wimlib.net Git - wimlib/blob - src/encoding.c

   1 /*
   2  * encoding.c:  Convert "multibyte" strings (the locale-default encoding---
   3  * generally, UTF-8 or something like ISO-8859-1) to UTF-16LE strings, and vice
   4  * versa.
   5  */
   6
   7 /*
   8  * Copyright (C) 2012, 2013 Eric Biggers
   9  *
  10  * This file is part of wimlib, a library for working with WIM files.
  11  *
  12  * wimlib is free software; you can redistribute it and/or modify it under the
  13  * terms of the GNU General Public License as published by the Free
  14  * Software Foundation; either version 3 of the License, or (at your option)
  15  * any later version.
  16  *
  17  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  18  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  19  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  20  * details.
  21  *
  22  * You should have received a copy of the GNU General Public License
  23  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  24  */
  25
  26 #include "config.h"
  27 #include "wimlib_internal.h"
  28 #include <pthread.h>
  29 #include "list.h"
  30
  31 #include <iconv.h>
  32 #include <stdlib.h>
  33
  34 bool wimlib_mbs_is_utf8 = false;
  35
  36 struct iconv_list_head {
  37         const char *from_encoding;
  38         const char *to_encoding;
  39         struct list_head list;
  40         pthread_mutex_t mutex;
  41 };
  42
  43 struct iconv_node {
  44         iconv_t cd;
  45         struct list_head list;
  46         struct iconv_list_head *head;
  47 };
  48
  49 #define ICONV_LIST(name, from, to)                      \
  50 struct iconv_list_head name = {                         \
  51         .from_encoding = from,                          \
  52         .to_encoding = to,                              \
  53         .list = LIST_HEAD_INIT(name.list),              \
  54         .mutex = PTHREAD_MUTEX_INITIALIZER,             \
  55 }
  56
  57 static iconv_t *
  58 get_iconv(struct iconv_list_head *head)
  59 {
  60         iconv_t cd;
  61         iconv_t *cd_p;
  62         struct iconv_node *i;
  63
  64         pthread_mutex_lock(&head->mutex);
  65         if (list_empty(&head->list)) {
  66                 cd = iconv_open(head->to_encoding, head->from_encoding);
  67                 if (cd == (iconv_t)-1) {
  68                         ERROR_WITH_ERRNO("Failed to open iconv from %s to %s",
  69                                          head->from_encoding, head->to_encoding);
  70                         cd_p = NULL;
  71                 } else {
  72                         i = MALLOC(sizeof(struct iconv_node));
  73                         if (i) {
  74                                 i->head = head;
  75                                 i->cd = cd;
  76                                 cd_p = &i->cd;
  77                         } else {
  78                                 iconv_close(cd);
  79                                 cd_p = NULL;
  80                         }
  81                 }
  82         } else {
  83                 i = container_of(head->list.next, struct iconv_node, list);
  84                 list_del(head->list.next);
  85                 cd_p = &i->cd;
  86         }
  87         pthread_mutex_unlock(&head->mutex);
  88         return cd_p;
  89 }
  90
  91 static void
  92 put_iconv(iconv_t *cd)
  93 {
  94         int errno_save = errno;
  95         struct iconv_node *i = container_of(cd, struct iconv_node, cd);
  96         struct iconv_list_head *head = i->head;
  97
  98         pthread_mutex_lock(&head->mutex);
  99         list_add(&i->list, &head->list);
 100         pthread_mutex_unlock(&head->mutex);
 101         errno = errno_save;
 102 }
 103
 104 /* Prevent printing an error message if a character conversion error occurs
 105  * while printing an error message.  (This variable is not per-thread but it
 106  * doesn't matter too much since it's just the error messages.) */
 107 static bool error_message_being_printed = false;
 108
 109 #define DEFINE_CHAR_CONVERSION_FUNCTIONS(varname1, longname1, chartype1,\
 110                                          varname2, longname2, chartype2,\
 111                                          worst_case_len_expr,           \
 112                                          err_return,                    \
 113                                          err_msg)                       \
 114 static ICONV_LIST(iconv_##varname1##_to_##varname2,                     \
 115                   longname1, longname2);                                \
 116                                                                         \
 117 int                                                                     \
 118 varname1##_to_##varname2##_nbytes(const chartype1 *in, size_t in_nbytes,\
 119                                   size_t *out_nbytes_ret)               \
 120 {                                                                       \
 121         iconv_t *cd = get_iconv(&iconv_##varname1##_to_##varname2);     \
 122         if (cd == NULL)                                                 \
 123                 return WIMLIB_ERR_ICONV_NOT_AVAILABLE;                  \
 124                                                                         \
 125         /* Worst case length */                                         \
 126         chartype2 buf[worst_case_len_expr];                             \
 127         char *inbuf = (char*)in;                                        \
 128         size_t inbytesleft = in_nbytes;                                 \
 129         char *outbuf = (char*)buf;                                      \
 130         size_t outbytesleft = sizeof(buf);                              \
 131         size_t len;                                                     \
 132         int ret;                                                        \
 133                                                                         \
 134         len = iconv(*cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft); \
 135         if (len == (size_t)-1) {                                        \
 136                 if (!error_message_being_printed) {                     \
 137                         error_message_being_printed = true;             \
 138                         err_msg;                                        \
 139                         error_message_being_printed = false;            \
 140                 }                                                       \
 141                 ret = err_return;                                       \
 142         } else {                                                        \
 143                 *out_nbytes_ret = sizeof(buf) - outbytesleft;           \
 144                 ret = 0;                                                \
 145         }                                                               \
 146         put_iconv(cd);                                                  \
 147         return ret;                                                     \
 148 }                                                                       \
 149                                                                         \
 150 int                                                                     \
 151 varname1##_to_##varname2##_buf(const chartype1 *in, size_t in_nbytes,   \
 152                                chartype2 *out)                          \
 153 {                                                                       \
 154         iconv_t *cd = get_iconv(&iconv_##varname1##_to_##varname2);     \
 155         if (cd == NULL)                                                 \
 156                 return WIMLIB_ERR_ICONV_NOT_AVAILABLE;                  \
 157                                                                         \
 158         char *inbuf = (char*)in;                                        \
 159         size_t inbytesleft = in_nbytes;                                 \
 160         char *outbuf = (char*)out;                                      \
 161         const size_t LARGE_NUMBER = 1000000000;                         \
 162         size_t outbytesleft = LARGE_NUMBER;                             \
 163         size_t len;                                                     \
 164         int ret;                                                        \
 165                                                                         \
 166         len = iconv(*cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft); \
 167         if (len == (size_t)-1) {                                        \
 168                 err_msg;                                                \
 169                 ret = err_return;                                       \
 170         } else {                                                        \
 171                 out[(LARGE_NUMBER-outbytesleft)/sizeof(chartype2)] = 0; \
 172                 ret = 0;                                                \
 173         }                                                               \
 174         put_iconv(cd);                                                  \
 175         return ret;                                                     \
 176 }                                                                       \
 177                                                                         \
 178 int                                                                     \
 179 varname1##_to_##varname2(const chartype1 *in, size_t in_nbytes,         \
 180                          chartype2 **out_ret,                           \
 181                          size_t *out_nbytes_ret)                        \
 182 {                                                                       \
 183         int ret;                                                        \
 184         chartype2 *out;                                                 \
 185         size_t out_nbytes;                                              \
 186                                                                         \
 187         ret = varname1##_to_##varname2##_nbytes(in, in_nbytes,          \
 188                                                 &out_nbytes);           \
 189         if (ret)                                                        \
 190                 return ret;                                             \
 191                                                                         \
 192         out = MALLOC(out_nbytes + sizeof(chartype2));                   \
 193         if (!out)                                                       \
 194                 return WIMLIB_ERR_NOMEM;                                \
 195                                                                         \
 196         ret = varname1##_to_##varname2##_buf(in, in_nbytes, out);       \
 197         if (ret) {                                                      \
 198                 int errno_save = errno;                                 \
 199                 FREE(out);                                              \
 200                 errno = errno_save;                                     \
 201         } else {                                                        \
 202                 *out_ret = out;                                         \
 203                 *out_nbytes_ret = out_nbytes;                           \
 204         }                                                               \
 205         return ret;                                                     \
 206 }
 207
 208 DEFINE_CHAR_CONVERSION_FUNCTIONS(utf16le, "UTF-16LE", utf16lechar,
 209                                  mbs, "", mbchar,
 210                                  in_nbytes / 2 * MB_CUR_MAX,
 211                                  WIMLIB_ERR_UNICODE_STRING_NOT_REPRESENTABLE,
 212                                  ERROR_WITH_ERRNO("Failed to convert UTF-16LE "
 213                                                   "string %U to multibyte string", in))
 214
 215 DEFINE_CHAR_CONVERSION_FUNCTIONS(mbs, "", mbchar,
 216                                  utf16le, "UTF-16LE", utf16lechar,
 217                                  in_nbytes * 2,
 218                                  WIMLIB_ERR_INVALID_MULTIBYTE_STRING,
 219                                  ERROR_WITH_ERRNO("Failed to convert multibyte "
 220                                                   "string %s to UTF-16LE string", in))
 221
 222 DEFINE_CHAR_CONVERSION_FUNCTIONS(utf8, "UTF-8", utf8char,
 223                                  mbs, "", mbchar,
 224                                  in_nbytes,
 225                                  WIMLIB_ERR_INVALID_UTF8_STRING,
 226                                  ERROR_WITH_ERRNO("Failed to convert UTF-8 "
 227                                                   "string %U to multibyte string", in))
 228
 229
 230 static void
 231 iconv_cleanup(struct iconv_list_head *head)
 232 {
 233         pthread_mutex_destroy(&head->mutex);
 234         while (!list_empty(&head->list)) {
 235                 struct iconv_node *i;
 236
 237                 i = container_of(head->list.next, struct iconv_node, list);
 238                 list_del(&i->list);
 239                 iconv_close(i->cd);
 240                 FREE(i);
 241         }
 242 }
 243
 244 void
 245 iconv_global_cleanup()
 246 {
 247         iconv_cleanup(&iconv_utf16le_to_mbs);
 248         iconv_cleanup(&iconv_mbs_to_utf16le);
 249         iconv_cleanup(&iconv_utf8_to_mbs);
 250 }
 251
 252
 253
 254 bool
 255 utf8_str_contains_nonascii_chars(const utf8char *utf8_str)
 256 {
 257         do {
 258                 if ((unsigned char)*utf8_str > 127)
 259                         return false;
 260         } while (*++utf8_str);
 261         return true;
 262 }