source: trunk/src/kernel32/codepage.cpp@ 6725

Last change on this file since 6725 was 6646, checked in by bird, 24 years ago

Added $Id:$ keyword.
File size: 13.4 KB
Line 
1/* $Id: codepage.cpp,v 1.17 2001-09-05 12:57:58 bird Exp $
2 *
3 * Code page functions
4 *
5 * Based on Wine code (memory\codepage.c)
6 *
7 * Copyright 2000 Alexandre Julliard
8 *
9 * Project Odin Software License can be found in LICENSE.TXT
10 *
11 */
12
13#include <assert.h>
14#include <stdio.h>
15#include <stdlib.h>
16#include <string.h>
17
18#include "winbase.h"
19#include "winerror.h"
20#include "winnls.h"
21#include "wine/unicode.h"
22#include "debugtools.h"
23
24#ifdef __WIN32OS2__
25#include <options.h>
26#include "codepage.h"
27
28#define DBG_LOCALLOG DBG_codepage
29#include "dbglocal.h"
30#endif
31
32DEFAULT_DEBUG_CHANNEL(string);
33
34/* current code pages */
35static const union cptable *ansi_cptable;
36static const union cptable *oem_cptable;
37static const union cptable *mac_cptable;
38
39/* retrieve a code page table from the locale info */
40static const union cptable *get_locale_cp( LCID lcid, LCTYPE type )
41{
42 const union cptable *table = NULL;
43 char buf[32];
44
45 if (GetLocaleInfoA( lcid, type, buf, sizeof(buf) )) table = cp_get_table( atoi(buf) );
46 return table;
47}
48
49/* setup default codepage info before we can get at the locale stuff */
50static void init_codepages(void)
51{
52 ansi_cptable = cp_get_table( 1252 );
53 oem_cptable = cp_get_table( 437 );
54 mac_cptable = cp_get_table( 10000 );
55 assert( ansi_cptable );
56 assert( oem_cptable );
57 assert( mac_cptable );
58}
59
60/* find the table for a given codepage, handling CP_ACP etc. pseudo-codepages */
61static const union cptable *get_codepage_table( unsigned int codepage )
62{
63 const union cptable *ret = NULL;
64
65 if (!ansi_cptable) init_codepages();
66
67 switch(codepage)
68 {
69 case CP_ACP: return ansi_cptable;
70 case CP_OEMCP: return oem_cptable;
71 case CP_MACCP: return mac_cptable;
72 case CP_THREAD_ACP: return get_locale_cp( GetThreadLocale(), LOCALE_IDEFAULTANSICODEPAGE );
73 case CP_UTF7:
74 case CP_UTF8:
75 break;
76 default:
77 if (codepage == ansi_cptable->info.codepage) return ansi_cptable;
78 if (codepage == oem_cptable->info.codepage) return oem_cptable;
79 if (codepage == mac_cptable->info.codepage) return mac_cptable;
80 ret = cp_get_table( codepage );
81 break;
82 }
83 return ret;
84}
85
86/* initialize default code pages from locale info */
87/* FIXME: should be done in init_codepages, but it can't right now */
88/* since it needs KERNEL32 to be loaded for the locale info. */
89void CODEPAGE_Init(void)
90{
91 const union cptable *table;
92 LCID lcid = GetUserDefaultLCID();
93
94 if (!ansi_cptable) init_codepages(); /* just in case */
95
96 if ((table = get_locale_cp( lcid, LOCALE_IDEFAULTANSICODEPAGE ))) ansi_cptable = table;
97 if ((table = get_locale_cp( lcid, LOCALE_IDEFAULTMACCODEPAGE ))) mac_cptable = table;
98 if ((table = get_locale_cp( lcid, LOCALE_IDEFAULTCODEPAGE ))) oem_cptable = table;
99
100 TRACE( "ansi=%03d oem=%03d mac=%03d\n", ansi_cptable->info.codepage,
101 oem_cptable->info.codepage, mac_cptable->info.codepage );
102#ifdef __WIN32OS2__
103 dprintf(("Language %s", getenv("LANG")));
104#endif
105}
106
107#ifdef __WIN32OS2__
108ULONG GetDisplayCodepage()
109{
110 if (!ansi_cptable) CODEPAGE_Init();
111
112 return ansi_cptable->info.codepage;
113}
114
115ULONG GetWindowsCodepage()
116{
117 if (!ansi_cptable) CODEPAGE_Init();
118
119 return ansi_cptable->info.codepage;
120}
121
122#endif
123
124/******************************************************************************
125 * GetACP (KERNEL32)
126 *
127 * RETURNS
128 * Current ANSI code-page identifier, default if no current defined
129 */
130UINT WINAPI GetACP(void)
131{
132 if (!ansi_cptable) init_codepages();
133#ifdef __WIN32OS2__
134 dprintf(("GetACP %d", ansi_cptable->info.codepage));
135#endif
136 return ansi_cptable->info.codepage;
137}
138
139
140/***********************************************************************
141 * GetOEMCP (KERNEL32)
142 */
143UINT WINAPI GetOEMCP(void)
144{
145 if (!oem_cptable) init_codepages();
146#ifdef __WIN32OS2__
147 dprintf(("GetOEMCP %d", oem_cptable->info.codepage));
148#endif
149 return oem_cptable->info.codepage;
150}
151
152
153/***********************************************************************
154 * IsValidCodePage (KERNEL32)
155 */
156BOOL WINAPI IsValidCodePage( UINT codepage )
157{
158#ifdef __WIN32OS2__
159 dprintf(("IsValidCodePage %d", codepage));
160#endif
161 return cp_get_table( codepage ) != NULL;
162}
163
164
165/***********************************************************************
166 * IsDBCSLeadByteEx (KERNEL32)
167 */
168BOOL WINAPI IsDBCSLeadByteEx( UINT codepage, BYTE testchar )
169{
170#ifdef __WIN32OS2__
171 dprintf2(("IsDBCSLeadByteEx %d %x", codepage, testchar));
172#endif
173
174 const union cptable *table = get_codepage_table( codepage );
175 return table && is_dbcs_leadbyte( table, testchar );
176}
177
178
179/***********************************************************************
180 * IsDBCSLeadByte (KERNEL32)
181 */
182BOOL WINAPI IsDBCSLeadByte( BYTE testchar )
183{
184 if (!ansi_cptable) init_codepages();
185 return is_dbcs_leadbyte( ansi_cptable, testchar );
186}
187
188
189/***********************************************************************
190 * GetCPInfo (KERNEL32)
191 */
192BOOL WINAPI GetCPInfo( UINT codepage, LPCPINFO cpinfo )
193{
194 const union cptable *table = get_codepage_table( codepage );
195
196#ifdef __WIN32OS2__
197 dprintf(("GetCPInfo %d %x", codepage, cpinfo));
198#endif
199
200 if (!table)
201 {
202 SetLastError( ERROR_INVALID_PARAMETER );
203 return FALSE;
204 }
205 if (table->info.def_char & 0xff00)
206 {
207 cpinfo->DefaultChar[0] = table->info.def_char & 0xff00;
208 cpinfo->DefaultChar[1] = table->info.def_char & 0x00ff;
209 }
210 else
211 {
212 cpinfo->DefaultChar[0] = table->info.def_char & 0xff;
213 cpinfo->DefaultChar[1] = 0;
214 }
215 if ((cpinfo->MaxCharSize = table->info.char_size) == 2)
216 memcpy( cpinfo->LeadByte, table->dbcs.lead_bytes, sizeof(cpinfo->LeadByte) );
217 else
218 cpinfo->LeadByte[0] = cpinfo->LeadByte[1] = 0;
219
220 return TRUE;
221}
222
223
224/***********************************************************************
225 * EnumSystemCodePagesA (KERNEL32)
226 */
227BOOL WINAPI EnumSystemCodePagesA( CODEPAGE_ENUMPROCA lpfnCodePageEnum, DWORD flags )
228{
229 const union cptable *table;
230 char buffer[10];
231 int index = 0;
232
233#ifdef __WIN32OS2__
234 dprintf(("EnumSystemCodePagesA %x %x", lpfnCodePageEnum, flags));
235#endif
236
237 for (;;)
238 {
239 if (!(table = cp_enum_table( index++ ))) break;
240 sprintf( buffer, "%d", table->info.codepage );
241 if (!lpfnCodePageEnum( buffer )) break;
242 }
243 return TRUE;
244}
245
246
247/***********************************************************************
248 * EnumSystemCodePagesW (KERNEL32)
249 */
250BOOL WINAPI EnumSystemCodePagesW( CODEPAGE_ENUMPROCW lpfnCodePageEnum, DWORD flags )
251{
252 const union cptable *table;
253 WCHAR buffer[10], *p;
254 int page, index = 0;
255
256#ifdef __WIN32OS2__
257 dprintf(("EnumSystemCodePagesW %x %x", lpfnCodePageEnum, flags));
258#endif
259
260 for (;;)
261 {
262 if (!(table = cp_enum_table( index++ ))) break;
263 p = buffer + sizeof(buffer)/sizeof(WCHAR);
264 *--p = 0;
265 page = table->info.codepage;
266 do
267 {
268 *--p = '0' + (page % 10);
269 page /= 10;
270 } while( page );
271 if (!lpfnCodePageEnum( p )) break;
272 }
273 return TRUE;
274}
275
276
277/***********************************************************************
278 * MultiByteToWideChar (KERNEL32)
279 *
280 * PARAMS
281 * page [in] Codepage character set to convert from
282 * flags [in] Character mapping flags
283 * src [in] Source string buffer
284 * srclen [in] Length of source string buffer
285 * dst [in] Destination buffer
286 * dstlen [in] Length of destination buffer
287 *
288 * NOTES
289 * The returned length includes the null terminator character.
290 *
291 * RETURNS
292 * Success: If dstlen > 0, number of characters written to destination
293 * buffer. If dstlen == 0, number of characters needed to do
294 * conversion.
295 * Failure: 0. Occurs if not enough space is available.
296 *
297 * ERRORS
298 * ERROR_INSUFFICIENT_BUFFER
299 * ERROR_INVALID_PARAMETER
300 * ERROR_NO_UNICODE_TRANSLATION
301 *
302 */
303INT WINAPI MultiByteToWideChar( UINT page, DWORD flags, LPCSTR src, INT srclen,
304 LPWSTR dst, INT dstlen )
305{
306 const union cptable *table;
307 int ret;
308
309#ifdef __WIN32OS2__
310 dprintf2(("MultiByteToWideChar %d %x %x %d %x %d", page, flags, src, srclen, dst, dstlen));
311#endif
312
313 //Docs say source ptr can't be the same as destination (Windows ME, NT4-SP6)
314 if (!src || (!dst && dstlen) || ((void *)src == (void *)dst))
315 {
316 SetLastError( ERROR_INVALID_PARAMETER );
317 return 0;
318 }
319
320 //Even though the docs claim this only works for -1, testing shows it
321 //is done for any negative value (Windows ME, NT4-SP6)
322 if (srclen <= -1) srclen = strlen(src) + 1;
323
324 if (flags & MB_USEGLYPHCHARS) FIXME("MB_USEGLYPHCHARS not supported\n");
325
326 switch(page)
327 {
328 case CP_UTF7:
329 FIXME("UTF not supported\n");
330 SetLastError( ERROR_CALL_NOT_IMPLEMENTED );
331 return 0;
332 case CP_UTF8:
333 ret = utf8_mbstowcs( flags, src, srclen, dst, dstlen );
334 break;
335 default:
336 if (!(table = get_codepage_table( page )))
337 {
338 SetLastError( ERROR_INVALID_PARAMETER );
339 return 0;
340 }
341 ret = cp_mbstowcs( table, flags, src, srclen, dst, dstlen );
342 break;
343 }
344
345 if (ret < 0)
346 {
347 switch(ret)
348 {
349 case -1: SetLastError( ERROR_INSUFFICIENT_BUFFER ); break;
350 case -2: SetLastError( ERROR_NO_UNICODE_TRANSLATION ); break;
351 }
352 ret = 0;
353 }
354 return ret;
355}
356
357
358/***********************************************************************
359 * WideCharToMultiByte (KERNEL32)
360 *
361 * PARAMS
362 * page [in] Codepage character set to convert to
363 * flags [in] Character mapping flags
364 * src [in] Source string buffer
365 * srclen [in] Length of source string buffer
366 * dst [in] Destination buffer
367 * dstlen [in] Length of destination buffer
368 * defchar [in] Default character to use for conversion if no exact
369 * conversion can be made
370 * used [out] Set if default character was used in the conversion
371 *
372 * NOTES
373 * The returned length includes the null terminator character.
374 *
375 * RETURNS
376 * Success: If dstlen > 0, number of characters written to destination
377 * buffer. If dstlen == 0, number of characters needed to do
378 * conversion.
379 * Failure: 0. Occurs if not enough space is available.
380 *
381 * ERRORS
382 * ERROR_INSUFFICIENT_BUFFER
383 * ERROR_INVALID_PARAMETER
384 */
385INT WINAPI WideCharToMultiByte( UINT page, DWORD flags, LPCWSTR src, INT srclen,
386 LPSTR dst, INT dstlen, LPCSTR defchar, BOOL *used )
387{
388 const union cptable *table;
389 int ret, used_tmp;
390
391#ifdef __WIN32OS2__
392 dprintf2(("WideCharToMultiByte %d %x %x %d %x %d", page, flags, src, srclen, dst, dstlen));
393#endif
394
395 //Docs say source ptr can't be the same as destination (Windows ME, NT4-SP6)
396 if (!src || (!dst && dstlen) || ((void *)src == (void *)dst))
397 {
398 SetLastError( ERROR_INVALID_PARAMETER );
399 return 0;
400 }
401
402 //Even though the docs claim this only works for -1, testing shows it
403 //is done for any negative value (Windows ME, NT4 - SP6)
404 if (srclen <= -1) srclen = strlenW(src) + 1;
405
406 switch(page)
407 {
408 case CP_UTF7:
409 FIXME("UTF-7 not supported\n");
410 SetLastError( ERROR_CALL_NOT_IMPLEMENTED );
411 return 0;
412 case CP_UTF8:
413 ret = utf8_wcstombs( src, srclen, dst, dstlen );
414 break;
415 default:
416 if (!(table = get_codepage_table( page )))
417 {
418 SetLastError( ERROR_INVALID_PARAMETER );
419 return 0;
420 }
421 ret = cp_wcstombs( table, flags, src, srclen, dst, dstlen,
422 defchar, used ? &used_tmp : NULL );
423 if (used) *used = used_tmp;
424 break;
425 }
426
427 if (ret == -1)
428 {
429 SetLastError( ERROR_INSUFFICIENT_BUFFER );
430 ret = 0;
431 }
432 return ret;
433}
434
435
436/******************************************************************************
437 * GetStringTypeW (KERNEL32)
438 *
439 */
440BOOL WINAPI GetStringTypeW( DWORD type, LPCWSTR src, INT count, LPWORD chartype )
441{
442#ifdef __WIN32OS2__
443 dprintf(("GetStringTypeW %x %x %d %x", type, src, count, chartype));
444#endif
445
446 if (count == -1) count = strlenW(src) + 1;
447 switch(type)
448 {
449 case CT_CTYPE1:
450 while (count--) *chartype++ = get_char_typeW( *src++ ) & 0xfff;
451 break;
452 case CT_CTYPE2:
453 while (count--) *chartype++ = get_char_typeW( *src++ ) >> 12;
454 break;
455 case CT_CTYPE3:
456 FIXME("CT_CTYPE3 not supported.\n");
457 default:
458 SetLastError( ERROR_INVALID_PARAMETER );
459 return FALSE;
460 }
461 return TRUE;
462}
463
464
465/******************************************************************************
466 * GetStringTypeExW (KERNEL32)
467 */
468BOOL WINAPI GetStringTypeExW( LCID locale, DWORD type, LPCWSTR src, INT count, LPWORD chartype )
469{
470 /* locale is ignored for Unicode */
471 return GetStringTypeW( type, src, count, chartype );
472}
473
474WCHAR WIN32API tolowerW( WCHAR ch )
475{
476 extern const WCHAR casemap_lower[];
477 return ch + casemap_lower[casemap_lower[ch >> 8] + (ch & 0xff)];
478}
479
480WCHAR WIN32API toupperW( WCHAR ch )
481{
482 extern const WCHAR casemap_upper[];
483 return ch + casemap_upper[casemap_upper[ch >> 8] + (ch & 0xff)];
484}
485
486/* the character type contains the C1_* flags in the low 12 bits */
487/* and the C2_* type in the high 4 bits */
488unsigned short get_char_typeW( WCHAR ch )
489{
490 extern const unsigned short wctype_table[];
491 return wctype_table[wctype_table[ch >> 8] + (ch & 0xff)];
492}
Note: See TracBrowser for help on using the repository browser.