lib/util/charset/charset.h

   1 /*
   2    Unix SMB/CIFS implementation.
   3    charset defines
   4    Copyright (C) Andrew Tridgell 2001
   5    Copyright (C) Jelmer Vernooij 2002
   6
   7    This program is free software; you can redistribute it and/or modify
   8    it under the terms of the GNU General Public License as published by
   9    the Free Software Foundation; either version 3 of the License, or
  10    (at your option) any later version.
  11
  12    This program is distributed in the hope that it will be useful,
  13    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15    GNU General Public License for more details.
  16
  17    You should have received a copy of the GNU General Public License
  18    along with this program.  If not, see <http://www.gnu.org/licenses/>.
  19 */
  20
  21 /* This is a public header file that is installed as part of Samba.
  22  * If you remove any functions or change their signature, update
  23  * the so version number. */
  24
  25 #ifndef __CHARSET_H__
  26 #define __CHARSET_H__
  27
  28 #include <talloc.h>
  29
  30 /* this defines the charset types used in samba */
  31 typedef enum {CH_UTF16LE=0, CH_UTF16=0, CH_UNIX, CH_DISPLAY, CH_DOS, CH_UTF8, CH_UTF16BE, CH_UTF16MUNGED} charset_t;
  32
  33 #define NUM_CHARSETS 7
  34
  35 /*
  36  * SMB UCS2 (16-bit unicode) internal type.
  37  * smb_ucs2_t is *always* in little endian format.
  38  */
  39
  40 typedef uint16_t smb_ucs2_t;
  41
  42 /*
  43  *   for each charset we have a function that pulls from that charset to
  44  *     a ucs2 buffer, and a function that pushes to a ucs2 buffer
  45  *     */
  46
  47 struct charset_functions {
  48         const char *name;
  49         size_t (*pull)(void *, const char **inbuf, size_t *inbytesleft,
  50                                    char **outbuf, size_t *outbytesleft);
  51         size_t (*push)(void *, const char **inbuf, size_t *inbytesleft,
  52                                    char **outbuf, size_t *outbytesleft);
  53         struct charset_functions *prev, *next;
  54 };
  55
  56 /* this type is used for manipulating unicode codepoints */
  57 typedef uint32_t codepoint_t;
  58
  59 #define INVALID_CODEPOINT ((codepoint_t)-1)
  60
  61 /*
  62  * This is auxiliary struct used by source/script/gen-8-bit-gap.sh script
  63  * during generation of an encoding table for charset module
  64  *     */
  65
  66 struct charset_gap_table {
  67   uint16_t start;
  68   uint16_t end;
  69   int32_t idx;
  70 };
  71
  72
  73 /* generic iconv conversion structure */
  74 typedef struct smb_iconv_s {
  75         size_t (*direct)(void *cd, const char **inbuf, size_t *inbytesleft,
  76                          char **outbuf, size_t *outbytesleft);
  77         size_t (*pull)(void *cd, const char **inbuf, size_t *inbytesleft,
  78                        char **outbuf, size_t *outbytesleft);
  79         size_t (*push)(void *cd, const char **inbuf, size_t *inbytesleft,
  80                        char **outbuf, size_t *outbytesleft);
  81         void *cd_direct, *cd_pull, *cd_push;
  82         char *from_name, *to_name;
  83 } *smb_iconv_t;
  84
  85 /* string manipulation flags */
  86 #define STR_TERMINATE 1
  87 #define STR_UPPER 2
  88 #define STR_ASCII 4
  89 #define STR_UNICODE 8
  90 #define STR_NOALIGN 16
  91 #define STR_NO_RANGE_CHECK 32
  92 #define STR_LEN8BIT 64
  93 #define STR_TERMINATE_ASCII 128 /* only terminate if ascii */
  94 #define STR_LEN_NOTERM 256 /* the length field is the unterminated length */
  95
  96 struct loadparm_context;
  97 struct smb_iconv_convenience;
  98
  99 /* replace some string functions with multi-byte
 100    versions */
 101 #define strlower(s) strlower_m(s)
 102 #define strupper(s) strupper_m(s)
 103
 104 char *strchr_m(const char *s, char c);
 105 size_t strlen_m_term(const char *s);
 106 size_t strlen_m_term_null(const char *s);
 107 size_t strlen_m(const char *s);
 108 char *alpha_strcpy(char *dest, const char *src, const char *other_safe_chars, size_t maxlength);
 109 void string_replace_m(char *s, char oldc, char newc);
 110 bool strcsequal_m(const char *s1,const char *s2);
 111 bool strequal_m(const char *s1, const char *s2);
 112 int strncasecmp_m(const char *s1, const char *s2, size_t n);
 113 bool next_token(const char **ptr,char *buff, const char *sep, size_t bufsize);
 114 int strcasecmp_m(const char *s1, const char *s2);
 115 size_t count_chars_m(const char *s, char c);
 116 void strupper_m(char *s);
 117 void strlower_m(char *s);
 118 char *strupper_talloc(TALLOC_CTX *ctx, const char *src);
 119 char *talloc_strdup_upper(TALLOC_CTX *ctx, const char *src);
 120 char *strupper_talloc_n(TALLOC_CTX *ctx, const char *src, size_t n);
 121 char *strlower_talloc(TALLOC_CTX *ctx, const char *src);
 122 bool strhasupper(const char *string);
 123 bool strhaslower(const char *string);
 124 char *strrchr_m(const char *s, char c);
 125 char *strchr_m(const char *s, char c);
 126
 127 bool push_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 128 bool push_ucs2_talloc(TALLOC_CTX *ctx, smb_ucs2_t **dest, const char *src, size_t *converted_size);
 129 bool push_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 130 bool pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 131 bool pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const smb_ucs2_t *src, size_t *converted_size);
 132 bool pull_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 133 ssize_t push_string(void *dest, const char *src, size_t dest_len, int flags);
 134 ssize_t pull_string(char *dest, const void *src, size_t dest_len, size_t src_len, int flags);
 135
 136 bool convert_string_talloc(TALLOC_CTX *ctx,
 137                                        charset_t from, charset_t to,
 138                                        void const *src, size_t srclen,
 139                                        void *dest, size_t *converted_size,
 140                                            bool allow_badcharcnv);
 141
 142 size_t convert_string(charset_t from, charset_t to,
 143                                 void const *src, size_t srclen,
 144                                 void *dest, size_t destlen, bool allow_badcharcnv);
 145
 146 ssize_t iconv_talloc(TALLOC_CTX *mem_ctx,
 147                                        smb_iconv_t cd,
 148                                        void const *src, size_t srclen,
 149                                        void *dest);
 150
 151 extern struct smb_iconv_convenience *global_iconv_convenience;
 152
 153 codepoint_t next_codepoint(const char *str, size_t *size);
 154 ssize_t push_codepoint(char *str, codepoint_t c);
 155
 156 /* codepoints */
 157 codepoint_t next_codepoint_convenience(struct smb_iconv_convenience *ic,
 158                             const char *str, size_t *size);
 159 ssize_t push_codepoint_convenience(struct smb_iconv_convenience *ic,
 160                                 char *str, codepoint_t c);
 161 codepoint_t toupper_m(codepoint_t val);
 162 codepoint_t tolower_m(codepoint_t val);
 163 int codepoint_cmpi(codepoint_t c1, codepoint_t c2);
 164
 165 /* Iconv convenience functions */
 166 struct smb_iconv_convenience *smb_iconv_convenience_init(TALLOC_CTX *mem_ctx,
 167                                                          const char *dos_charset,
 168                                                          const char *unix_charset,
 169                                                          bool native_iconv);
 170
 171 bool convert_string_convenience(struct smb_iconv_convenience *ic,
 172                                 charset_t from, charset_t to,
 173                                 void const *src, size_t srclen,
 174                                 void *dest, size_t destlen, size_t *converted_size,
 175                                 bool allow_badcharcnv);
 176 bool convert_string_talloc_convenience(TALLOC_CTX *ctx,
 177                                        struct smb_iconv_convenience *ic,
 178                                        charset_t from, charset_t to,
 179                                        void const *src, size_t srclen,
 180                                        void *dest, size_t *converted_size, bool allow_badcharcnv);
 181 /* iconv */
 182 smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode);
 183 int smb_iconv_close(smb_iconv_t cd);
 184 size_t smb_iconv(smb_iconv_t cd,
 185                  const char **inbuf, size_t *inbytesleft,
 186                  char **outbuf, size_t *outbytesleft);
 187 smb_iconv_t smb_iconv_open_ex(TALLOC_CTX *mem_ctx, const char *tocode,
 188                               const char *fromcode, bool native_iconv);
 189
 190 void load_case_tables(void);
 191 bool charset_register_backend(const void *_funcs);
 192
 193 /*
 194  *   Define stub for charset module which implements 8-bit encoding with gaps.
 195  *   Encoding tables for such module should be produced from glibc's CHARMAPs
 196  *   using script source/script/gen-8bit-gap.sh
 197  *   CHARSETNAME is CAPITALIZED charset name
 198  *
 199  *     */
 200 #define SMB_GENERATE_CHARSET_MODULE_8_BIT_GAP(CHARSETNAME)                                      \
 201 static size_t CHARSETNAME ## _push(void *cd, const char **inbuf, size_t *inbytesleft,                   \
 202                          char **outbuf, size_t *outbytesleft)                                   \
 203 {                                                                                               \
 204         while (*inbytesleft >= 2 && *outbytesleft >= 1) {                                       \
 205                 int i;                                                                          \
 206                 int done = 0;                                                                   \
 207                                                                                                 \
 208                 uint16 ch = SVAL(*inbuf,0);                                                     \
 209                                                                                                 \
 210                 for (i=0; from_idx[i].start != 0xffff; i++) {                                   \
 211                         if ((from_idx[i].start <= ch) && (from_idx[i].end >= ch)) {             \
 212                                 ((unsigned char*)(*outbuf))[0] = from_ucs2[from_idx[i].idx+ch]; \
 213                                 (*inbytesleft) -= 2;                                            \
 214                                 (*outbytesleft) -= 1;                                           \
 215                                 (*inbuf)  += 2;                                                 \
 216                                 (*outbuf) += 1;                                                 \
 217                                 done = 1;                                                       \
 218                                 break;                                                          \
 219                         }                                                                       \
 220                 }                                                                               \
 221                 if (!done) {                                                                    \
 222                         errno = EINVAL;                                                         \
 223                         return -1;                                                              \
 224                 }                                                                               \
 225                                                                                                 \
 226         }                                                                                       \
 227                                                                                                 \
 228         if (*inbytesleft == 1) {                                                                \
 229                 errno = EINVAL;                                                                 \
 230                 return -1;                                                                      \
 231         }                                                                                       \
 232                                                                                                 \
 233         if (*inbytesleft > 1) {                                                                 \
 234                 errno = E2BIG;                                                                  \
 235                 return -1;                                                                      \
 236         }                                                                                       \
 237                                                                                                 \
 238         return 0;                                                                               \
 239 }                                                                                               \
 240                                                                                                 \
 241 static size_t CHARSETNAME ## _pull(void *cd, const char **inbuf, size_t *inbytesleft,                           \
 242                          char **outbuf, size_t *outbytesleft)                                   \
 243 {                                                                                               \
 244         while (*inbytesleft >= 1 && *outbytesleft >= 2) {                                       \
 245                 *(uint16*)(*outbuf) = to_ucs2[((unsigned char*)(*inbuf))[0]];                   \
 246                 (*inbytesleft)  -= 1;                                                           \
 247                 (*outbytesleft) -= 2;                                                           \
 248                 (*inbuf)  += 1;                                                                 \
 249                 (*outbuf) += 2;                                                                 \
 250         }                                                                                       \
 251                                                                                                 \
 252         if (*inbytesleft > 0) {                                                                 \
 253                 errno = E2BIG;                                                                  \
 254                 return -1;                                                                      \
 255         }                                                                                       \
 256                                                                                                 \
 257         return 0;                                                                               \
 258 }                                                                                               \
 259                                                                                                 \
 260 struct charset_functions CHARSETNAME ## _functions =                                            \
 261                 {#CHARSETNAME, CHARSETNAME ## _pull, CHARSETNAME ## _push};                     \
 262                                                                                                 \
 263 NTSTATUS charset_ ## CHARSETNAME ## _init(void);                                                        \
 264 NTSTATUS charset_ ## CHARSETNAME ## _init(void)                                                 \
 265 {                                                                                               \
 266         return smb_register_charset(& CHARSETNAME ## _functions);                               \
 267 }                                                                                               \
 268
 269
 270 #endif /* __CHARSET_H__ */