1 | /** |
---|
2 | * @file |
---|
3 | * |
---|
4 | * @ingroup DOSFS |
---|
5 | * |
---|
6 | * @brief UTF-8 Converter |
---|
7 | */ |
---|
8 | |
---|
9 | /* |
---|
10 | * Copyright (c) 2013 embedded brains GmbH. All rights reserved. |
---|
11 | * |
---|
12 | * embedded brains GmbH |
---|
13 | * Dornierstr. 4 |
---|
14 | * 82178 Puchheim |
---|
15 | * Germany |
---|
16 | * <rtems@embedded-brains.de> |
---|
17 | * |
---|
18 | * The license and distribution terms for this file may be |
---|
19 | * found in the file LICENSE in this distribution or at |
---|
20 | * http://www.rtems.org/license/LICENSE. |
---|
21 | */ |
---|
22 | |
---|
23 | #include <stddef.h> |
---|
24 | #include <assert.h> |
---|
25 | #include <errno.h> |
---|
26 | #include <iconv.h> |
---|
27 | #include <rtems/dosfs.h> |
---|
28 | #include <utf8proc/utf8proc.h> |
---|
29 | #include "msdos.h" |
---|
30 | |
---|
31 | #define INVALID_ICONV_DESC ( (iconv_t) -1 ) |
---|
32 | |
---|
33 | typedef struct { |
---|
34 | /* |
---|
35 | * This structure must be the first field, since otherwise the cast |
---|
36 | * operations later in the file are invalid. |
---|
37 | */ |
---|
38 | rtems_dosfs_convert_control super; |
---|
39 | |
---|
40 | iconv_t desc_codepage_to_utf8; |
---|
41 | iconv_t desc_utf8_to_codepage; |
---|
42 | iconv_t desc_utf16_to_utf8; |
---|
43 | iconv_t desc_utf8_to_utf16; |
---|
44 | uint8_t buffer[MSDOS_NAME_MAX_UTF8_LFN_BYTES]; |
---|
45 | } msdos_utf8_convert_control; |
---|
46 | |
---|
47 | static int msdos_utf8_convert_with_iconv( |
---|
48 | iconv_t desc, |
---|
49 | const void *src, |
---|
50 | size_t src_size, |
---|
51 | void *dst, |
---|
52 | size_t *dst_size |
---|
53 | ) |
---|
54 | { |
---|
55 | int eno; |
---|
56 | size_t inbytes_left = src_size; |
---|
57 | size_t outbytes_left = *dst_size; |
---|
58 | char *inbuf = (void *) (uintptr_t) src; |
---|
59 | char *outbuf = dst; |
---|
60 | size_t iconv_status; |
---|
61 | |
---|
62 | iconv_status = iconv( |
---|
63 | desc, |
---|
64 | &inbuf, |
---|
65 | &inbytes_left, |
---|
66 | &outbuf, |
---|
67 | &outbytes_left |
---|
68 | ); |
---|
69 | |
---|
70 | *dst_size -= outbytes_left; |
---|
71 | |
---|
72 | if ( iconv_status == 0 ) { |
---|
73 | eno = 0; |
---|
74 | } else if ( iconv_status == (size_t) -1 ) { |
---|
75 | /* |
---|
76 | * iconv() has detected an error. The most likely reason seems to be a too |
---|
77 | * small outbuf. |
---|
78 | */ |
---|
79 | eno = ENOMEM; |
---|
80 | } else { |
---|
81 | /* |
---|
82 | * The iconv_status contains the number of characters converted in a |
---|
83 | * non-reversible way. We want to use reversible conversions only. |
---|
84 | * Characters permitted within DOSFS names seem to be reversible. |
---|
85 | */ |
---|
86 | eno = EINVAL; |
---|
87 | } |
---|
88 | |
---|
89 | return eno; |
---|
90 | } |
---|
91 | |
---|
92 | static int msdos_utf8_codepage_to_utf8( |
---|
93 | rtems_dosfs_convert_control *super, |
---|
94 | const char *src, |
---|
95 | size_t src_size, |
---|
96 | uint8_t *dst, |
---|
97 | size_t *dst_size |
---|
98 | ) |
---|
99 | { |
---|
100 | msdos_utf8_convert_control *self = (msdos_utf8_convert_control *) super; |
---|
101 | |
---|
102 | return msdos_utf8_convert_with_iconv( |
---|
103 | self->desc_codepage_to_utf8, |
---|
104 | src, |
---|
105 | src_size, |
---|
106 | dst, |
---|
107 | dst_size |
---|
108 | ); |
---|
109 | } |
---|
110 | |
---|
111 | static int msdos_utf8_utf8_to_codepage( |
---|
112 | rtems_dosfs_convert_control *super, |
---|
113 | const uint8_t *src, |
---|
114 | size_t src_size, |
---|
115 | char *dst, |
---|
116 | size_t *dst_size |
---|
117 | ) |
---|
118 | { |
---|
119 | msdos_utf8_convert_control *self = (msdos_utf8_convert_control *) super; |
---|
120 | |
---|
121 | return msdos_utf8_convert_with_iconv( |
---|
122 | self->desc_utf8_to_codepage, |
---|
123 | src, |
---|
124 | src_size, |
---|
125 | dst, |
---|
126 | dst_size |
---|
127 | ); |
---|
128 | } |
---|
129 | |
---|
130 | static int msdos_utf8_utf16_to_utf8( |
---|
131 | rtems_dosfs_convert_control *super, |
---|
132 | const uint16_t *src, |
---|
133 | size_t src_size, |
---|
134 | uint8_t *dst, |
---|
135 | size_t *dst_size |
---|
136 | ) |
---|
137 | { |
---|
138 | msdos_utf8_convert_control *self = (msdos_utf8_convert_control *) super; |
---|
139 | |
---|
140 | return msdos_utf8_convert_with_iconv( |
---|
141 | self->desc_utf16_to_utf8, |
---|
142 | src, |
---|
143 | src_size, |
---|
144 | dst, |
---|
145 | dst_size |
---|
146 | ); |
---|
147 | } |
---|
148 | |
---|
149 | static int msdos_utf8_utf8_to_utf16( |
---|
150 | rtems_dosfs_convert_control *super, |
---|
151 | const uint8_t *src, |
---|
152 | size_t src_size, |
---|
153 | uint16_t *dst, |
---|
154 | size_t *dst_size |
---|
155 | ) |
---|
156 | { |
---|
157 | msdos_utf8_convert_control *self = (msdos_utf8_convert_control *) super; |
---|
158 | |
---|
159 | return msdos_utf8_convert_with_iconv( |
---|
160 | self->desc_utf8_to_utf16, |
---|
161 | src, |
---|
162 | src_size, |
---|
163 | dst, |
---|
164 | dst_size |
---|
165 | ); |
---|
166 | } |
---|
167 | |
---|
168 | static int msdos_utf8proc_errmsg_to_errno( ssize_t errcode ) |
---|
169 | { |
---|
170 | int eno = 0; |
---|
171 | |
---|
172 | |
---|
173 | switch ( errcode ) { |
---|
174 | case 0: |
---|
175 | eno = 0; |
---|
176 | break; |
---|
177 | case UTF8PROC_ERROR_NOMEM: |
---|
178 | eno = ENOMEM; |
---|
179 | break; |
---|
180 | case UTF8PROC_ERROR_OVERFLOW: |
---|
181 | eno = EOVERFLOW; |
---|
182 | break; |
---|
183 | case UTF8PROC_ERROR_INVALIDUTF8: |
---|
184 | eno = EINVAL; |
---|
185 | break; |
---|
186 | case UTF8PROC_ERROR_NOTASSIGNED: |
---|
187 | eno = EINVAL; |
---|
188 | break; |
---|
189 | case UTF8PROC_ERROR_INVALIDOPTS: |
---|
190 | eno = EINVAL; |
---|
191 | break; |
---|
192 | default: |
---|
193 | eno = ENOENT; |
---|
194 | break; |
---|
195 | } |
---|
196 | |
---|
197 | return eno; |
---|
198 | } |
---|
199 | |
---|
200 | static int msdos_utf8_normalize_and_fold( |
---|
201 | rtems_dosfs_convert_control *super, |
---|
202 | const uint8_t *src, |
---|
203 | const size_t src_size, |
---|
204 | uint8_t *dst, |
---|
205 | size_t *dst_size |
---|
206 | ) |
---|
207 | { |
---|
208 | int eno = 0; |
---|
209 | int32_t *unicode_buf = (int32_t *) dst; |
---|
210 | ssize_t unicode_buf_size = *dst_size / sizeof( *unicode_buf ); |
---|
211 | ssize_t unicodes_to_reencode; |
---|
212 | ssize_t result; |
---|
213 | |
---|
214 | (void) super; |
---|
215 | |
---|
216 | result = utf8proc_decompose( |
---|
217 | src, |
---|
218 | (ssize_t) src_size, |
---|
219 | unicode_buf, |
---|
220 | unicode_buf_size, |
---|
221 | UTF8PROC_STABLE | UTF8PROC_DECOMPOSE | UTF8PROC_CASEFOLD |
---|
222 | ); |
---|
223 | |
---|
224 | if ( result >= 0 ) { |
---|
225 | if ( result <= unicode_buf_size ) { |
---|
226 | unicodes_to_reencode = result; |
---|
227 | } else { |
---|
228 | unicodes_to_reencode = unicode_buf_size; |
---|
229 | eno = ENOMEM; |
---|
230 | } |
---|
231 | |
---|
232 | result = utf8proc_reencode( |
---|
233 | unicode_buf, |
---|
234 | unicodes_to_reencode, |
---|
235 | UTF8PROC_STABLE | UTF8PROC_DECOMPOSE |
---|
236 | ); |
---|
237 | |
---|
238 | if ( result >= 0 ) { |
---|
239 | *dst_size = result; |
---|
240 | } else { |
---|
241 | eno = msdos_utf8proc_errmsg_to_errno( result ); |
---|
242 | } |
---|
243 | } else { |
---|
244 | eno = msdos_utf8proc_errmsg_to_errno( result ); |
---|
245 | } |
---|
246 | |
---|
247 | return eno; |
---|
248 | } |
---|
249 | |
---|
250 | static void msdos_utf8_destroy( |
---|
251 | rtems_dosfs_convert_control *super |
---|
252 | ) |
---|
253 | { |
---|
254 | msdos_utf8_convert_control *self = (msdos_utf8_convert_control *) super; |
---|
255 | int rv; |
---|
256 | |
---|
257 | if ( self->desc_utf16_to_utf8 != INVALID_ICONV_DESC ) { |
---|
258 | rv = iconv_close( self->desc_utf16_to_utf8 ); |
---|
259 | assert( rv == 0 ); |
---|
260 | } |
---|
261 | |
---|
262 | if ( self->desc_codepage_to_utf8 != INVALID_ICONV_DESC ) { |
---|
263 | rv = iconv_close( self->desc_codepage_to_utf8 ); |
---|
264 | assert( rv == 0 ); |
---|
265 | } |
---|
266 | |
---|
267 | if ( self->desc_utf8_to_codepage != INVALID_ICONV_DESC ) { |
---|
268 | rv = iconv_close( self->desc_utf8_to_codepage ); |
---|
269 | assert( rv == 0 ); |
---|
270 | } |
---|
271 | |
---|
272 | if ( self->desc_utf8_to_utf16 != INVALID_ICONV_DESC ) { |
---|
273 | rv = iconv_close( self->desc_utf8_to_utf16 ); |
---|
274 | assert( rv == 0 ); |
---|
275 | } |
---|
276 | |
---|
277 | free( self ); |
---|
278 | } |
---|
279 | |
---|
280 | static const rtems_dosfs_convert_handler msdos_utf8_convert_handler = { |
---|
281 | .utf8_to_codepage = msdos_utf8_utf8_to_codepage, |
---|
282 | .codepage_to_utf8 = msdos_utf8_codepage_to_utf8, |
---|
283 | .utf8_to_utf16 = msdos_utf8_utf8_to_utf16, |
---|
284 | .utf16_to_utf8 = msdos_utf8_utf16_to_utf8, |
---|
285 | .utf8_normalize_and_fold = msdos_utf8_normalize_and_fold, |
---|
286 | .destroy = msdos_utf8_destroy |
---|
287 | }; |
---|
288 | |
---|
289 | rtems_dosfs_convert_control *rtems_dosfs_create_utf8_converter( |
---|
290 | const char *codepage |
---|
291 | ) |
---|
292 | { |
---|
293 | msdos_utf8_convert_control *self = malloc( sizeof( *self ) ); |
---|
294 | |
---|
295 | if ( self != NULL ) { |
---|
296 | self->desc_codepage_to_utf8 = iconv_open( "UTF-8", codepage ); |
---|
297 | self->desc_utf8_to_codepage = iconv_open( codepage, "UTF-8" ); |
---|
298 | self->desc_utf16_to_utf8 = iconv_open( "UTF-8", "UTF-16LE" ); |
---|
299 | self->desc_utf8_to_utf16 = iconv_open( "UTF-16LE", "UTF-8" ); |
---|
300 | |
---|
301 | if ( |
---|
302 | self->desc_utf16_to_utf8 != INVALID_ICONV_DESC |
---|
303 | && self->desc_utf8_to_codepage != INVALID_ICONV_DESC |
---|
304 | && self->desc_codepage_to_utf8 != INVALID_ICONV_DESC |
---|
305 | && self->desc_utf8_to_utf16 != INVALID_ICONV_DESC |
---|
306 | ) { |
---|
307 | rtems_dosfs_convert_control *super = &self->super; |
---|
308 | |
---|
309 | super->handler = &msdos_utf8_convert_handler; |
---|
310 | super->buffer.data = &self->buffer; |
---|
311 | super->buffer.size = sizeof( self->buffer ); |
---|
312 | } else { |
---|
313 | msdos_utf8_destroy( &self->super ); |
---|
314 | self = NULL; |
---|
315 | } |
---|
316 | } |
---|
317 | |
---|
318 | return &self->super; |
---|
319 | } |
---|