1/* Find matching transformation algorithms and initialize steps.
2 Copyright (C) 1997-2020 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
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 <errno.h>
21#include <locale.h>
22#include "../locale/localeinfo.h"
23#include <stdlib.h>
24#include <string.h>
25
26#include <gconv_int.h>
27
28
29/* How many character should be converted in one call? */
30#define GCONV_NCHAR_GOAL 8160
31
32
33int
34__gconv_open (const char *toset, const char *fromset, __gconv_t *handle,
35 int flags)
36{
37 struct __gconv_step *steps;
38 size_t nsteps;
39 __gconv_t result = NULL;
40 size_t cnt = 0;
41 int res;
42 int conv_flags = 0;
43 const char *errhand;
44 const char *ignore;
45 bool translit = false;
46
47 /* Find out whether any error handling method is specified. */
48 errhand = strchr (toset, '/');
49 if (errhand != NULL)
50 errhand = strchr (errhand + 1, '/');
51 if (__glibc_likely (errhand != NULL))
52 {
53 if (*++errhand == '\0')
54 errhand = NULL;
55 else
56 {
57 /* Make copy without the error handling description. */
58 char *newtoset = (char *) alloca (errhand - toset + 1);
59 char *tok;
60 char *ptr = NULL /* Work around a bogus warning */;
61
62 newtoset[errhand - toset] = '\0';
63 toset = memcpy (newtoset, toset, errhand - toset);
64
65 /* Find the appropriate transliteration handlers. */
66 tok = strdupa (errhand);
67
68 tok = __strtok_r (tok, ",", &ptr);
69 while (tok != NULL)
70 {
71 if (__strcasecmp_l (tok, "TRANSLIT", _nl_C_locobj_ptr) == 0)
72 translit = true;
73 else if (__strcasecmp_l (tok, "IGNORE", _nl_C_locobj_ptr) == 0)
74 /* Set the flag to ignore all errors. */
75 conv_flags |= __GCONV_IGNORE_ERRORS;
76
77 tok = __strtok_r (NULL, ",", &ptr);
78 }
79 }
80 }
81
82 /* For the source character set we ignore the error handler specification.
83 XXX Is this really always the best? */
84 ignore = strchr (fromset, '/');
85 if (ignore != NULL && (ignore = strchr (ignore + 1, '/')) != NULL
86 && *++ignore != '\0')
87 {
88 char *newfromset = (char *) alloca (ignore - fromset + 1);
89
90 newfromset[ignore - fromset] = '\0';
91 fromset = memcpy (newfromset, fromset, ignore - fromset);
92 }
93
94 /* If the string is empty define this to mean the charset of the
95 currently selected locale. */
96 if (strcmp (toset, "//") == 0)
97 {
98 const char *codeset = _NL_CURRENT (LC_CTYPE, CODESET);
99 size_t len = strlen (codeset);
100 char *dest;
101 toset = dest = (char *) alloca (len + 3);
102 memcpy (__mempcpy (dest, codeset, len), "//", 3);
103 }
104 if (strcmp (fromset, "//") == 0)
105 {
106 const char *codeset = _NL_CURRENT (LC_CTYPE, CODESET);
107 size_t len = strlen (codeset);
108 char *dest;
109 fromset = dest = (char *) alloca (len + 3);
110 memcpy (__mempcpy (dest, codeset, len), "//", 3);
111 }
112
113 res = __gconv_find_transform (toset, fromset, &steps, &nsteps, flags);
114 if (res == __GCONV_OK)
115 {
116 /* Allocate room for handle. */
117 result = (__gconv_t) malloc (sizeof (struct __gconv_info)
118 + (nsteps
119 * sizeof (struct __gconv_step_data)));
120 if (result == NULL)
121 res = __GCONV_NOMEM;
122 else
123 {
124 /* Remember the list of steps. */
125 result->__steps = steps;
126 result->__nsteps = nsteps;
127
128 /* Clear the array for the step data. */
129 memset (result->__data, '\0',
130 nsteps * sizeof (struct __gconv_step_data));
131
132 /* Call all initialization functions for the transformation
133 step implementations. */
134 for (cnt = 0; cnt < nsteps; ++cnt)
135 {
136 size_t size;
137
138 /* Would have to be done if we would not clear the whole
139 array above. */
140#if 0
141 /* Reset the counter. */
142 result->__data[cnt].__invocation_counter = 0;
143
144 /* It's a regular use. */
145 result->__data[cnt].__internal_use = 0;
146#endif
147
148 /* We use the `mbstate_t' member in DATA. */
149 result->__data[cnt].__statep = &result->__data[cnt].__state;
150
151 /* The builtin transliteration handling only
152 supports the internal encoding. */
153 if (translit
154 && __strcasecmp_l (steps[cnt].__from_name,
155 "INTERNAL", _nl_C_locobj_ptr) == 0)
156 conv_flags |= __GCONV_TRANSLIT;
157
158 /* If this is the last step we must not allocate an
159 output buffer. */
160 if (cnt < nsteps - 1)
161 {
162 result->__data[cnt].__flags = conv_flags;
163
164 /* Allocate the buffer. */
165 size = (GCONV_NCHAR_GOAL * steps[cnt].__max_needed_to);
166
167 result->__data[cnt].__outbuf = malloc (size);
168 if (result->__data[cnt].__outbuf == NULL)
169 {
170 res = __GCONV_NOMEM;
171 goto bail;
172 }
173
174 result->__data[cnt].__outbufend =
175 result->__data[cnt].__outbuf + size;
176 }
177 else
178 {
179 /* Handle the last entry. */
180 result->__data[cnt].__flags = conv_flags | __GCONV_IS_LAST;
181
182 break;
183 }
184 }
185 }
186
187 if (res != __GCONV_OK)
188 {
189 /* Something went wrong. Free all the resources. */
190 int serrno;
191 bail:
192 serrno = errno;
193
194 if (result != NULL)
195 {
196 while (cnt-- > 0)
197 free (result->__data[cnt].__outbuf);
198
199 free (result);
200 result = NULL;
201 }
202
203 __gconv_close_transform (steps, nsteps);
204
205 __set_errno (serrno);
206 }
207 }
208
209 *handle = result;
210 return res;
211}
212