"}, /* IA5 German (7-bit) */
495 {20107, "x-IA5-Swedish"}, /* IA5 Swedish (7-bit) */
496 {20108, "x-IA5-Norwegian"}, /* IA5 Norwegian (7-bit) */
497 {20127, "us-ascii"}, /* US-ASCII (7-bit) */
498 {20261, "x-cp20261"}, /* T.61 */
499 {20269, "x-cp20269"}, /* ISO 6937 Non-Spacing Accent */
500 {20273, "IBM273"}, /* IBM EBCDIC Germany */
501 {20277, "IBM277"}, /* IBM EBCDIC Denmark-Norway */
502 {20278, "IBM278"}, /* IBM EBCDIC Finland-Sweden */
503 {20280, "IBM280"}, /* IBM EBCDIC Italy */
504 {20284, "IBM284"}, /* IBM EBCDIC Latin America-Spain */
505 {20285, "IBM285"}, /* IBM EBCDIC United Kingdom */
506 {20290, "IBM290"}, /* IBM EBCDIC Japanese Katakana Extended */
507 {20297, "IBM297"}, /* IBM EBCDIC France */
508 {20420, "IBM420"}, /* IBM EBCDIC Arabic */
509 {20423, "IBM423"}, /* IBM EBCDIC Greek */
510 {20424, "IBM424"}, /* IBM EBCDIC Hebrew */
511 {20833, "x-EBCDIC-KoreanExtended"}, /* IBM EBCDIC Korean Extended */
512 {20838, "IBM-Thai"}, /* IBM EBCDIC Thai */
513 {20866, "koi8-r"}, /* Russian (KOI8-R); Cyrillic (KOI8-R) */
514 {20871, "IBM871"}, /* IBM EBCDIC Icelandic */
515 {20880, "IBM880"}, /* IBM EBCDIC Cyrillic Russian */
516 {20905, "IBM905"}, /* IBM EBCDIC Turkish */
517 {20924, "IBM00924"}, /* IBM EBCDIC Latin 1/Open System (1047 + Euro symbol) */
518 {20932, "EUC-JP"}, /* Japanese (JIS 0208-1990 and 0121-1990) */
519 {20936, "x-cp20936"}, /* Simplified Chinese (GB2312); Chinese Simplified (GB2312-80) */
520 {20949, "x-cp20949"}, /* Korean Wansung */
521 {21025, "cp1025"}, /* IBM EBCDIC Cyrillic Serbian-Bulgarian */
522 /* 21027 (deprecated) */
523 {21866, "koi8-u"}, /* Ukrainian (KOI8-U); Cyrillic (KOI8-U) */
524 {28591, "iso-8859-1"}, /* ISO 8859-1 Latin 1; Western European (ISO) */
525 {28591, "iso8859-1"}, /* ISO 8859-1 Latin 1; Western European (ISO) */
526 {28591, "iso_8859-1"},
527 {28591, "iso_8859_1"},
528 {28592, "iso-8859-2"}, /* ISO 8859-2 Central European; Central European (ISO) */
529 {28592, "iso8859-2"}, /* ISO 8859-2 Central European; Central European (ISO) */
530 {28592, "iso_8859-2"},
531 {28592, "iso_8859_2"},
532 {28593, "iso-8859-3"}, /* ISO 8859-3 Latin 3 */
533 {28593, "iso8859-3"}, /* ISO 8859-3 Latin 3 */
534 {28593, "iso_8859-3"},
535 {28593, "iso_8859_3"},
536 {28594, "iso-8859-4"}, /* ISO 8859-4 Baltic */
537 {28594, "iso8859-4"}, /* ISO 8859-4 Baltic */
538 {28594, "iso_8859-4"},
539 {28594, "iso_8859_4"},
540 {28595, "iso-8859-5"}, /* ISO 8859-5 Cyrillic */
541 {28595, "iso8859-5"}, /* ISO 8859-5 Cyrillic */
542 {28595, "iso_8859-5"},
543 {28595, "iso_8859_5"},
544 {28596, "iso-8859-6"}, /* ISO 8859-6 Arabic */
545 {28596, "iso8859-6"}, /* ISO 8859-6 Arabic */
546 {28596, "iso_8859-6"},
547 {28596, "iso_8859_6"},
548 {28597, "iso-8859-7"}, /* ISO 8859-7 Greek */
549 {28597, "iso8859-7"}, /* ISO 8859-7 Greek */
550 {28597, "iso_8859-7"},
551 {28597, "iso_8859_7"},
552 {28598, "iso-8859-8"}, /* ISO 8859-8 Hebrew; Hebrew (ISO-Visual) */
553 {28598, "iso8859-8"}, /* ISO 8859-8 Hebrew; Hebrew (ISO-Visual) */
554 {28598, "iso_8859-8"},
555 {28598, "iso_8859_8"},
556 {28599, "iso-8859-9"}, /* ISO 8859-9 Turkish */
557 {28599, "iso8859-9"}, /* ISO 8859-9 Turkish */
558 {28599, "iso_8859-9"},
559 {28599, "iso_8859_9"},
560 {28603, "iso-8859-13"}, /* ISO 8859-13 Estonian */
561 {28603, "iso8859-13"}, /* ISO 8859-13 Estonian */
562 {28603, "iso_8859-13"},
563 {28603, "iso_8859_13"},
564 {28605, "iso-8859-15"}, /* ISO 8859-15 Latin 9 */
565 {28605, "iso8859-15"}, /* ISO 8859-15 Latin 9 */
566 {28605, "iso_8859-15"},
567 {28605, "iso_8859_15"},
568 {29001, "x-Europa"}, /* Europa 3 */
569 {38598, "iso-8859-8-i"}, /* ISO 8859-8 Hebrew; Hebrew (ISO-Logical) */
570 {38598, "iso8859-8-i"}, /* ISO 8859-8 Hebrew; Hebrew (ISO-Logical) */
571 {38598, "iso_8859-8-i"},
572 {38598, "iso_8859_8-i"},
573 {50220, "iso-2022-jp"}, /* ISO 2022 Japanese with no halfwidth Katakana; Japanese (JIS) */
574 {50221, "csISO2022JP"}, /* ISO 2022 Japanese with halfwidth Katakana; Japanese (JIS-Allow 1 byte Kana) */
575 {50222, "iso-2022-jp"}, /* ISO 2022 Japanese JIS X 0201-1989; Japanese (JIS-Allow 1 byte Kana - SO/SI) */
576 {50225, "iso-2022-kr"}, /* ISO 2022 Korean */
577 {50225, "iso2022-kr"}, /* ISO 2022 Korean */
578 {50227, "x-cp50227"}, /* ISO 2022 Simplified Chinese; Chinese Simplified (ISO 2022) */
579 /* 50229 ISO 2022 Traditional Chinese */
580 /* 50930 EBCDIC Japanese (Katakana) Extended */
581 /* 50931 EBCDIC US-Canada and Japanese */
582 /* 50933 EBCDIC Korean Extended and Korean */
583 /* 50935 EBCDIC Simplified Chinese Extended and Simplified Chinese */
584 /* 50936 EBCDIC Simplified Chinese */
585 /* 50937 EBCDIC US-Canada and Traditional Chinese */
586 /* 50939 EBCDIC Japanese (Latin) Extended and Japanese */
587 {51932, "euc-jp"}, /* EUC Japanese */
588 {51936, "EUC-CN"}, /* EUC Simplified Chinese; Chinese Simplified (EUC) */
589 {51949, "euc-kr"}, /* EUC Korean */
590 /* 51950 EUC Traditional Chinese */
591 {52936, "hz-gb-2312"}, /* HZ-GB2312 Simplified Chinese; Chinese Simplified (HZ) */
592 {54936, "GB18030"}, /* Windows XP and later: GB18030 Simplified Chinese (4 byte); Chinese Simplified (GB18030) */
593 {57002, "x-iscii-de"}, /* ISCII Devanagari */
594 {57003, "x-iscii-be"}, /* ISCII Bengali */
595 {57004, "x-iscii-ta"}, /* ISCII Tamil */
596 {57005, "x-iscii-te"}, /* ISCII Telugu */
597 {57006, "x-iscii-as"}, /* ISCII Assamese */
598 {57007, "x-iscii-or"}, /* ISCII Oriya */
599 {57008, "x-iscii-ka"}, /* ISCII Kannada */
600 {57009, "x-iscii-ma"}, /* ISCII Malayalam */
601 {57010, "x-iscii-gu"}, /* ISCII Gujarati */
602 {57011, "x-iscii-pa"}, /* ISCII Punjabi */
603
604 {0, NULL}
605 };
606
607 /*
608 * SJIS SHIFTJIS table CP932 table
609 * ---- --------------------------- --------------------------------
610 * 5C U+00A5 YEN SIGN U+005C REVERSE SOLIDUS
611 * 7E U+203E OVERLINE U+007E TILDE
612 * 815C U+2014 EM DASH U+2015 HORIZONTAL BAR
613 * 815F U+005C REVERSE SOLIDUS U+FF3C FULLWIDTH REVERSE SOLIDUS
614 * 8160 U+301C WAVE DASH U+FF5E FULLWIDTH TILDE
615 * 8161 U+2016 DOUBLE VERTICAL LINE U+2225 PARALLEL TO
616 * 817C U+2212 MINUS SIGN U+FF0D FULLWIDTH HYPHEN-MINUS
617 * 8191 U+00A2 CENT SIGN U+FFE0 FULLWIDTH CENT SIGN
618 * 8192 U+00A3 POUND SIGN U+FFE1 FULLWIDTH POUND SIGN
619 * 81CA U+00AC NOT SIGN U+FFE2 FULLWIDTH NOT SIGN
620 *
621 * EUC-JP and ISO-2022-JP should be compatible with CP932.
622 *
623 * Kernel and MLang have different Unicode mapping table. Make sure
624 * which API is used.
625 */
626 static compat_t cp932_compat[] = {
627 {0x00A5, 0x005C, COMPAT_OUT},
628 {0x203E, 0x007E, COMPAT_OUT},
629 {0x2014, 0x2015, COMPAT_OUT},
630 {0x301C, 0xFF5E, COMPAT_OUT},
631 {0x2016, 0x2225, COMPAT_OUT},
632 {0x2212, 0xFF0D, COMPAT_OUT},
633 {0x00A2, 0xFFE0, COMPAT_OUT},
634 {0x00A3, 0xFFE1, COMPAT_OUT},
635 {0x00AC, 0xFFE2, COMPAT_OUT},
636 {0, 0, 0}
637 };
638
639 static compat_t cp20932_compat[] = {
640 {0x00A5, 0x005C, COMPAT_OUT},
641 {0x203E, 0x007E, COMPAT_OUT},
642 {0x2014, 0x2015, COMPAT_OUT},
643 {0xFF5E, 0x301C, COMPAT_OUT|COMPAT_IN},
644 {0x2225, 0x2016, COMPAT_OUT|COMPAT_IN},
645 {0xFF0D, 0x2212, COMPAT_OUT|COMPAT_IN},
646 {0xFFE0, 0x00A2, COMPAT_OUT|COMPAT_IN},
647 {0xFFE1, 0x00A3, COMPAT_OUT|COMPAT_IN},
648 {0xFFE2, 0x00AC, COMPAT_OUT|COMPAT_IN},
649 {0, 0, 0}
650 };
651
652 static compat_t *cp51932_compat = cp932_compat;
653
654 /* cp20932_compat for kernel. cp932_compat for mlang. */
655 static compat_t *cp5022x_compat = cp932_compat;
656
657 typedef HRESULT (WINAPI *CONVERTINETSTRING)(
658 LPDWORD lpdwMode,
659 DWORD dwSrcEncoding,
660 DWORD dwDstEncoding,
661 LPCSTR lpSrcStr,
662 LPINT lpnSrcSize,
663 LPBYTE lpDstStr,
664 LPINT lpnDstSize
665 );
666 typedef HRESULT (WINAPI *CONVERTINETMULTIBYTETOUNICODE)(
667 LPDWORD lpdwMode,
668 DWORD dwSrcEncoding,
669 LPCSTR lpSrcStr,
670 LPINT lpnMultiCharCount,
671 LPWSTR lpDstStr,
672 LPINT lpnWideCharCount
673 );
674 typedef HRESULT (WINAPI *CONVERTINETUNICODETOMULTIBYTE)(
675 LPDWORD lpdwMode,
676 DWORD dwEncoding,
677 LPCWSTR lpSrcStr,
678 LPINT lpnWideCharCount,
679 LPSTR lpDstStr,
680 LPINT lpnMultiCharCount
681 );
682 typedef HRESULT (WINAPI *ISCONVERTINETSTRINGAVAILABLE)(
683 DWORD dwSrcEncoding,
684 DWORD dwDstEncoding
685 );
686 typedef HRESULT (WINAPI *LCIDTORFC1766A)(
687 LCID Locale,
688 LPSTR pszRfc1766,
689 int nChar
690 );
691 typedef HRESULT (WINAPI *LCIDTORFC1766W)(
692 LCID Locale,
693 LPWSTR pszRfc1766,
694 int nChar
695 );
696 typedef HRESULT (WINAPI *RFC1766TOLCIDA)(
697 LCID *pLocale,
698 LPSTR pszRfc1766
699 );
700 typedef HRESULT (WINAPI *RFC1766TOLCIDW)(
701 LCID *pLocale,
702 LPWSTR pszRfc1766
703 );
704 static CONVERTINETSTRING ConvertINetString;
705 static CONVERTINETMULTIBYTETOUNICODE ConvertINetMultiByteToUnicode;
706 static CONVERTINETUNICODETOMULTIBYTE ConvertINetUnicodeToMultiByte;
707 static ISCONVERTINETSTRINGAVAILABLE IsConvertINetStringAvailable;
708 static LCIDTORFC1766A LcidToRfc1766A;
709 static RFC1766TOLCIDA Rfc1766ToLcidA;
710
711 static int
load_mlang(void)712 load_mlang(void)
713 {
714 HMODULE h;
715 if (ConvertINetString != NULL)
716 return TRUE;
717 h = LoadLibrary(TEXT("mlang.dll"));
718 if (!h)
719 return FALSE;
720 ConvertINetString = (CONVERTINETSTRING)GetProcAddressA(h, "ConvertINetString");
721 ConvertINetMultiByteToUnicode = (CONVERTINETMULTIBYTETOUNICODE)GetProcAddressA(h, "ConvertINetMultiByteToUnicode");
722 ConvertINetUnicodeToMultiByte = (CONVERTINETUNICODETOMULTIBYTE)GetProcAddressA(h, "ConvertINetUnicodeToMultiByte");
723 IsConvertINetStringAvailable = (ISCONVERTINETSTRINGAVAILABLE)GetProcAddressA(h, "IsConvertINetStringAvailable");
724 LcidToRfc1766A = (LCIDTORFC1766A)GetProcAddressA(h, "LcidToRfc1766A");
725 Rfc1766ToLcidA = (RFC1766TOLCIDA)GetProcAddressA(h, "Rfc1766ToLcidA");
726 return TRUE;
727 }
728
729 iconv_t
iconv_open(const char * tocode,const char * fromcode)730 iconv_open(const char *tocode, const char *fromcode)
731 {
732 rec_iconv_t *cd;
733
734 cd = (rec_iconv_t *)calloc(1, sizeof(rec_iconv_t));
735 if (cd == NULL)
736 return (iconv_t)(-1);
737
738 #if defined(USE_LIBICONV_DLL)
739 errno = 0;
740 if (libiconv_iconv_open(cd, tocode, fromcode))
741 return (iconv_t)cd;
742 #endif
743
744 /* reset the errno to prevent reporting wrong error code.
745 * 0 for unsorted error. */
746 errno = 0;
747 if (win_iconv_open(cd, tocode, fromcode))
748 return (iconv_t)cd;
749
750 free(cd);
751
752 return (iconv_t)(-1);
753 }
754
755 int
iconv_close(iconv_t _cd)756 iconv_close(iconv_t _cd)
757 {
758 rec_iconv_t *cd = (rec_iconv_t *)_cd;
759 int r = cd->iconv_close(cd->cd);
760 int e = *(cd->_errno());
761 #if defined(USE_LIBICONV_DLL)
762 if (cd->hlibiconv != NULL)
763 FreeLibrary(cd->hlibiconv);
764 #endif
765 free(cd);
766 errno = e;
767 return r;
768 }
769
770 size_t
iconv(iconv_t _cd,const char ** inbuf,size_t * inbytesleft,char ** outbuf,size_t * outbytesleft)771 iconv(iconv_t _cd, const char **inbuf, size_t *inbytesleft, char **outbuf, size_t *outbytesleft)
772 {
773 rec_iconv_t *cd = (rec_iconv_t *)_cd;
774 size_t r = cd->iconv(cd->cd, inbuf, inbytesleft, outbuf, outbytesleft);
775 errno = *(cd->_errno());
776 return r;
777 }
778
779 static int
win_iconv_open(rec_iconv_t * cd,const char * tocode,const char * fromcode)780 win_iconv_open(rec_iconv_t *cd, const char *tocode, const char *fromcode)
781 {
782 if (!make_csconv(fromcode, &cd->from) || !make_csconv(tocode, &cd->to))
783 return FALSE;
784 cd->iconv_close = win_iconv_close;
785 cd->iconv = win_iconv;
786 cd->_errno = _errno;
787 cd->cd = (iconv_t)cd;
788 return TRUE;
789 }
790
791 static int
win_iconv_close(iconv_t cd UNUSED)792 win_iconv_close(iconv_t cd UNUSED)
793 {
794 return 0;
795 }
796
797 static size_t
win_iconv(iconv_t _cd,const char ** inbuf,size_t * inbytesleft,char ** outbuf,size_t * outbytesleft)798 win_iconv(iconv_t _cd, const char **inbuf, size_t *inbytesleft, char **outbuf, size_t *outbytesleft)
799 {
800 rec_iconv_t *cd = (rec_iconv_t *)_cd;
801 ushort wbuf[MB_CHAR_MAX]; /* enough room for one character */
802 int insize;
803 int outsize;
804 int wsize;
805 DWORD frommode;
806 DWORD tomode;
807 uint wc;
808 compat_t *cp;
809 int i;
810
811 if (inbuf == NULL || *inbuf == NULL)
812 {
813 if (outbuf != NULL && *outbuf != NULL && cd->to.flush != NULL)
814 {
815 tomode = cd->to.mode;
816 outsize = cd->to.flush(&cd->to, (uchar *)*outbuf, *outbytesleft);
817 if (outsize == -1)
818 {
819 if ((cd->to.flags & FLAG_IGNORE) && errno != E2BIG)
820 {
821 outsize = 0;
822 }
823 else
824 {
825 cd->to.mode = tomode;
826 return (size_t)(-1);
827 }
828 }
829 *outbuf += outsize;
830 *outbytesleft -= outsize;
831 }
832 cd->from.mode = 0;
833 cd->to.mode = 0;
834 return 0;
835 }
836
837 while (*inbytesleft != 0)
838 {
839 frommode = cd->from.mode;
840 tomode = cd->to.mode;
841 wsize = MB_CHAR_MAX;
842
843 insize = cd->from.mbtowc(&cd->from, (const uchar *)*inbuf, *inbytesleft, wbuf, &wsize);
844 if (insize == -1)
845 {
846 if (cd->to.flags & FLAG_IGNORE)
847 {
848 cd->from.mode = frommode;
849 insize = 1;
850 wsize = 0;
851 }
852 else
853 {
854 cd->from.mode = frommode;
855 return (size_t)(-1);
856 }
857 }
858
859 if (wsize == 0)
860 {
861 *inbuf += insize;
862 *inbytesleft -= insize;
863 continue;
864 }
865
866 if (cd->from.compat != NULL)
867 {
868 wc = utf16_to_ucs4(wbuf);
869 cp = cd->from.compat;
870 for (i = 0; cp[i].in != 0; ++i)
871 {
872 if ((cp[i].flag & COMPAT_IN) && cp[i].out == wc)
873 {
874 ucs4_to_utf16(cp[i].in, wbuf, &wsize);
875 break;
876 }
877 }
878 }
879
880 if (cd->to.compat != NULL)
881 {
882 wc = utf16_to_ucs4(wbuf);
883 cp = cd->to.compat;
884 for (i = 0; cp[i].in != 0; ++i)
885 {
886 if ((cp[i].flag & COMPAT_OUT) && cp[i].in == wc)
887 {
888 ucs4_to_utf16(cp[i].out, wbuf, &wsize);
889 break;
890 }
891 }
892 }
893
894 outsize = cd->to.wctomb(&cd->to, wbuf, wsize, (uchar *)*outbuf, *outbytesleft);
895 if (outsize == -1)
896 {
897 if ((cd->to.flags & FLAG_IGNORE) && errno != E2BIG)
898 {
899 cd->to.mode = tomode;
900 outsize = 0;
901 }
902 else
903 {
904 cd->from.mode = frommode;
905 cd->to.mode = tomode;
906 return (size_t)(-1);
907 }
908 }
909
910 *inbuf += insize;
911 *outbuf += outsize;
912 *inbytesleft -= insize;
913 *outbytesleft -= outsize;
914 }
915
916 return 0;
917 }
918
919 static int
make_csconv(const char * _name,csconv_t * cv)920 make_csconv(const char *_name, csconv_t *cv)
921 {
922 CPINFO cpinfo;
923 int use_compat = TRUE;
924 int flag = 0;
925 char *name;
926 char *p;
927
928 name = xstrndup(_name, strlen(_name));
929 if (name == NULL)
930 return FALSE;
931
932 /* check for option "enc_name//opt1//opt2" */
933 while ((p = strrstr(name, "//")) != NULL)
934 {
935 if (_stricmp(p + 2, "nocompat") == 0)
936 use_compat = FALSE;
937 else if (_stricmp(p + 2, "translit") == 0)
938 flag |= FLAG_TRANSLIT;
939 else if (_stricmp(p + 2, "ignore") == 0)
940 flag |= FLAG_IGNORE;
941 *p = 0;
942 }
943
944 cv->mode = 0;
945 cv->flags = flag;
946 cv->mblen = NULL;
947 cv->flush = NULL;
948 cv->compat = NULL;
949 cv->codepage = name_to_codepage(name);
950 if (cv->codepage == 1200 || cv->codepage == 1201)
951 {
952 cv->mbtowc = utf16_mbtowc;
953 cv->wctomb = utf16_wctomb;
954 if (_stricmp(name, "UTF-16") == 0 || _stricmp(name, "UTF16") == 0 ||
955 _stricmp(name, "UCS-2") == 0 || _stricmp(name, "UCS2") == 0)
956 cv->flags |= FLAG_USE_BOM;
957 }
958 else if (cv->codepage == 12000 || cv->codepage == 12001)
959 {
960 cv->mbtowc = utf32_mbtowc;
961 cv->wctomb = utf32_wctomb;
962 if (_stricmp(name, "UTF-32") == 0 || _stricmp(name, "UTF32") == 0 ||
963 _stricmp(name, "UCS-4") == 0 || _stricmp(name, "UCS4") == 0)
964 cv->flags |= FLAG_USE_BOM;
965 }
966 else if (cv->codepage == 65001)
967 {
968 cv->mbtowc = kernel_mbtowc;
969 cv->wctomb = kernel_wctomb;
970 cv->mblen = utf8_mblen;
971 }
972 else if ((cv->codepage == 50220 || cv->codepage == 50221 || cv->codepage == 50222) && load_mlang())
973 {
974 cv->mbtowc = iso2022jp_mbtowc;
975 cv->wctomb = iso2022jp_wctomb;
976 cv->flush = iso2022jp_flush;
977 }
978 else if (cv->codepage == 51932 && load_mlang())
979 {
980 cv->mbtowc = mlang_mbtowc;
981 cv->wctomb = mlang_wctomb;
982 cv->mblen = eucjp_mblen;
983 }
984 else if (IsValidCodePage(cv->codepage)
985 && GetCPInfo(cv->codepage, &cpinfo) != 0)
986 {
987 cv->mbtowc = kernel_mbtowc;
988 cv->wctomb = kernel_wctomb;
989 if (cpinfo.MaxCharSize == 1)
990 cv->mblen = sbcs_mblen;
991 else if (cpinfo.MaxCharSize == 2)
992 cv->mblen = dbcs_mblen;
993 else
994 cv->mblen = mbcs_mblen;
995 }
996 else
997 {
998 /* not supported */
999 free(name);
1000 errno = EINVAL;
1001 return FALSE;
1002 }
1003
1004 if (use_compat)
1005 {
1006 switch (cv->codepage)
1007 {
1008 case 932: cv->compat = cp932_compat; break;
1009 case 20932: cv->compat = cp20932_compat; break;
1010 case 51932: cv->compat = cp51932_compat; break;
1011 case 50220: case 50221: case 50222: cv->compat = cp5022x_compat; break;
1012 }
1013 }
1014
1015 free(name);
1016
1017 return TRUE;
1018 }
1019
1020 static int
name_to_codepage(const char * name)1021 name_to_codepage(const char *name)
1022 {
1023 int i;
1024
1025 if (*name == '\0' ||
1026 strcmp(name, "char") == 0)
1027 return GetACP();
1028 else if (strcmp(name, "wchar_t") == 0)
1029 return 1200;
1030 else if (_strnicmp(name, "cp", 2) == 0)
1031 return atoi(name + 2); /* CP123 */
1032 else if ('0' <= name[0] && name[0] <= '9')
1033 return atoi(name); /* 123 */
1034 else if (_strnicmp(name, "xx", 2) == 0)
1035 return atoi(name + 2); /* XX123 for debug */
1036
1037 for (i = 0; codepage_alias[i].name != NULL; ++i)
1038 if (_stricmp(name, codepage_alias[i].name) == 0)
1039 return codepage_alias[i].codepage;
1040 return -1;
1041 }
1042
1043 /*
1044 * http://www.faqs.org/rfcs/rfc2781.html
1045 */
1046 static uint
utf16_to_ucs4(const ushort * wbuf)1047 utf16_to_ucs4(const ushort *wbuf)
1048 {
1049 uint wc = wbuf[0];
1050 if (0xD800 <= wbuf[0] && wbuf[0] <= 0xDBFF)
1051 wc = ((wbuf[0] & 0x3FF) << 10) + (wbuf[1] & 0x3FF) + 0x10000;
1052 return wc;
1053 }
1054
1055 static void
ucs4_to_utf16(uint wc,ushort * wbuf,int * wbufsize)1056 ucs4_to_utf16(uint wc, ushort *wbuf, int *wbufsize)
1057 {
1058 if (wc < 0x10000)
1059 {
1060 wbuf[0] = wc;
1061 *wbufsize = 1;
1062 }
1063 else
1064 {
1065 wc -= 0x10000;
1066 wbuf[0] = 0xD800 | ((wc >> 10) & 0x3FF);
1067 wbuf[1] = 0xDC00 | (wc & 0x3FF);
1068 *wbufsize = 2;
1069 }
1070 }
1071
1072 /*
1073 * Check if codepage is one of those for which the dwFlags parameter
1074 * to MultiByteToWideChar() must be zero. Return zero or
1075 * MB_ERR_INVALID_CHARS. The docs in Platform SDK for for Windows
1076 * Server 2003 R2 claims that also codepage 65001 is one of these, but
1077 * that doesn't seem to be the case. The MSDN docs for MSVS2008 leave
1078 * out 65001 (UTF-8), and that indeed seems to be the case on XP, it
1079 * works fine to pass MB_ERR_INVALID_CHARS in dwFlags when converting
1080 * from UTF-8.
1081 */
1082 static int
mbtowc_flags(int codepage)1083 mbtowc_flags(int codepage)
1084 {
1085 return (codepage == 50220 || codepage == 50221 ||
1086 codepage == 50222 || codepage == 50225 ||
1087 codepage == 50227 || codepage == 50229 ||
1088 codepage == 52936 || codepage == 54936 ||
1089 (codepage >= 57002 && codepage <= 57011) ||
1090 codepage == 65000 || codepage == 42) ? 0 : MB_ERR_INVALID_CHARS;
1091 }
1092
1093 /*
1094 * Check if codepage is one those for which the lpUsedDefaultChar
1095 * parameter to WideCharToMultiByte() must be NULL. The docs in
1096 * Platform SDK for Windows Server 2003 R2 claims that this is the
1097 * list below, while the MSDN docs for MSVS2008 claim that it is only
1098 * for 65000 (UTF-7) and 65001 (UTF-8). This time the earlier Platform
1099 * SDK seems to be correct, at least for XP.
1100 */
1101 static int
must_use_null_useddefaultchar(int codepage)1102 must_use_null_useddefaultchar(int codepage)
1103 {
1104 return (codepage == 65000 || codepage == 65001 ||
1105 codepage == 50220 || codepage == 50221 ||
1106 codepage == 50222 || codepage == 50225 ||
1107 codepage == 50227 || codepage == 50229 ||
1108 codepage == 52936 || codepage == 54936 ||
1109 (codepage >= 57002 && codepage <= 57011) ||
1110 codepage == 42);
1111 }
1112
1113 static char *
strrstr(const char * str,const char * token)1114 strrstr(const char *str, const char *token)
1115 {
1116 int len = strlen(token);
1117 const char *p = str + strlen(str);
1118
1119 while (str <= --p)
1120 if (p[0] == token[0] && strncmp(p, token, len) == 0)
1121 return (char *)p;
1122 return NULL;
1123 }
1124
1125 static char *
xstrndup(const char * s,size_t n)1126 xstrndup(const char *s, size_t n)
1127 {
1128 char *p;
1129
1130 p = (char *)malloc(n + 1);
1131 if (p == NULL)
1132 return NULL;
1133 memcpy(p, s, n);
1134 p[n] = '\0';
1135 return p;
1136 }
1137
1138 static int
seterror(int err)1139 seterror(int err)
1140 {
1141 errno = err;
1142 return -1;
1143 }
1144
1145 #if defined(USE_LIBICONV_DLL)
1146 static int
libiconv_iconv_open(rec_iconv_t * cd,const char * tocode,const char * fromcode)1147 libiconv_iconv_open(rec_iconv_t *cd, const char *tocode, const char *fromcode)
1148 {
1149 HMODULE hlibiconv = NULL;
1150 char *dllname;
1151 const char *p;
1152 const char *e;
1153 f_iconv_open _iconv_open;
1154
1155 /*
1156 * always try to load dll, so that we can switch dll in runtime.
1157 */
1158
1159 /* XXX: getenv() can't get variable set by SetEnvironmentVariable() */
1160 p = getenv("WINICONV_LIBICONV_DLL");
1161 if (p == NULL)
1162 p = DEFAULT_LIBICONV_DLL;
1163 /* parse comma separated value */
1164 for ( ; *p != 0; p = (*e == ',') ? e + 1 : e)
1165 {
1166 e = strchr(p, ',');
1167 if (p == e)
1168 continue;
1169 else if (e == NULL)
1170 e = p + strlen(p);
1171 dllname = xstrndup(p, e - p);
1172 if (dllname == NULL)
1173 return FALSE;
1174 hlibiconv = LoadLibraryA(dllname);
1175 free(dllname);
1176 if (hlibiconv != NULL)
1177 {
1178 if (hlibiconv == hwiniconv)
1179 {
1180 FreeLibrary(hlibiconv);
1181 hlibiconv = NULL;
1182 continue;
1183 }
1184 break;
1185 }
1186 }
1187
1188 if (hlibiconv == NULL)
1189 goto failed;
1190
1191 _iconv_open = (f_iconv_open)GetProcAddressA(hlibiconv, "libiconv_open");
1192 if (_iconv_open == NULL)
1193 _iconv_open = (f_iconv_open)GetProcAddressA(hlibiconv, "iconv_open");
1194 cd->iconv_close = (f_iconv_close)GetProcAddressA(hlibiconv, "libiconv_close");
1195 if (cd->iconv_close == NULL)
1196 cd->iconv_close = (f_iconv_close)GetProcAddressA(hlibiconv, "iconv_close");
1197 cd->iconv = (f_iconv)GetProcAddressA(hlibiconv, "libiconv");
1198 if (cd->iconv == NULL)
1199 cd->iconv = (f_iconv)GetProcAddressA(hlibiconv, "iconv");
1200 cd->_errno = (f_errno)find_imported_function(hlibiconv, "_errno");
1201 if (_iconv_open == NULL || cd->iconv_close == NULL
1202 || cd->iconv == NULL || cd->_errno == NULL)
1203 goto failed;
1204
1205 cd->cd = _iconv_open(tocode, fromcode);
1206 if (cd->cd == (iconv_t)(-1))
1207 goto failed;
1208
1209 cd->hlibiconv = hlibiconv;
1210 return TRUE;
1211
1212 failed:
1213 if (hlibiconv != NULL)
1214 FreeLibrary(hlibiconv);
1215 return FALSE;
1216 }
1217
1218 /*
1219 * Reference:
1220 * http://forums.belution.com/ja/vc/000/234/78s.shtml
1221 * http://nienie.com/~masapico/api_ImageDirectoryEntryToData.html
1222 *
1223 * The formal way is
1224 * imagehlp.h or dbghelp.h
1225 * imagehlp.lib or dbghelp.lib
1226 * ImageDirectoryEntryToData()
1227 */
1228 #define TO_DOS_HEADER(base) ((PIMAGE_DOS_HEADER)(base))
1229 #define TO_NT_HEADERS(base) ((PIMAGE_NT_HEADERS)((LPBYTE)(base) + TO_DOS_HEADER(base)->e_lfanew))
1230 static PVOID
MyImageDirectoryEntryToData(LPVOID Base,BOOLEAN MappedAsImage,USHORT DirectoryEntry,PULONG Size)1231 MyImageDirectoryEntryToData(LPVOID Base, BOOLEAN MappedAsImage, USHORT DirectoryEntry, PULONG Size)
1232 {
1233 /* TODO: MappedAsImage? */
1234 PIMAGE_DATA_DIRECTORY p;
1235 p = TO_NT_HEADERS(Base)->OptionalHeader.DataDirectory + DirectoryEntry;
1236 if (p->VirtualAddress == 0) {
1237 *Size = 0;
1238 return NULL;
1239 }
1240 *Size = p->Size;
1241 return (PVOID)((LPBYTE)Base + p->VirtualAddress);
1242 }
1243
1244 static FARPROC
find_imported_function(HMODULE hModule,const char * funcname)1245 find_imported_function(HMODULE hModule, const char *funcname)
1246 {
1247 DWORD_PTR Base;
1248 ULONG Size;
1249 PIMAGE_IMPORT_DESCRIPTOR Imp;
1250 PIMAGE_THUNK_DATA Address; /* Import Address Table */
1251 PIMAGE_THUNK_DATA Name; /* Import Name Table */
1252 PIMAGE_IMPORT_BY_NAME ImpName;
1253
1254 Base = (DWORD_PTR)hModule;
1255 Imp = (PIMAGE_IMPORT_DESCRIPTOR)MyImageDirectoryEntryToData(
1256 (LPVOID)Base,
1257 TRUE,
1258 IMAGE_DIRECTORY_ENTRY_IMPORT,
1259 &Size);
1260 if (Imp == NULL)
1261 return NULL;
1262 for ( ; Imp->OriginalFirstThunk != 0; ++Imp)
1263 {
1264 Address = (PIMAGE_THUNK_DATA)(Base + Imp->FirstThunk);
1265 Name = (PIMAGE_THUNK_DATA)(Base + Imp->OriginalFirstThunk);
1266 for ( ; Name->u1.Ordinal != 0; ++Name, ++Address)
1267 {
1268 if (!IMAGE_SNAP_BY_ORDINAL(Name->u1.Ordinal))
1269 {
1270 ImpName = (PIMAGE_IMPORT_BY_NAME)
1271 (Base + (DWORD_PTR)Name->u1.AddressOfData);
1272 if (strcmp((char *)ImpName->Name, funcname) == 0)
1273 return (FARPROC)Address->u1.Function;
1274 }
1275 }
1276 }
1277 return NULL;
1278 }
1279 #endif
1280
1281 static int
sbcs_mblen(csconv_t * cv UNUSED,const uchar * buf UNUSED,int bufsize UNUSED)1282 sbcs_mblen(csconv_t *cv UNUSED, const uchar *buf UNUSED, int bufsize UNUSED)
1283 {
1284 return 1;
1285 }
1286
1287 static int
dbcs_mblen(csconv_t * cv,const uchar * buf,int bufsize)1288 dbcs_mblen(csconv_t *cv, const uchar *buf, int bufsize)
1289 {
1290 int len = IsDBCSLeadByteEx(cv->codepage, buf[0]) ? 2 : 1;
1291 if (bufsize < len)
1292 return seterror(EINVAL);
1293 return len;
1294 }
1295
1296 static int
mbcs_mblen(csconv_t * cv,const uchar * buf,int bufsize)1297 mbcs_mblen(csconv_t *cv, const uchar *buf, int bufsize)
1298 {
1299 int len = 0;
1300
1301 if (cv->codepage == 54936) {
1302 if (buf[0] <= 0x7F) len = 1;
1303 else if (buf[0] >= 0x81 && buf[0] <= 0xFE &&
1304 bufsize >= 2 &&
1305 ((buf[1] >= 0x40 && buf[1] <= 0x7E) ||
1306 (buf[1] >= 0x80 && buf[1] <= 0xFE))) len = 2;
1307 else if (buf[0] >= 0x81 && buf[0] <= 0xFE &&
1308 bufsize >= 4 &&
1309 buf[1] >= 0x30 && buf[1] <= 0x39) len = 4;
1310 else
1311 return seterror(EINVAL);
1312 return len;
1313 }
1314 else
1315 return seterror(EINVAL);
1316 }
1317
1318 static int
utf8_mblen(csconv_t * cv UNUSED,const uchar * buf,int bufsize)1319 utf8_mblen(csconv_t *cv UNUSED, const uchar *buf, int bufsize)
1320 {
1321 int len = 0;
1322
1323 if (buf[0] < 0x80) len = 1;
1324 else if ((buf[0] & 0xE0) == 0xC0) len = 2;
1325 else if ((buf[0] & 0xF0) == 0xE0) len = 3;
1326 else if ((buf[0] & 0xF8) == 0xF0) len = 4;
1327 else if ((buf[0] & 0xFC) == 0xF8) len = 5;
1328 else if ((buf[0] & 0xFE) == 0xFC) len = 6;
1329
1330 if (len == 0)
1331 return seterror(EILSEQ);
1332 else if (bufsize < len)
1333 return seterror(EINVAL);
1334 return len;
1335 }
1336
1337 static int
eucjp_mblen(csconv_t * cv UNUSED,const uchar * buf,int bufsize)1338 eucjp_mblen(csconv_t *cv UNUSED, const uchar *buf, int bufsize)
1339 {
1340 if (buf[0] < 0x80) /* ASCII */
1341 return 1;
1342 else if (buf[0] == 0x8E) /* JIS X 0201 */
1343 {
1344 if (bufsize < 2)
1345 return seterror(EINVAL);
1346 else if (!(0xA1 <= buf[1] && buf[1] <= 0xDF))
1347 return seterror(EILSEQ);
1348 return 2;
1349 }
1350 else if (buf[0] == 0x8F) /* JIS X 0212 */
1351 {
1352 if (bufsize < 3)
1353 return seterror(EINVAL);
1354 else if (!(0xA1 <= buf[1] && buf[1] <= 0xFE)
1355 || !(0xA1 <= buf[2] && buf[2] <= 0xFE))
1356 return seterror(EILSEQ);
1357 return 3;
1358 }
1359 else /* JIS X 0208 */
1360 {
1361 if (bufsize < 2)
1362 return seterror(EINVAL);
1363 else if (!(0xA1 <= buf[0] && buf[0] <= 0xFE)
1364 || !(0xA1 <= buf[1] && buf[1] <= 0xFE))
1365 return seterror(EILSEQ);
1366 return 2;
1367 }
1368 }
1369
1370 static int
kernel_mbtowc(csconv_t * cv,const uchar * buf,int bufsize,ushort * wbuf,int * wbufsize)1371 kernel_mbtowc(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize)
1372 {
1373 int len;
1374
1375 len = cv->mblen(cv, buf, bufsize);
1376 if (len == -1)
1377 return -1;
1378 *wbufsize = MultiByteToWideChar(cv->codepage, mbtowc_flags (cv->codepage),
1379 (const char *)buf, len, (wchar_t *)wbuf, *wbufsize);
1380 if (*wbufsize == 0)
1381 return seterror(EILSEQ);
1382 return len;
1383 }
1384
1385 static int
kernel_wctomb(csconv_t * cv,ushort * wbuf,int wbufsize,uchar * buf,int bufsize)1386 kernel_wctomb(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize)
1387 {
1388 BOOL usedDefaultChar = 0;
1389 BOOL *p = NULL;
1390 int flags = 0;
1391 int len;
1392
1393 if (bufsize == 0)
1394 return seterror(E2BIG);
1395 if (!must_use_null_useddefaultchar(cv->codepage))
1396 {
1397 p = &usedDefaultChar;
1398 #ifdef WC_NO_BEST_FIT_CHARS
1399 if (!(cv->flags & FLAG_TRANSLIT))
1400 flags |= WC_NO_BEST_FIT_CHARS;
1401 #endif
1402 }
1403 len = WideCharToMultiByte(cv->codepage, flags,
1404 (const wchar_t *)wbuf, wbufsize, (char *)buf, bufsize, NULL, p);
1405 if (len == 0)
1406 {
1407 if (GetLastError() == ERROR_INSUFFICIENT_BUFFER)
1408 return seterror(E2BIG);
1409 return seterror(EILSEQ);
1410 }
1411 else if (usedDefaultChar && !(cv->flags & FLAG_TRANSLIT))
1412 return seterror(EILSEQ);
1413 else if (cv->mblen(cv, buf, len) != len) /* validate result */
1414 return seterror(EILSEQ);
1415 return len;
1416 }
1417
1418 /*
1419 * It seems that the mode (cv->mode) is fixnum.
1420 * For example, when converting iso-2022-jp(cp50221) to unicode:
1421 * in ascii sequence: mode=0xC42C0000
1422 * in jisx0208 sequence: mode=0xC42C0001
1423 * "C42C" is same for each convert session.
1424 * It should be: ((codepage-1)<<16)|state
1425 */
1426 static int
mlang_mbtowc(csconv_t * cv,const uchar * buf,int bufsize,ushort * wbuf,int * wbufsize)1427 mlang_mbtowc(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize)
1428 {
1429 int len;
1430 int insize;
1431 HRESULT hr;
1432
1433 len = cv->mblen(cv, buf, bufsize);
1434 if (len == -1)
1435 return -1;
1436 insize = len;
1437 hr = ConvertINetMultiByteToUnicode(&cv->mode, cv->codepage,
1438 (const char *)buf, &insize, (wchar_t *)wbuf, wbufsize);
1439 if (hr != S_OK || insize != len)
1440 return seterror(EILSEQ);
1441 return len;
1442 }
1443
1444 static int
mlang_wctomb(csconv_t * cv,ushort * wbuf,int wbufsize,uchar * buf,int bufsize)1445 mlang_wctomb(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize)
1446 {
1447 char tmpbuf[MB_CHAR_MAX]; /* enough room for one character */
1448 int tmpsize = MB_CHAR_MAX;
1449 int insize = wbufsize;
1450 HRESULT hr;
1451
1452 hr = ConvertINetUnicodeToMultiByte(&cv->mode, cv->codepage,
1453 (const wchar_t *)wbuf, &wbufsize, tmpbuf, &tmpsize);
1454 if (hr != S_OK || insize != wbufsize)
1455 return seterror(EILSEQ);
1456 else if (bufsize < tmpsize)
1457 return seterror(E2BIG);
1458 else if (cv->mblen(cv, (uchar *)tmpbuf, tmpsize) != tmpsize)
1459 return seterror(EILSEQ);
1460 memcpy(buf, tmpbuf, tmpsize);
1461 return tmpsize;
1462 }
1463
1464 static int
utf16_mbtowc(csconv_t * cv,const uchar * buf,int bufsize,ushort * wbuf,int * wbufsize)1465 utf16_mbtowc(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize)
1466 {
1467 int codepage = cv->codepage;
1468
1469 /* swap endian: 1200 <-> 1201 */
1470 if (cv->mode & UNICODE_MODE_SWAPPED)
1471 codepage ^= 1;
1472
1473 if (bufsize < 2)
1474 return seterror(EINVAL);
1475 if (codepage == 1200) /* little endian */
1476 wbuf[0] = (buf[1] << 8) | buf[0];
1477 else if (codepage == 1201) /* big endian */
1478 wbuf[0] = (buf[0] << 8) | buf[1];
1479
1480 if ((cv->flags & FLAG_USE_BOM) && !(cv->mode & UNICODE_MODE_BOM_DONE))
1481 {
1482 cv->mode |= UNICODE_MODE_BOM_DONE;
1483 if (wbuf[0] == 0xFFFE)
1484 {
1485 cv->mode |= UNICODE_MODE_SWAPPED;
1486 *wbufsize = 0;
1487 return 2;
1488 }
1489 else if (wbuf[0] == 0xFEFF)
1490 {
1491 *wbufsize = 0;
1492 return 2;
1493 }
1494 }
1495
1496 if (0xDC00 <= wbuf[0] && wbuf[0] <= 0xDFFF)
1497 return seterror(EILSEQ);
1498 if (0xD800 <= wbuf[0] && wbuf[0] <= 0xDBFF)
1499 {
1500 if (bufsize < 4)
1501 return seterror(EINVAL);
1502 if (codepage == 1200) /* little endian */
1503 wbuf[1] = (buf[3] << 8) | buf[2];
1504 else if (codepage == 1201) /* big endian */
1505 wbuf[1] = (buf[2] << 8) | buf[3];
1506 if (!(0xDC00 <= wbuf[1] && wbuf[1] <= 0xDFFF))
1507 return seterror(EILSEQ);
1508 *wbufsize = 2;
1509 return 4;
1510 }
1511 *wbufsize = 1;
1512 return 2;
1513 }
1514
1515 static int
utf16_wctomb(csconv_t * cv,ushort * wbuf,int wbufsize,uchar * buf,int bufsize)1516 utf16_wctomb(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize)
1517 {
1518 if ((cv->flags & FLAG_USE_BOM) && !(cv->mode & UNICODE_MODE_BOM_DONE))
1519 {
1520 int r;
1521
1522 cv->mode |= UNICODE_MODE_BOM_DONE;
1523 if (bufsize < 2)
1524 return seterror(E2BIG);
1525 if (cv->codepage == 1200) /* little endian */
1526 memcpy(buf, "\xFF\xFE", 2);
1527 else if (cv->codepage == 1201) /* big endian */
1528 memcpy(buf, "\xFE\xFF", 2);
1529
1530 r = utf16_wctomb(cv, wbuf, wbufsize, buf + 2, bufsize - 2);
1531 if (r == -1)
1532 return -1;
1533 return r + 2;
1534 }
1535
1536 if (bufsize < 2)
1537 return seterror(E2BIG);
1538 if (cv->codepage == 1200) /* little endian */
1539 {
1540 buf[0] = (wbuf[0] & 0x00FF);
1541 buf[1] = (wbuf[0] & 0xFF00) >> 8;
1542 }
1543 else if (cv->codepage == 1201) /* big endian */
1544 {
1545 buf[0] = (wbuf[0] & 0xFF00) >> 8;
1546 buf[1] = (wbuf[0] & 0x00FF);
1547 }
1548 if (0xD800 <= wbuf[0] && wbuf[0] <= 0xDBFF)
1549 {
1550 if (bufsize < 4)
1551 return seterror(E2BIG);
1552 if (cv->codepage == 1200) /* little endian */
1553 {
1554 buf[2] = (wbuf[1] & 0x00FF);
1555 buf[3] = (wbuf[1] & 0xFF00) >> 8;
1556 }
1557 else if (cv->codepage == 1201) /* big endian */
1558 {
1559 buf[2] = (wbuf[1] & 0xFF00) >> 8;
1560 buf[3] = (wbuf[1] & 0x00FF);
1561 }
1562 return 4;
1563 }
1564 return 2;
1565 }
1566
1567 static int
utf32_mbtowc(csconv_t * cv,const uchar * buf,int bufsize,ushort * wbuf,int * wbufsize)1568 utf32_mbtowc(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize)
1569 {
1570 int codepage = cv->codepage;
1571 uint wc;
1572
1573 /* swap endian: 12000 <-> 12001 */
1574 if (cv->mode & UNICODE_MODE_SWAPPED)
1575 codepage ^= 1;
1576
1577 if (bufsize < 4)
1578 return seterror(EINVAL);
1579 if (codepage == 12000) /* little endian */
1580 wc = (buf[3] << 24) | (buf[2] << 16) | (buf[1] << 8) | buf[0];
1581 else if (codepage == 12001) /* big endian */
1582 wc = (buf[0] << 24) | (buf[1] << 16) | (buf[2] << 8) | buf[3];
1583
1584 if ((cv->flags & FLAG_USE_BOM) && !(cv->mode & UNICODE_MODE_BOM_DONE))
1585 {
1586 cv->mode |= UNICODE_MODE_BOM_DONE;
1587 if (wc == 0xFFFE0000)
1588 {
1589 cv->mode |= UNICODE_MODE_SWAPPED;
1590 *wbufsize = 0;
1591 return 4;
1592 }
1593 else if (wc == 0x0000FEFF)
1594 {
1595 *wbufsize = 0;
1596 return 4;
1597 }
1598 }
1599
1600 if ((0xD800 <= wc && wc <= 0xDFFF) || 0x10FFFF < wc)
1601 return seterror(EILSEQ);
1602 ucs4_to_utf16(wc, wbuf, wbufsize);
1603 return 4;
1604 }
1605
1606 static int
utf32_wctomb(csconv_t * cv,ushort * wbuf,int wbufsize,uchar * buf,int bufsize)1607 utf32_wctomb(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize)
1608 {
1609 uint wc;
1610
1611 if ((cv->flags & FLAG_USE_BOM) && !(cv->mode & UNICODE_MODE_BOM_DONE))
1612 {
1613 int r;
1614
1615 cv->mode |= UNICODE_MODE_BOM_DONE;
1616 if (bufsize < 4)
1617 return seterror(E2BIG);
1618 if (cv->codepage == 12000) /* little endian */
1619 memcpy(buf, "\xFF\xFE\x00\x00", 4);
1620 else if (cv->codepage == 12001) /* big endian */
1621 memcpy(buf, "\x00\x00\xFE\xFF", 4);
1622
1623 r = utf32_wctomb(cv, wbuf, wbufsize, buf + 4, bufsize - 4);
1624 if (r == -1)
1625 return -1;
1626 return r + 4;
1627 }
1628
1629 if (bufsize < 4)
1630 return seterror(E2BIG);
1631 wc = utf16_to_ucs4(wbuf);
1632 if (cv->codepage == 12000) /* little endian */
1633 {
1634 buf[0] = wc & 0x000000FF;
1635 buf[1] = (wc & 0x0000FF00) >> 8;
1636 buf[2] = (wc & 0x00FF0000) >> 16;
1637 buf[3] = (wc & 0xFF000000) >> 24;
1638 }
1639 else if (cv->codepage == 12001) /* big endian */
1640 {
1641 buf[0] = (wc & 0xFF000000) >> 24;
1642 buf[1] = (wc & 0x00FF0000) >> 16;
1643 buf[2] = (wc & 0x0000FF00) >> 8;
1644 buf[3] = wc & 0x000000FF;
1645 }
1646 return 4;
1647 }
1648
1649 /*
1650 * 50220: ISO 2022 Japanese with no halfwidth Katakana; Japanese (JIS)
1651 * 50221: ISO 2022 Japanese with halfwidth Katakana; Japanese (JIS-Allow
1652 * 1 byte Kana)
1653 * 50222: ISO 2022 Japanese JIS X 0201-1989; Japanese (JIS-Allow 1 byte
1654 * Kana - SO/SI)
1655 *
1656 * MultiByteToWideChar() and WideCharToMultiByte() behave differently
1657 * depending on Windows version. On XP, WideCharToMultiByte() doesn't
1658 * terminate result sequence with ascii escape. But Vista does.
1659 * Use MLang instead.
1660 */
1661
1662 #define ISO2022_MODE(cs, shift) (((cs) << 8) | (shift))
1663 #define ISO2022_MODE_CS(mode) (((mode) >> 8) & 0xFF)
1664 #define ISO2022_MODE_SHIFT(mode) ((mode) & 0xFF)
1665
1666 #define ISO2022_SI 0
1667 #define ISO2022_SO 1
1668
1669 /* shift in */
1670 static const char iso2022_SI_seq[] = "\x0F";
1671 /* shift out */
1672 static const char iso2022_SO_seq[] = "\x0E";
1673
1674 typedef struct iso2022_esc_t iso2022_esc_t;
1675 struct iso2022_esc_t {
1676 const char *esc;
1677 int esc_len;
1678 int len;
1679 int cs;
1680 };
1681
1682 #define ISO2022JP_CS_ASCII 0
1683 #define ISO2022JP_CS_JISX0201_ROMAN 1
1684 #define ISO2022JP_CS_JISX0201_KANA 2
1685 #define ISO2022JP_CS_JISX0208_1978 3
1686 #define ISO2022JP_CS_JISX0208_1983 4
1687 #define ISO2022JP_CS_JISX0212 5
1688
1689 static iso2022_esc_t iso2022jp_esc[] = {
1690 {"\x1B\x28\x42", 3, 1, ISO2022JP_CS_ASCII},
1691 {"\x1B\x28\x4A", 3, 1, ISO2022JP_CS_JISX0201_ROMAN},
1692 {"\x1B\x28\x49", 3, 1, ISO2022JP_CS_JISX0201_KANA},
1693 {"\x1B\x24\x40", 3, 2, ISO2022JP_CS_JISX0208_1983}, /* unify 1978 with 1983 */
1694 {"\x1B\x24\x42", 3, 2, ISO2022JP_CS_JISX0208_1983},
1695 {"\x1B\x24\x28\x44", 4, 2, ISO2022JP_CS_JISX0212},
1696 {NULL, 0, 0, 0}
1697 };
1698
1699 static int
iso2022jp_mbtowc(csconv_t * cv,const uchar * buf,int bufsize,ushort * wbuf,int * wbufsize)1700 iso2022jp_mbtowc(csconv_t *cv, const uchar *buf, int bufsize, ushort *wbuf, int *wbufsize)
1701 {
1702 iso2022_esc_t *iesc = iso2022jp_esc;
1703 char tmp[MB_CHAR_MAX];
1704 int insize;
1705 HRESULT hr;
1706 DWORD dummy = 0;
1707 int len;
1708 int esc_len;
1709 int cs;
1710 int shift;
1711 int i;
1712
1713 if (buf[0] == 0x1B)
1714 {
1715 for (i = 0; iesc[i].esc != NULL; ++i)
1716 {
1717 esc_len = iesc[i].esc_len;
1718 if (bufsize < esc_len)
1719 {
1720 if (strncmp((char *)buf, iesc[i].esc, bufsize) == 0)
1721 return seterror(EINVAL);
1722 }
1723 else
1724 {
1725 if (strncmp((char *)buf, iesc[i].esc, esc_len) == 0)
1726 {
1727 cv->mode = ISO2022_MODE(iesc[i].cs, ISO2022_SI);
1728 *wbufsize = 0;
1729 return esc_len;
1730 }
1731 }
1732 }
1733 /* not supported escape sequence */
1734 return seterror(EILSEQ);
1735 }
1736 else if (buf[0] == iso2022_SO_seq[0])
1737 {
1738 cv->mode = ISO2022_MODE(ISO2022_MODE_CS(cv->mode), ISO2022_SO);
1739 *wbufsize = 0;
1740 return 1;
1741 }
1742 else if (buf[0] == iso2022_SI_seq[0])
1743 {
1744 cv->mode = ISO2022_MODE(ISO2022_MODE_CS(cv->mode), ISO2022_SI);
1745 *wbufsize = 0;
1746 return 1;
1747 }
1748
1749 cs = ISO2022_MODE_CS(cv->mode);
1750 shift = ISO2022_MODE_SHIFT(cv->mode);
1751
1752 /* reset the mode for informal sequence */
1753 if (buf[0] < 0x20)
1754 {
1755 cs = ISO2022JP_CS_ASCII;
1756 shift = ISO2022_SI;
1757 }
1758
1759 len = iesc[cs].len;
1760 if (bufsize < len)
1761 return seterror(EINVAL);
1762 for (i = 0; i < len; ++i)
1763 if (!(buf[i] < 0x80))
1764 return seterror(EILSEQ);
1765 esc_len = iesc[cs].esc_len;
1766 memcpy(tmp, iesc[cs].esc, esc_len);
1767 if (shift == ISO2022_SO)
1768 {
1769 memcpy(tmp + esc_len, iso2022_SO_seq, 1);
1770 esc_len += 1;
1771 }
1772 memcpy(tmp + esc_len, buf, len);
1773
1774 if ((cv->codepage == 50220 || cv->codepage == 50221
1775 || cv->codepage == 50222) && shift == ISO2022_SO)
1776 {
1777 /* XXX: shift-out cannot be used for mbtowc (both kernel and
1778 * mlang) */
1779 esc_len = iesc[ISO2022JP_CS_JISX0201_KANA].esc_len;
1780 memcpy(tmp, iesc[ISO2022JP_CS_JISX0201_KANA].esc, esc_len);
1781 memcpy(tmp + esc_len, buf, len);
1782 }
1783
1784 insize = len + esc_len;
1785 hr = ConvertINetMultiByteToUnicode(&dummy, cv->codepage,
1786 (const char *)tmp, &insize, (wchar_t *)wbuf, wbufsize);
1787 if (hr != S_OK || insize != len + esc_len)
1788 return seterror(EILSEQ);
1789
1790 /* Check for conversion error. Assuming defaultChar is 0x3F. */
1791 /* ascii should be converted from ascii */
1792 if (wbuf[0] == buf[0]
1793 && cv->mode != ISO2022_MODE(ISO2022JP_CS_ASCII, ISO2022_SI))
1794 return seterror(EILSEQ);
1795
1796 /* reset the mode for informal sequence */
1797 if (cv->mode != ISO2022_MODE(cs, shift))
1798 cv->mode = ISO2022_MODE(cs, shift);
1799
1800 return len;
1801 }
1802
1803 static int
iso2022jp_wctomb(csconv_t * cv,ushort * wbuf,int wbufsize,uchar * buf,int bufsize)1804 iso2022jp_wctomb(csconv_t *cv, ushort *wbuf, int wbufsize, uchar *buf, int bufsize)
1805 {
1806 iso2022_esc_t *iesc = iso2022jp_esc;
1807 char tmp[MB_CHAR_MAX];
1808 int tmpsize = MB_CHAR_MAX;
1809 int insize = wbufsize;
1810 HRESULT hr;
1811 DWORD dummy = 0;
1812 int len;
1813 int esc_len;
1814 int cs;
1815 int shift;
1816 int i;
1817
1818 /*
1819 * MultiByte = [escape sequence] + character + [escape sequence]
1820 *
1821 * Whether trailing escape sequence is added depends on which API is
1822 * used (kernel or MLang, and its version).
1823 */
1824 hr = ConvertINetUnicodeToMultiByte(&dummy, cv->codepage,
1825 (const wchar_t *)wbuf, &wbufsize, tmp, &tmpsize);
1826 if (hr != S_OK || insize != wbufsize)
1827 return seterror(EILSEQ);
1828 else if (bufsize < tmpsize)
1829 return seterror(E2BIG);
1830
1831 if (tmpsize == 1)
1832 {
1833 cs = ISO2022JP_CS_ASCII;
1834 esc_len = 0;
1835 }
1836 else
1837 {
1838 for (i = 1; iesc[i].esc != NULL; ++i)
1839 {
1840 esc_len = iesc[i].esc_len;
1841 if (strncmp(tmp, iesc[i].esc, esc_len) == 0)
1842 {
1843 cs = iesc[i].cs;
1844 break;
1845 }
1846 }
1847 if (iesc[i].esc == NULL)
1848 /* not supported escape sequence */
1849 return seterror(EILSEQ);
1850 }
1851
1852 shift = ISO2022_SI;
1853 if (tmp[esc_len] == iso2022_SO_seq[0])
1854 {
1855 shift = ISO2022_SO;
1856 esc_len += 1;
1857 }
1858
1859 len = iesc[cs].len;
1860
1861 /* Check for converting error. Assuming defaultChar is 0x3F. */
1862 /* ascii should be converted from ascii */
1863 if (cs == ISO2022JP_CS_ASCII && !(wbuf[0] < 0x80))
1864 return seterror(EILSEQ);
1865 else if (tmpsize < esc_len + len)
1866 return seterror(EILSEQ);
1867
1868 if (cv->mode == ISO2022_MODE(cs, shift))
1869 {
1870 /* remove escape sequence */
1871 if (esc_len != 0)
1872 memmove(tmp, tmp + esc_len, len);
1873 esc_len = 0;
1874 }
1875 else
1876 {
1877 if (cs == ISO2022JP_CS_ASCII)
1878 {
1879 esc_len = iesc[ISO2022JP_CS_ASCII].esc_len;
1880 memmove(tmp + esc_len, tmp, len);
1881 memcpy(tmp, iesc[ISO2022JP_CS_ASCII].esc, esc_len);
1882 }
1883 if (ISO2022_MODE_SHIFT(cv->mode) == ISO2022_SO)
1884 {
1885 /* shift-in before changing to other mode */
1886 memmove(tmp + 1, tmp, len + esc_len);
1887 memcpy(tmp, iso2022_SI_seq, 1);
1888 esc_len += 1;
1889 }
1890 }
1891
1892 if (bufsize < len + esc_len)
1893 return seterror(E2BIG);
1894 memcpy(buf, tmp, len + esc_len);
1895 cv->mode = ISO2022_MODE(cs, shift);
1896 return len + esc_len;
1897 }
1898
1899 static int
iso2022jp_flush(csconv_t * cv,uchar * buf,int bufsize)1900 iso2022jp_flush(csconv_t *cv, uchar *buf, int bufsize)
1901 {
1902 iso2022_esc_t *iesc = iso2022jp_esc;
1903 int esc_len;
1904
1905 if (cv->mode != ISO2022_MODE(ISO2022JP_CS_ASCII, ISO2022_SI))
1906 {
1907 esc_len = 0;
1908 if (ISO2022_MODE_SHIFT(cv->mode) != ISO2022_SI)
1909 esc_len += 1;
1910 if (ISO2022_MODE_CS(cv->mode) != ISO2022JP_CS_ASCII)
1911 esc_len += iesc[ISO2022JP_CS_ASCII].esc_len;
1912 if (bufsize < esc_len)
1913 return seterror(E2BIG);
1914
1915 esc_len = 0;
1916 if (ISO2022_MODE_SHIFT(cv->mode) != ISO2022_SI)
1917 {
1918 memcpy(buf, iso2022_SI_seq, 1);
1919 esc_len += 1;
1920 }
1921 if (ISO2022_MODE_CS(cv->mode) != ISO2022JP_CS_ASCII)
1922 {
1923 memcpy(buf + esc_len, iesc[ISO2022JP_CS_ASCII].esc,
1924 iesc[ISO2022JP_CS_ASCII].esc_len);
1925 esc_len += iesc[ISO2022JP_CS_ASCII].esc_len;
1926 }
1927 return esc_len;
1928 }
1929 return 0;
1930 }
1931
1932 #if defined(MAKE_DLL) && defined(USE_LIBICONV_DLL)
1933 BOOL WINAPI
DllMain(HINSTANCE hinstDLL,DWORD fdwReason,LPVOID lpReserved)1934 DllMain(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpReserved)
1935 {
1936 switch( fdwReason )
1937 {
1938 case DLL_PROCESS_ATTACH:
1939 hwiniconv = (HMODULE)hinstDLL;
1940 break;
1941 case DLL_THREAD_ATTACH:
1942 case DLL_THREAD_DETACH:
1943 case DLL_PROCESS_DETACH:
1944 break;
1945 }
1946 return TRUE;
1947 }
1948 #endif
1949
1950 #if defined(MAKE_EXE)
1951 #include <stdio.h>
1952 #include <fcntl.h>
1953 #include <io.h>
1954 int
main(int argc,char ** argv)1955 main(int argc, char **argv)
1956 {
1957 char *fromcode = NULL;
1958 char *tocode = NULL;
1959 int i;
1960 char inbuf[BUFSIZ];
1961 char outbuf[BUFSIZ];
1962 const char *pin;
1963 char *pout;
1964 size_t inbytesleft;
1965 size_t outbytesleft;
1966 size_t rest = 0;
1967 iconv_t cd;
1968 size_t r;
1969 FILE *in = stdin;
1970 FILE *out = stdout;
1971 int ignore = 0;
1972 char *p;
1973
1974 _setmode(_fileno(stdin), _O_BINARY);
1975 _setmode(_fileno(stdout), _O_BINARY);
1976
1977 for (i = 1; i < argc; ++i)
1978 {
1979 if (strcmp(argv[i], "-l") == 0)
1980 {
1981 for (i = 0; codepage_alias[i].name != NULL; ++i)
1982 printf("%s\n", codepage_alias[i].name);
1983 return 0;
1984 }
1985
1986 if (strcmp(argv[i], "-f") == 0)
1987 fromcode = argv[++i];
1988 else if (strcmp(argv[i], "-t") == 0)
1989 tocode = argv[++i];
1990 else if (strcmp(argv[i], "-c") == 0)
1991 ignore = 1;
1992 else if (strcmp(argv[i], "--output") == 0)
1993 {
1994 out = fopen(argv[++i], "wb");
1995 if(out == NULL)
1996 {
1997 fprintf(stderr, "cannot open %s\n", argv[i]);
1998 return 1;
1999 }
2000 }
2001 else
2002 {
2003 in = fopen(argv[i], "rb");
2004 if (in == NULL)
2005 {
2006 fprintf(stderr, "cannot open %s\n", argv[i]);
2007 return 1;
2008 }
2009 break;
2010 }
2011 }
2012
2013 if (fromcode == NULL || tocode == NULL)
2014 {
2015 printf("usage: %s [-c] -f from-enc -t to-enc [file]\n", argv[0]);
2016 return 0;
2017 }
2018
2019 if (ignore)
2020 {
2021 p = tocode;
2022 tocode = (char *)malloc(strlen(p) + strlen("//IGNORE") + 1);
2023 if (tocode == NULL)
2024 {
2025 perror("fatal error");
2026 return 1;
2027 }
2028 strcpy(tocode, p);
2029 strcat(tocode, "//IGNORE");
2030 }
2031
2032 cd = iconv_open(tocode, fromcode);
2033 if (cd == (iconv_t)(-1))
2034 {
2035 perror("iconv_open error");
2036 return 1;
2037 }
2038
2039 while ((inbytesleft = fread(inbuf + rest, 1, sizeof(inbuf) - rest, in)) != 0
2040 || rest != 0)
2041 {
2042 inbytesleft += rest;
2043 pin = inbuf;
2044 pout = outbuf;
2045 outbytesleft = sizeof(outbuf);
2046 r = iconv(cd, &pin, &inbytesleft, &pout, &outbytesleft);
2047 fwrite(outbuf, 1, sizeof(outbuf) - outbytesleft, out);
2048 if (r == (size_t)(-1) && errno != E2BIG && (errno != EINVAL || feof(in)))
2049 {
2050 perror("conversion error");
2051 return 1;
2052 }
2053 memmove(inbuf, pin, inbytesleft);
2054 rest = inbytesleft;
2055 }
2056 pout = outbuf;
2057 outbytesleft = sizeof(outbuf);
2058 r = iconv(cd, NULL, NULL, &pout, &outbytesleft);
2059 fwrite(outbuf, 1, sizeof(outbuf) - outbytesleft, out);
2060 if (r == (size_t)(-1))
2061 {
2062 perror("conversion error");
2063 return 1;
2064 }
2065
2066 iconv_close(cd);
2067
2068 return 0;
2069 }
2070 #endif
2071
2072