Fix handling of state. Optimize a bit.
[kopensolaris-gnu/glibc.git] / iconvdata / ibm939.c
1 /* Conversion to and from IBM939.
2    Copyright (C) 2000, 2001 Free Software Foundation, Inc.
3    This file is part of the GNU C Library.
4    Contributed by Masahide Washizawa <washi@yamato.ibm.co.jp>, 2000.
5
6    The GNU C Library is free software; you can redistribute it and/or
7    modify it under the terms of the GNU Library General Public License as
8    published by the Free Software Foundation; either version 2 of the
9    License, or (at your option) any later version.
10
11    The GNU C Library is distributed in the hope that it will be useful,
12    but WITHOUT ANY WARRANTY; without even the implied warranty of
13    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14    Library General Public License for more details.
15
16    You should have received a copy of the GNU Library General Public
17    License along with the GNU C Library; see the file COPYING.LIB.  If not,
18    write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
19    Boston, MA 02111-1307, USA.  */
20
21 #include <dlfcn.h>
22 #include <stdint.h>
23 #include <wchar.h>
24 #include <byteswap.h>
25 #include "ibm939.h"
26
27 /* The shift sequences for this charset (it does not use ESC).  */
28 #define SI              0x0F  /* Shift In, host code to turn DBCS off.  */
29 #define SO              0x0E  /* Shift Out, host code to turn DBCS on.  */
30
31 /* Definitions used in the body of the `gconv' function.  */
32 #define CHARSET_NAME    "IBM939//"
33 #define FROM_LOOP       from_ibm939
34 #define TO_LOOP         to_ibm939
35 #define MIN_NEEDED_FROM 1
36 #define MAX_NEEDED_FROM 2
37 #define MIN_NEEDED_TO   4
38 #define MAX_NEEDED_TO   4
39 #define PREPARE_LOOP \
40   int save_curcs;                                                             \
41   int *curcsp = &data->__statep->__count;
42 #define EXTRA_LOOP_ARGS         , curcsp
43
44 /* Definitions of initialization and destructor function.  */
45 #define DEFINE_INIT     1
46 #define DEFINE_FINI     1
47
48
49 /* Since this is a stateful encoding we have to provide code which resets
50    the output state to the initial state.  This has to be done during the
51    flushing.  */
52 #define EMIT_SHIFT_TO_INIT \
53   if ((data->__statep->__count & ~7) != sb)                                   \
54     {                                                                         \
55       if (FROM_DIRECTION)                                                     \
56         data->__statep->__count &= 7;                                         \
57       else                                                                    \
58         {                                                                     \
59           unsigned char *outbuf = data->__outbuf;                             \
60                                                                               \
61           /* We are not in the initial state.  To switch back we have         \
62              to emit `SI'.  */                                                \
63           if (__builtin_expect (outbuf >= data->__outbufend, 0))              \
64             /* We don't have enough room in the output buffer.  */            \
65             status = __GCONV_FULL_OUTPUT;                                     \
66           else                                                                \
67             {                                                                 \
68               /* Write out the shift sequence.  */                            \
69               *outbuf++ = SI;                                                 \
70               data->__outbuf = outbuf;                                        \
71               data->__statep->__count &= 7;                                   \
72             }                                                                 \
73         }                                                                     \
74     }
75
76
77 /* Since we might have to reset input pointer we must be able to save
78    and retore the state.  */
79 #define SAVE_RESET_STATE(Save) \
80   if (Save)                                                                   \
81     save_curcs = *curcsp;                                                     \
82   else                                                                        \
83     *curcsp = save_curcs
84
85
86 /* Current codeset type.  */
87 enum
88 {
89   sb = 0,
90   db = 64
91 };
92
93 /* First, define the conversion function from IBM-939 to UCS4.  */
94 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
95 #define MAX_NEEDED_INPUT        MAX_NEEDED_FROM
96 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
97 #define LOOPFCT                 FROM_LOOP
98 #define BODY \
99   {                                                                           \
100     uint32_t ch = *inptr;                                                     \
101     uint32_t res;                                                             \
102                                                                               \
103     if (__builtin_expect (ch, 0) == SO)                                       \
104       {                                                                       \
105         /* Shift OUT, change to DBCS converter.  */                           \
106         if (curcs == db)                                                      \
107           {                                                                   \
108             result = __GCONV_ILLEGAL_INPUT;                                   \
109             break;                                                            \
110           }                                                                   \
111         curcs = db;                                                           \
112         ++inptr;                                                              \
113         continue;                                                             \
114       }                                                                       \
115     else if (__builtin_expect (ch, 0) == SI)                                  \
116       {                                                                       \
117         /* Shift IN, change to SBCS converter.  */                            \
118         if (curcs == sb)                                                      \
119           {                                                                   \
120             result = __GCONV_ILLEGAL_INPUT;                                   \
121             break;                                                            \
122           }                                                                   \
123         curcs = sb;                                                           \
124         ++inptr;                                                              \
125         continue;                                                             \
126       }                                                                       \
127                                                                               \
128     if (curcs == sb)                                                          \
129       {                                                                       \
130         /* Use the IBM939 table for single byte.  */                          \
131         res = __ibm939sb_to_ucs4[ch];                                         \
132         if (__builtin_expect (res, L'\1') == L'\0' && ch != '\0')             \
133           {                                                                   \
134             /* This is an illegal character.  */                              \
135             if (! ignore_errors_p ())                                         \
136               {                                                               \
137                 result = __GCONV_ILLEGAL_INPUT;                               \
138                 break;                                                        \
139               }                                                               \
140             ++*irreversible;                                                  \
141           }                                                                   \
142         else                                                                  \
143           {                                                                   \
144             put32 (outptr, res);                                              \
145             outptr += 4;                                                      \
146           }                                                                   \
147         ++inptr;                                                              \
148       }                                                                       \
149     else                                                                      \
150       {                                                                       \
151         /* Use the IBM939 table for double byte.  */                          \
152         const struct gap *rp2 = __ibm939db_to_ucs4_idx;                       \
153                                                                               \
154         assert (curcs == db);                                                 \
155                                                                               \
156         if (__builtin_expect (inptr + 1 >= inend, 0))                         \
157           {                                                                   \
158             /* The second character is not available.  Store the              \
159                intermediate result. */                                        \
160             result = __GCONV_INCOMPLETE_INPUT;                                \
161             break;                                                            \
162           }                                                                   \
163                                                                               \
164         ch = (ch * 0x100) + inptr[1];                                         \
165         while (ch > rp2->end)                                                 \
166           ++rp2;                                                              \
167                                                                               \
168         if (__builtin_expect (rp2 == NULL, 0)                                 \
169             || __builtin_expect (ch < rp2->start, 0)                          \
170             || (res = __ibm939db_to_ucs4[ch + rp2->idx],                      \
171                 __builtin_expect (res, L'\1') == L'\0' && ch != '\0'))        \
172           {                                                                   \
173             /* This is an illegal character.  */                              \
174             if (! ignore_errors_p ())                                         \
175               {                                                               \
176                 result = __GCONV_ILLEGAL_INPUT;                               \
177                 break;                                                        \
178               }                                                               \
179             ++*irreversible;                                                  \
180           }                                                                   \
181         else                                                                  \
182           {                                                                   \
183             put32 (outptr, res);                                              \
184             outptr += 4;                                                      \
185           }                                                                   \
186         inptr += 2;                                                           \
187       }                                                                       \
188   }
189 #define LOOP_NEED_FLAGS
190 #define EXTRA_LOOP_DECLS        , int *curcsp
191 #define INIT_PARAMS             int curcs = *curcsp & ~7
192 #define UPDATE_PARAMS           *curcsp = curcs
193 #include <iconv/loop.c>
194
195 /* Next, define the other direction */
196 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
197 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
198 #define MAX_NEEDED_OUTPUT       MAX_NEEDED_FROM
199 #define LOOPFCT                 TO_LOOP
200 #define BODY \
201   {                                                                           \
202     uint32_t ch = get32 (inptr);                                              \
203     const struct gap *rp1 = __ucs4_to_ibm939sb_idx;                           \
204     const struct gap *rp2 = __ucs4_to_ibm939db_idx;                           \
205     const char *cp;                                                           \
206                                                                               \
207     if (__builtin_expect (ch, 0) >= 0xffff)                                   \
208       {                                                                       \
209         rp1 = NULL;                                                           \
210         rp2 = NULL;                                                           \
211       }                                                                       \
212     else                                                                      \
213       while (ch > rp1->end)                                                   \
214         ++rp1;                                                                \
215                                                                               \
216     /* Use the UCS4 table for single byte.  */                                \
217     if (__builtin_expect (rp1 == NULL, 0)                                     \
218         || __builtin_expect (ch < rp1->start, 0)                              \
219         || (cp = __ucs4_to_ibm939sb[ch + rp1->idx],                           \
220             __builtin_expect (cp[0], L'\1') == L'\0' && ch != '\0'))          \
221       {                                                                       \
222         /* Use the UCS4 table for double byte.  */                            \
223         while (ch > rp2->end)                                                 \
224           ++rp2;                                                              \
225                                                                               \
226         if (__builtin_expect (rp2 == NULL, 0)                                 \
227             || __builtin_expect (ch < rp2->start, 0)                          \
228             || (cp = __ucs4_to_ibm939db[ch + rp2->idx],                       \
229                 __builtin_expect (cp[0], L'\1')==L'\0' && ch != '\0'))        \
230           {                                                                   \
231             /* This is an illegal character.  */                              \
232             if (! ignore_errors_p ())                                         \
233               {                                                               \
234                 result = __GCONV_ILLEGAL_INPUT;                               \
235                 break;                                                        \
236               }                                                               \
237             ++*irreversible;                                                  \
238           }                                                                   \
239         else                                                                  \
240           {                                                                   \
241             if (curcs == sb)                                                  \
242               {                                                               \
243                 if (__builtin_expect (outptr+1 > outend, 0))                  \
244                   {                                                           \
245                     result = __GCONV_FULL_OUTPUT;                             \
246                     break;                                                    \
247                   }                                                           \
248                 *outptr++ = SO;                                               \
249                 curcs = db;                                                   \
250               }                                                               \
251                                                                               \
252             if (__builtin_expect (outptr + 2 > outend, 0))                    \
253               {                                                               \
254                 result = __GCONV_FULL_OUTPUT;                                 \
255                 break;                                                        \
256               }                                                               \
257             *outptr++ = cp[0];                                                \
258             *outptr++ = cp[1];                                                \
259           }                                                                   \
260       }                                                                       \
261     else                                                                      \
262       {                                                                       \
263         if (curcs == db)                                                      \
264           {                                                                   \
265             if (__builtin_expect (outptr + 1 > outend, 0))                    \
266               {                                                               \
267                 result = __GCONV_FULL_OUTPUT;                                 \
268                 break;                                                        \
269               }                                                               \
270             *outptr++ = SI;                                                   \
271           }                                                                   \
272                                                                               \
273         if (__builtin_expect (outptr + 1 > outend, 0))                        \
274           {                                                                   \
275             result = __GCONV_FULL_OUTPUT;                                     \
276             break;                                                            \
277           }                                                                   \
278         if (ch == 0x7e)                                                       \
279           *outptr++ = 0xa0;                                                   \
280         else if (ch == 0x5c)                                                  \
281           *outptr++ = 0xb2;                                                   \
282         else                                                                  \
283           *outptr++ = cp[0];                                                  \
284         curcs = sb;                                                           \
285       }                                                                       \
286                                                                               \
287     /* Now that we wrote the output increment the input pointer.  */          \
288     inptr += 4;                                                               \
289   }
290 #define LOOP_NEED_FLAGS
291 #define EXTRA_LOOP_DECLS        , int *curcsp
292 #define INIT_PARAMS             int curcs = *curcsp & ~7
293 #define UPDATE_PARAMS           *curcsp = curcs
294 #include <iconv/loop.c>
295
296 /* Now define the toplevel functions. */
297 #include <iconv/skeleton.c>