1/* Conversion module for UTF-16.
2 Copyright (C) 1999-2021 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Ulrich Drepper <drepper@cygnus.com>, 1999.
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#include <byteswap.h>
21#include <dlfcn.h>
22#include <gconv.h>
23#include <stddef.h>
24#include <stdint.h>
25#include <stdlib.h>
26#include <string.h>
27
28/* This is the Byte Order Mark character (BOM). */
29#define BOM 0xfeff
30/* And in the other byte order. */
31#define BOM_OE 0xfffe
32
33
34/* Definitions used in the body of the `gconv' function. */
35#define FROM_LOOP from_utf16_loop
36#define TO_LOOP to_utf16_loop
37#define DEFINE_INIT 0
38#define DEFINE_FINI 0
39#define MIN_NEEDED_FROM 2
40#define MAX_NEEDED_FROM 4
41#define MIN_NEEDED_TO 4
42#define ONE_DIRECTION 0
43#define FROM_DIRECTION (dir == from_utf16)
44#define PREPARE_LOOP \
45 enum direction dir = ((struct utf16_data *) step->__data)->dir; \
46 enum variant var = ((struct utf16_data *) step->__data)->var; \
47 if (__glibc_unlikely (data->__invocation_counter == 0)) \
48 { \
49 if (var == UTF_16) \
50 { \
51 if (FROM_DIRECTION) \
52 { \
53 /* We have to find out which byte order the file is \
54 encoded in. */ \
55 if (inptr + 2 > inend) \
56 return (inptr == inend \
57 ? __GCONV_EMPTY_INPUT : __GCONV_INCOMPLETE_INPUT); \
58 \
59 if (get16u (inptr) == BOM) \
60 /* Simply ignore the BOM character. */ \
61 *inptrp = inptr += 2; \
62 else if (get16u (inptr) == BOM_OE) \
63 { \
64 data->__flags |= __GCONV_SWAP; \
65 *inptrp = inptr += 2; \
66 } \
67 } \
68 else if (!FROM_DIRECTION && !data->__internal_use) \
69 { \
70 /* Emit the Byte Order Mark. */ \
71 if (__glibc_unlikely (outbuf + 2 > outend)) \
72 return __GCONV_FULL_OUTPUT; \
73 \
74 put16u (outbuf, BOM); \
75 outbuf += 2; \
76 } \
77 } \
78 else if ((var == UTF_16LE && BYTE_ORDER == BIG_ENDIAN) \
79 || (var == UTF_16BE && BYTE_ORDER == LITTLE_ENDIAN)) \
80 data->__flags |= __GCONV_SWAP; \
81 } \
82 const int swap = data->__flags & __GCONV_SWAP;
83#define EXTRA_LOOP_ARGS , swap
84
85
86/* Direction of the transformation. */
87enum direction
88{
89 illegal_dir,
90 to_utf16,
91 from_utf16
92};
93
94enum variant
95{
96 illegal_var,
97 UTF_16,
98 UTF_16LE,
99 UTF_16BE
100};
101
102struct utf16_data
103{
104 enum direction dir;
105 enum variant var;
106};
107
108
109extern int gconv_init (struct __gconv_step *step);
110int
111gconv_init (struct __gconv_step *step)
112{
113 /* Determine which direction. */
114 struct utf16_data *new_data;
115 enum direction dir = illegal_dir;
116 enum variant var = illegal_var;
117 int result;
118
119 if (__strcasecmp (step->__from_name, "UTF-16//") == 0)
120 {
121 dir = from_utf16;
122 var = UTF_16;
123 }
124 else if (__strcasecmp (step->__to_name, "UTF-16//") == 0)
125 {
126 dir = to_utf16;
127 var = UTF_16;
128 }
129 else if (__strcasecmp (step->__from_name, "UTF-16BE//") == 0)
130 {
131 dir = from_utf16;
132 var = UTF_16BE;
133 }
134 else if (__strcasecmp (step->__to_name, "UTF-16BE//") == 0)
135 {
136 dir = to_utf16;
137 var = UTF_16BE;
138 }
139 else if (__strcasecmp (step->__from_name, "UTF-16LE//") == 0)
140 {
141 dir = from_utf16;
142 var = UTF_16LE;
143 }
144 else if (__strcasecmp (step->__to_name, "UTF-16LE//") == 0)
145 {
146 dir = to_utf16;
147 var = UTF_16LE;
148 }
149
150 result = __GCONV_NOCONV;
151 if (__builtin_expect (dir, to_utf16) != illegal_dir)
152 {
153 new_data = (struct utf16_data *) malloc (sizeof (struct utf16_data));
154
155 result = __GCONV_NOMEM;
156 if (new_data != NULL)
157 {
158 new_data->dir = dir;
159 new_data->var = var;
160 step->__data = new_data;
161
162 if (dir == from_utf16)
163 {
164 step->__min_needed_from = MIN_NEEDED_FROM;
165 step->__max_needed_from = MAX_NEEDED_FROM;
166 step->__min_needed_to = MIN_NEEDED_TO;
167 step->__max_needed_to = MIN_NEEDED_TO;
168 }
169 else
170 {
171 step->__min_needed_from = MIN_NEEDED_TO;
172 step->__max_needed_from = MIN_NEEDED_TO;
173 step->__min_needed_to = MIN_NEEDED_FROM;
174 step->__max_needed_to = MAX_NEEDED_FROM;
175 }
176
177 step->__stateful = 0;
178
179 result = __GCONV_OK;
180 }
181 }
182
183 return result;
184}
185
186
187extern void gconv_end (struct __gconv_step *data);
188void
189gconv_end (struct __gconv_step *data)
190{
191 free (data->__data);
192}
193
194
195/* Convert from the internal (UCS4-like) format to UTF-16. */
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 c = get32 (inptr); \
203 \
204 if (__glibc_unlikely (c >= 0xd800 && c < 0xe000)) \
205 { \
206 /* Surrogate characters in UCS-4 input are not valid. \
207 We must catch this. If we let surrogates pass through, \
208 attackers could make a security hole exploit by \
209 synthesizing any desired plane 1-16 character. */ \
210 result = __GCONV_ILLEGAL_INPUT; \
211 if (! ignore_errors_p ()) \
212 break; \
213 inptr += 4; \
214 ++*irreversible; \
215 continue; \
216 } \
217 \
218 if (swap) \
219 { \
220 if (__glibc_unlikely (c >= 0x10000)) \
221 { \
222 if (__glibc_unlikely (c >= 0x110000)) \
223 { \
224 STANDARD_TO_LOOP_ERR_HANDLER (4); \
225 } \
226 \
227 /* Generate a surrogate character. */ \
228 if (__glibc_unlikely (outptr + 4 > outend)) \
229 { \
230 /* Overflow in the output buffer. */ \
231 result = __GCONV_FULL_OUTPUT; \
232 break; \
233 } \
234 \
235 put16 (outptr, bswap_16 (0xd7c0 + (c >> 10))); \
236 outptr += 2; \
237 put16 (outptr, bswap_16 (0xdc00 + (c & 0x3ff))); \
238 } \
239 else \
240 put16 (outptr, bswap_16 (c)); \
241 } \
242 else \
243 { \
244 if (__glibc_unlikely (c >= 0x10000)) \
245 { \
246 if (__glibc_unlikely (c >= 0x110000)) \
247 { \
248 STANDARD_TO_LOOP_ERR_HANDLER (4); \
249 } \
250 \
251 /* Generate a surrogate character. */ \
252 if (__glibc_unlikely (outptr + 4 > outend)) \
253 { \
254 /* Overflow in the output buffer. */ \
255 result = __GCONV_FULL_OUTPUT; \
256 break; \
257 } \
258 \
259 put16 (outptr, 0xd7c0 + (c >> 10)); \
260 outptr += 2; \
261 put16 (outptr, 0xdc00 + (c & 0x3ff)); \
262 } \
263 else \
264 put16 (outptr, c); \
265 } \
266 outptr += 2; \
267 inptr += 4; \
268 }
269#define LOOP_NEED_FLAGS
270#define EXTRA_LOOP_DECLS \
271 , int swap
272#include <iconv/loop.c>
273
274
275/* Convert from UTF-16 to the internal (UCS4-like) format. */
276#define MIN_NEEDED_INPUT MIN_NEEDED_FROM
277#define MAX_NEEDED_INPUT MAX_NEEDED_FROM
278#define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
279#define LOOPFCT FROM_LOOP
280#define BODY \
281 { \
282 uint16_t u1 = get16 (inptr); \
283 \
284 if (swap) \
285 { \
286 u1 = bswap_16 (u1); \
287 \
288 if (__builtin_expect (u1 < 0xd800, 1) || u1 > 0xdfff) \
289 { \
290 /* No surrogate. */ \
291 put32 (outptr, u1); \
292 inptr += 2; \
293 } \
294 else \
295 { \
296 uint16_t u2; \
297 \
298 if (__glibc_unlikely (u1 >= 0xdc00)) \
299 { \
300 /* This is no valid first word for a surrogate. */ \
301 STANDARD_FROM_LOOP_ERR_HANDLER (2); \
302 } \
303 \
304 /* It's a surrogate character. At least the first word says \
305 it is. */ \
306 if (__glibc_unlikely (inptr + 4 > inend)) \
307 { \
308 /* We don't have enough input for another complete input \
309 character. */ \
310 result = __GCONV_INCOMPLETE_INPUT; \
311 break; \
312 } \
313 \
314 inptr += 2; \
315 u2 = bswap_16 (get16 (inptr)); \
316 if (__builtin_expect (u2 < 0xdc00, 0) \
317 || __builtin_expect (u2 > 0xdfff, 0)) \
318 { \
319 /* This is no valid second word for a surrogate. */ \
320 inptr -= 2; \
321 STANDARD_FROM_LOOP_ERR_HANDLER (2); \
322 } \
323 \
324 put32 (outptr, ((u1 - 0xd7c0) << 10) + (u2 - 0xdc00)); \
325 inptr += 2; \
326 } \
327 } \
328 else \
329 { \
330 if (__builtin_expect (u1 < 0xd800, 1) || u1 > 0xdfff) \
331 { \
332 /* No surrogate. */ \
333 put32 (outptr, u1); \
334 inptr += 2; \
335 } \
336 else \
337 { \
338 if (__glibc_unlikely (u1 >= 0xdc00)) \
339 { \
340 /* This is no valid first word for a surrogate. */ \
341 STANDARD_FROM_LOOP_ERR_HANDLER (2); \
342 } \
343 \
344 /* It's a surrogate character. At least the first word says \
345 it is. */ \
346 if (__glibc_unlikely (inptr + 4 > inend)) \
347 { \
348 /* We don't have enough input for another complete input \
349 character. */ \
350 result = __GCONV_INCOMPLETE_INPUT; \
351 break; \
352 } \
353 \
354 inptr += 2; \
355 uint16_t u2 = get16 (inptr); \
356 if (__builtin_expect (u2 < 0xdc00, 0) \
357 || __builtin_expect (u2 > 0xdfff, 0)) \
358 { \
359 /* This is no valid second word for a surrogate. */ \
360 inptr -= 2; \
361 STANDARD_FROM_LOOP_ERR_HANDLER (2); \
362 } \
363 \
364 put32 (outptr, ((u1 - 0xd7c0) << 10) + (u2 - 0xdc00)); \
365 inptr += 2; \
366 } \
367 } \
368 outptr += 4; \
369 }
370#define LOOP_NEED_FLAGS
371#define EXTRA_LOOP_DECLS \
372 , int swap
373#include <iconv/loop.c>
374
375
376/* Now define the toplevel functions. */
377#include <iconv/skeleton.c>
378