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_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 #ifdef WORDS_BIGENDIAN
  43 #define UCS2_SHIFT 8
  44 #else
  45 #define UCS2_SHIFT 0
  46 #endif
  47
  48 /* turn a 7 bit character into a ucs2 character */
  49 #define UCS2_CHAR(c) ((c) << UCS2_SHIFT)
  50
  51 /*
  52  *   for each charset we have a function that pulls from that charset to
  53  *     a ucs2 buffer, and a function that pushes to a ucs2 buffer
  54  *     */
  55
  56 struct charset_functions {
  57         const char *name;
  58         size_t (*pull)(void *, const char **inbuf, size_t *inbytesleft,
  59                                    char **outbuf, size_t *outbytesleft);
  60         size_t (*push)(void *, const char **inbuf, size_t *inbytesleft,
  61                                    char **outbuf, size_t *outbytesleft);
  62         struct charset_functions *prev, *next;
  63 };
  64
  65 /* this type is used for manipulating unicode codepoints */
  66 typedef uint32_t codepoint_t;
  67
  68 #define INVALID_CODEPOINT ((codepoint_t)-1)
  69
  70 /*
  71  * This is auxiliary struct used by source/script/gen-8-bit-gap.sh script
  72  * during generation of an encoding table for charset module
  73  *     */
  74
  75 struct charset_gap_table {
  76   uint16_t start;
  77   uint16_t end;
  78   int32_t idx;
  79 };
  80
  81
  82 /* generic iconv conversion structure */
  83 typedef struct smb_iconv_s {
  84         size_t (*direct)(void *cd, const char **inbuf, size_t *inbytesleft,
  85                          char **outbuf, size_t *outbytesleft);
  86         size_t (*pull)(void *cd, const char **inbuf, size_t *inbytesleft,
  87                        char **outbuf, size_t *outbytesleft);
  88         size_t (*push)(void *cd, const char **inbuf, size_t *inbytesleft,
  89                        char **outbuf, size_t *outbytesleft);
  90         void *cd_direct, *cd_pull, *cd_push;
  91         char *from_name, *to_name;
  92 } *smb_iconv_t;
  93
  94 /* string manipulation flags */
  95 #define STR_TERMINATE 1
  96 #define STR_UPPER 2
  97 #define STR_ASCII 4
  98 #define STR_UNICODE 8
  99 #define STR_NOALIGN 16
 100 #define STR_NO_RANGE_CHECK 32
 101 #define STR_LEN8BIT 64
 102 #define STR_TERMINATE_ASCII 128 /* only terminate if ascii */
 103 #define STR_LEN_NOTERM 256 /* the length field is the unterminated length */
 104
 105 struct loadparm_context;
 106 struct smb_iconv_handle;
 107
 108 char *strchr_m(const char *s, char c);
 109 /**
 110  * Calculate the number of units (8 or 16-bit, depending on the
 111  * destination charset), that would be needed to convert the input
 112  * string which is expected to be in in src_charset encoding to the
 113  * destination charset (which should be a unicode charset).
 114  */
 115 size_t strlen_m_ext_handle(struct smb_iconv_handle *ic,
 116                            const char *s, charset_t src_charset, charset_t dst_charset);
 117 size_t strlen_m_ext(const char *s, charset_t src_charset, charset_t dst_charset);
 118 size_t strlen_m_ext_term(const char *s, charset_t src_charset,
 119                          charset_t dst_charset);
 120 size_t strlen_m_term(const char *s);
 121 size_t strlen_m_term_null(const char *s);
 122 size_t strlen_m(const char *s);
 123 char *alpha_strcpy(char *dest, const char *src, const char *other_safe_chars, size_t maxlength);
 124 void string_replace_m(char *s, char oldc, char newc);
 125 bool strcsequal(const char *s1,const char *s2);
 126 bool strequal_m(const char *s1, const char *s2);
 127 int strncasecmp_m(const char *s1, const char *s2, size_t n);
 128 int strncasecmp_m_handle(struct smb_iconv_handle *iconv_handle,
 129                          const char *s1, const char *s2, size_t n);
 130 bool next_token(const char **ptr,char *buff, const char *sep, size_t bufsize);
 131 int strcasecmp_m_handle(struct smb_iconv_handle *iconv_handle,
 132                         const char *s1, const char *s2);
 133 int strcasecmp_m(const char *s1, const char *s2);
 134 size_t count_chars_m(const char *s, char c);
 135 char *strupper_talloc(TALLOC_CTX *ctx, const char *src);
 136 char *talloc_strdup_upper(TALLOC_CTX *ctx, const char *src);
 137 char *strupper_talloc_n_handle(struct smb_iconv_handle *iconv_handle,
 138                                 TALLOC_CTX *ctx, const char *src, size_t n);
 139 char *strupper_talloc_n(TALLOC_CTX *ctx, const char *src, size_t n);
 140  char *strlower_talloc_handle(struct smb_iconv_handle *iconv_handle,
 141                               TALLOC_CTX *ctx, const char *src);
 142 char *strlower_talloc(TALLOC_CTX *ctx, const char *src);
 143 bool strhasupper(const char *string);
 144 bool strhaslower_handle(struct smb_iconv_handle *ic,
 145                         const char *string);
 146 bool strhaslower(const char *string);
 147 bool strhasupper_handle(struct smb_iconv_handle *ic,
 148                         const char *string);
 149 char *strrchr_m(const char *s, char c);
 150 char *strchr_m(const char *s, char c);
 151 char *strstr_m(const char *src, const char *findstr);
 152
 153 bool push_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 154 bool push_ucs2_talloc(TALLOC_CTX *ctx, smb_ucs2_t **dest, const char *src, size_t *converted_size);
 155 bool push_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 156 bool pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 157 bool pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const smb_ucs2_t *src, size_t *converted_size);
 158 bool pull_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 159 ssize_t push_string(void *dest, const char *src, size_t dest_len, int flags);
 160 ssize_t pull_string(char *dest, const void *src, size_t dest_len, size_t src_len, int flags);
 161
 162 bool convert_string_talloc(TALLOC_CTX *ctx,
 163                            charset_t from, charset_t to,
 164                            void const *src, size_t srclen,
 165                            void *dest, size_t *converted_size);
 166
 167 bool convert_string(charset_t from, charset_t to,
 168                       void const *src, size_t srclen,
 169                       void *dest, size_t destlen,
 170                       size_t *converted_size);
 171 bool convert_string_error(charset_t from, charset_t to,
 172                           void const *src, size_t srclen,
 173                           void *dest, size_t destlen,
 174                           size_t *converted_size);
 175
 176 ssize_t iconv_talloc(TALLOC_CTX *mem_ctx,
 177                                        smb_iconv_t cd,
 178                                        void const *src, size_t srclen,
 179                                        void *dest);
 180
 181 extern struct smb_iconv_handle *global_iconv_handle;
 182 struct smb_iconv_handle *get_iconv_handle(void);
 183 struct smb_iconv_handle *get_iconv_testing_handle(TALLOC_CTX *mem_ctx,
 184                                                   const char *dos_charset,
 185                                                   const char *unix_charset);
 186 smb_iconv_t get_conv_handle(struct smb_iconv_handle *ic,
 187                             charset_t from, charset_t to);
 188 const char *charset_name(struct smb_iconv_handle *ic, charset_t ch);
 189
 190 codepoint_t next_codepoint_ext(const char *str, charset_t src_charset,
 191                                size_t *size);
 192 codepoint_t next_codepoint(const char *str, size_t *size);
 193 ssize_t push_codepoint(char *str, codepoint_t c);
 194
 195 /* codepoints */
 196 codepoint_t next_codepoint_handle_ext(struct smb_iconv_handle *ic,
 197                             const char *str, charset_t src_charset,
 198                             size_t *size);
 199 codepoint_t next_codepoint_handle(struct smb_iconv_handle *ic,
 200                             const char *str, size_t *size);
 201 ssize_t push_codepoint_handle(struct smb_iconv_handle *ic,
 202                                 char *str, codepoint_t c);
 203
 204 codepoint_t toupper_m(codepoint_t val);
 205 codepoint_t tolower_m(codepoint_t val);
 206 bool islower_m(codepoint_t val);
 207 bool isupper_m(codepoint_t val);
 208 int codepoint_cmpi(codepoint_t c1, codepoint_t c2);
 209
 210 /* Iconv convenience functions */
 211 struct smb_iconv_handle *smb_iconv_handle_reinit(TALLOC_CTX *mem_ctx,
 212                                                            const char *dos_charset,
 213                                                            const char *unix_charset,
 214                                                            bool native_iconv,
 215                                                            struct smb_iconv_handle *old_ic);
 216
 217 bool convert_string_handle(struct smb_iconv_handle *ic,
 218                                 charset_t from, charset_t to,
 219                                 void const *src, size_t srclen,
 220                                 void *dest, size_t destlen, size_t *converted_size);
 221 bool convert_string_error_handle(struct smb_iconv_handle *ic,
 222                                  charset_t from, charset_t to,
 223                                  void const *src, size_t srclen,
 224                                  void *dest, size_t destlen,
 225                                  size_t *converted_size);
 226
 227 bool convert_string_talloc_handle(TALLOC_CTX *ctx,
 228                                        struct smb_iconv_handle *ic,
 229                                        charset_t from, charset_t to,
 230                                        void const *src, size_t srclen,
 231                                        void *dest, size_t *converted_size);
 232 /* iconv */
 233 smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode);
 234 int smb_iconv_close(smb_iconv_t cd);
 235 size_t smb_iconv(smb_iconv_t cd,
 236                  const char **inbuf, size_t *inbytesleft,
 237                  char **outbuf, size_t *outbytesleft);
 238 smb_iconv_t smb_iconv_open_ex(TALLOC_CTX *mem_ctx, const char *tocode,
 239                               const char *fromcode, bool native_iconv);
 240
 241 void load_case_tables(void);
 242 void load_case_tables_library(void);
 243 bool smb_register_charset(const struct charset_functions *funcs_in);
 244
 245 /* The following definitions come from util_unistr_w.c  */
 246
 247 size_t strlen_w(const smb_ucs2_t *src);
 248 size_t strnlen_w(const smb_ucs2_t *src, size_t max);
 249 smb_ucs2_t *strchr_w(const smb_ucs2_t *s, smb_ucs2_t c);
 250 smb_ucs2_t *strchr_wa(const smb_ucs2_t *s, char c);
 251 smb_ucs2_t *strrchr_w(const smb_ucs2_t *s, smb_ucs2_t c);
 252 smb_ucs2_t *strnrchr_w(const smb_ucs2_t *s, smb_ucs2_t c, unsigned int n);
 253 smb_ucs2_t *strstr_w(const smb_ucs2_t *s, const smb_ucs2_t *ins);
 254 bool strlower_w(smb_ucs2_t *s);
 255 bool strupper_w(smb_ucs2_t *s);
 256 int strcmp_w(const smb_ucs2_t *a, const smb_ucs2_t *b);
 257 int strcasecmp_w(const smb_ucs2_t *a, const smb_ucs2_t *b);
 258 int strncasecmp_w(const smb_ucs2_t *a, const smb_ucs2_t *b, size_t len);
 259 int strcmp_wa(const smb_ucs2_t *a, const char *b);
 260 int toupper_ascii(int c);
 261 int tolower_ascii(int c);
 262 int isupper_ascii(int c);
 263 int islower_ascii(int c);
 264
 265 /*
 266  *   Define stub for charset module which implements 8-bit encoding with gaps.
 267  *   Encoding tables for such module should be produced from glibc's CHARMAPs
 268  *   using script source/script/gen-8bit-gap.sh
 269  *   CHARSETNAME is CAPITALIZED charset name
 270  *
 271  *     */
 272 #define SMB_GENERATE_CHARSET_MODULE_8_BIT_GAP(CHARSETNAME)                                      \
 273 static size_t CHARSETNAME ## _push(void *cd, const char **inbuf, size_t *inbytesleft,                   \
 274                          char **outbuf, size_t *outbytesleft)                                   \
 275 {                                                                                               \
 276         while (*inbytesleft >= 2 && *outbytesleft >= 1) {                                       \
 277                 int i;                                                                          \
 278                 int done = 0;                                                                   \
 279                                                                                                 \
 280                 uint16_t ch = SVAL(*inbuf,0);                                                   \
 281                                                                                                 \
 282                 for (i=0; from_idx[i].start != 0xffff; i++) {                                   \
 283                         if ((from_idx[i].start <= ch) && (from_idx[i].end >= ch)) {             \
 284                                 ((unsigned char*)(*outbuf))[0] = from_ucs2[from_idx[i].idx+ch]; \
 285                                 (*inbytesleft) -= 2;                                            \
 286                                 (*outbytesleft) -= 1;                                           \
 287                                 (*inbuf)  += 2;                                                 \
 288                                 (*outbuf) += 1;                                                 \
 289                                 done = 1;                                                       \
 290                                 break;                                                          \
 291                         }                                                                       \
 292                 }                                                                               \
 293                 if (!done) {                                                                    \
 294                         errno = EINVAL;                                                         \
 295                         return -1;                                                              \
 296                 }                                                                               \
 297                                                                                                 \
 298         }                                                                                       \
 299                                                                                                 \
 300         if (*inbytesleft == 1) {                                                                \
 301                 errno = EINVAL;                                                                 \
 302                 return -1;                                                                      \
 303         }                                                                                       \
 304                                                                                                 \
 305         if (*inbytesleft > 1) {                                                                 \
 306                 errno = E2BIG;                                                                  \
 307                 return -1;                                                                      \
 308         }                                                                                       \
 309                                                                                                 \
 310         return 0;                                                                               \
 311 }                                                                                               \
 312                                                                                                 \
 313 static size_t CHARSETNAME ## _pull(void *cd, const char **inbuf, size_t *inbytesleft,                           \
 314                          char **outbuf, size_t *outbytesleft)                                   \
 315 {                                                                                               \
 316         while (*inbytesleft >= 1 && *outbytesleft >= 2) {                                       \
 317                 SSVAL(*outbuf, 0, to_ucs2[((unsigned char*)(*inbuf))[0]]);                      \
 318                 (*inbytesleft)  -= 1;                                                           \
 319                 (*outbytesleft) -= 2;                                                           \
 320                 (*inbuf)  += 1;                                                                 \
 321                 (*outbuf) += 2;                                                                 \
 322         }                                                                                       \
 323                                                                                                 \
 324         if (*inbytesleft > 0) {                                                                 \
 325                 errno = E2BIG;                                                                  \
 326                 return -1;                                                                      \
 327         }                                                                                       \
 328                                                                                                 \
 329         return 0;                                                                               \
 330 }                                                                                               \
 331                                                                                                 \
 332 struct charset_functions CHARSETNAME ## _functions =                                            \
 333                 {#CHARSETNAME, CHARSETNAME ## _pull, CHARSETNAME ## _push};                     \
 334                                                                                                 \
 335 NTSTATUS charset_ ## CHARSETNAME ## _init(void);                                                        \
 336 NTSTATUS charset_ ## CHARSETNAME ## _init(void)                                                 \
 337 {                                                                                               \
 338         if (!smb_register_charset(& CHARSETNAME ## _functions)) {       \
 339                 return NT_STATUS_INTERNAL_ERROR;                        \
 340         }                                                               \
 341         return NT_STATUS_OK; \
 342 }                                               \
 343
 344
 345 #endif /* __CHARSET_H__ */