1/* Conversion from and to IBM930.
2 Copyright (C) 2000-2020 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 Lesser General Public
8 License as published by the Free Software Foundation; either
9 version 2.1 of the 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 Lesser General Public License for more details.
15
16 You should have received a copy of the GNU Lesser General Public
17 License along with the GNU C Library; if not, see
18 <https://www.gnu.org/licenses/>. */
19
20/* IBM930 is designed for the representation of Japanese Katakana/Kanji
21 using a stateful EBCDIC encoding scheme. It is also known as
22 CCSID 930 or CP930. See:
23 https://www-01.ibm.com/software/globalization/ccsid/ccsid930.html */
24
25#include <dlfcn.h>
26#include <stdint.h>
27#include <wchar.h>
28#include <byteswap.h>
29#include "ibm930.h"
30
31/* The shift sequences for this charset (it does not use ESC). */
32#define SI 0x0F /* Shift In, host code to turn DBCS off. */
33#define SO 0x0E /* Shift Out, host code to turn DBCS on. */
34
35/* Definitions used in the body of the `gconv' function. */
36#define CHARSET_NAME "IBM930//"
37#define FROM_LOOP from_ibm930
38#define TO_LOOP to_ibm930
39#define ONE_DIRECTION 0
40#define FROM_LOOP_MIN_NEEDED_FROM 1
41#define FROM_LOOP_MAX_NEEDED_FROM 2
42#define FROM_LOOP_MIN_NEEDED_TO 4
43#define FROM_LOOP_MAX_NEEDED_TO 4
44#define TO_LOOP_MIN_NEEDED_FROM 4
45#define TO_LOOP_MAX_NEEDED_FROM 4
46#define TO_LOOP_MIN_NEEDED_TO 1
47#define TO_LOOP_MAX_NEEDED_TO 3
48#define PREPARE_LOOP \
49 int save_curcs; \
50 int *curcsp = &data->__statep->__count;
51#define EXTRA_LOOP_ARGS , curcsp
52
53/* Definitions of initialization and destructor function. */
54#define DEFINE_INIT 1
55#define DEFINE_FINI 1
56
57
58/* Since this is a stateful encoding we have to provide code which resets
59 the output state to the initial state. This has to be done during the
60 flushing. */
61#define EMIT_SHIFT_TO_INIT \
62 if ((data->__statep->__count & ~7) != sb) \
63 { \
64 if (FROM_DIRECTION) \
65 data->__statep->__count &= 7; \
66 else \
67 { \
68 /* We are not in the initial state. To switch back we have \
69 to emit `SI'. */ \
70 if (__glibc_unlikely (outbuf >= outend)) \
71 /* We don't have enough room in the output buffer. */ \
72 status = __GCONV_FULL_OUTPUT; \
73 else \
74 { \
75 /* Write out the shift sequence. */ \
76 *outbuf++ = SI; \
77 data->__statep->__count &= 7; \
78 } \
79 } \
80 }
81
82
83/* Since we might have to reset input pointer we must be able to save
84 and retore the state. */
85#define SAVE_RESET_STATE(Save) \
86 if (Save) \
87 save_curcs = *curcsp; \
88 else \
89 *curcsp = save_curcs
90
91
92/* Current codeset type. */
93enum
94{
95 sb = 0,
96 db = 64
97};
98
99
100/* First, define the conversion function from IBM-930 to UCS4. */
101#define MIN_NEEDED_INPUT FROM_LOOP_MIN_NEEDED_FROM
102#define MAX_NEEDED_INPUT FROM_LOOP_MAX_NEEDED_FROM
103#define MIN_NEEDED_OUTPUT FROM_LOOP_MIN_NEEDED_TO
104#define MAX_NEEDED_OUTPUT FROM_LOOP_MAX_NEEDED_TO
105#define LOOPFCT FROM_LOOP
106#define BODY \
107 { \
108 uint32_t ch = *inptr; \
109 uint32_t res; \
110 \
111 if (__builtin_expect (ch, 0) == SO) \
112 { \
113 /* Shift OUT, change to DBCS converter (redundant escape okay). */ \
114 curcs = db; \
115 ++inptr; \
116 continue; \
117 } \
118 else if (__builtin_expect (ch, 0) == SI) \
119 { \
120 /* Shift IN, change to SBCS converter (redundant escape okay). */ \
121 curcs = sb; \
122 ++inptr; \
123 continue; \
124 } \
125 \
126 if (curcs == sb) \
127 { \
128 /* Use the IBM930 table for single byte. */ \
129 res = __ibm930sb_to_ucs4[ch]; \
130 if (__builtin_expect (res, L'\1') == L'\0' && ch != '\0') \
131 { \
132 /* This is an illegal character. */ \
133 STANDARD_FROM_LOOP_ERR_HANDLER (1); \
134 } \
135 else \
136 { \
137 put32 (outptr, res); \
138 outptr += 4; \
139 } \
140 ++inptr; \
141 } \
142 else \
143 { \
144 /* Use the IBM930 table for double byte. */ \
145 const struct gap *rp2 = __ibm930db_to_ucs4_idx; \
146 \
147 assert (curcs == db); \
148 \
149 if (__glibc_unlikely (inptr + 1 >= inend)) \
150 { \
151 /* The second character is not available. Store the \
152 intermediate result. */ \
153 result = __GCONV_INCOMPLETE_INPUT; \
154 break; \
155 } \
156 \
157 ch = (ch * 0x100) + inptr[1]; \
158 while (ch > rp2->end) \
159 ++rp2; \
160 \
161 if (__builtin_expect (rp2->start == 0xffff, 0) \
162 || __builtin_expect (ch < rp2->start, 0) \
163 || (res = __ibm930db_to_ucs4[ch + rp2->idx], \
164 __builtin_expect (res, L'\1') == L'\0' && ch != '\0')) \
165 { \
166 /* This is an illegal character. */ \
167 STANDARD_FROM_LOOP_ERR_HANDLER (2); \
168 } \
169 else \
170 { \
171 put32 (outptr, res); \
172 outptr += 4; \
173 } \
174 inptr += 2; \
175 } \
176 }
177#define LOOP_NEED_FLAGS
178#define EXTRA_LOOP_DECLS , int *curcsp
179#define INIT_PARAMS int curcs = *curcsp & ~7
180#define UPDATE_PARAMS *curcsp = curcs
181#include <iconv/loop.c>
182
183/* Next, define the other direction. */
184#define MIN_NEEDED_INPUT TO_LOOP_MIN_NEEDED_FROM
185#define MAX_NEEDED_INPUT TO_LOOP_MAX_NEEDED_FROM
186#define MIN_NEEDED_OUTPUT TO_LOOP_MIN_NEEDED_TO
187#define MAX_NEEDED_OUTPUT TO_LOOP_MAX_NEEDED_TO
188#define LOOPFCT TO_LOOP
189#define BODY \
190 { \
191 uint32_t ch = get32 (inptr); \
192 const struct gap *rp1 = __ucs4_to_ibm930sb_idx; \
193 const struct gap *rp2 = __ucs4_to_ibm930db_idx; \
194 \
195 if (__glibc_unlikely (ch >= 0xffff)) \
196 { \
197 UNICODE_TAG_HANDLER (ch, 4); \
198 \
199 STANDARD_TO_LOOP_ERR_HANDLER (4); \
200 } \
201 \
202 while (ch > rp1->end) \
203 ++rp1; \
204 \
205 /* Use the UCS4 table for single byte. */ \
206 unsigned char sbconv; \
207 if (__builtin_expect (ch < rp1->start, 0) \
208 || (sbconv = __ucs4_to_ibm930sb[ch + rp1->idx], \
209 __builtin_expect (sbconv, L'\1') == L'\0' && ch != '\0')) \
210 { \
211 /* Use the UCS4 table for double byte. */ \
212 while (ch > rp2->end) \
213 ++rp2; \
214 \
215 const char *cp; \
216 if (__builtin_expect (ch < rp2->start, 0) \
217 || (cp = __ucs4_to_ibm930db[ch + rp2->idx], \
218 __builtin_expect (cp[0], L'\1')== L'\0' && ch != '\0')) \
219 { \
220 /* This is an illegal character. */ \
221 STANDARD_TO_LOOP_ERR_HANDLER (4); \
222 } \
223 else \
224 { \
225 if (curcs == sb) \
226 { \
227 if (__glibc_unlikely (outptr + 1 > outend)) \
228 { \
229 result = __GCONV_FULL_OUTPUT; \
230 break; \
231 } \
232 *outptr++ = SO; \
233 curcs = db; \
234 } \
235 \
236 if (__glibc_unlikely (outptr + 2 > outend)) \
237 { \
238 result = __GCONV_FULL_OUTPUT; \
239 break; \
240 } \
241 *outptr++ = cp[0]; \
242 *outptr++ = cp[1]; \
243 } \
244 } \
245 else \
246 { \
247 if (curcs == db) \
248 { \
249 if (__glibc_unlikely (outptr + 1 > outend)) \
250 { \
251 result = __GCONV_FULL_OUTPUT; \
252 break; \
253 } \
254 *outptr++ = SI; \
255 curcs = sb; \
256 } \
257 \
258 if (__glibc_unlikely (outptr + 1 > outend)) \
259 { \
260 result = __GCONV_FULL_OUTPUT; \
261 break; \
262 } \
263 if (ch == 0x7e) \
264 *outptr++ = 0xa1; \
265 else if (ch == 0x5c) \
266 *outptr++ = 0x5b; \
267 else \
268 *outptr++ = sbconv; \
269 } \
270 \
271 /* Now that we wrote the output increment the input pointer. */ \
272 inptr += 4; \
273 }
274#define LOOP_NEED_FLAGS
275#define EXTRA_LOOP_DECLS , int *curcsp
276#define INIT_PARAMS int curcs = *curcsp & ~7
277#define REINIT_PARAMS curcs = *curcsp & ~7
278#define UPDATE_PARAMS *curcsp = curcs
279#include <iconv/loop.c>
280
281/* Now define the toplevel functions. */
282#include <iconv/skeleton.c>
283