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 #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 /* replace some string functions with multi-byte
 109    versions */
 110 #define strlower(s) strlower_m(s)
 111 #define strupper(s) strupper_m(s)
 112
 113 char *strchr_m(const char *s, char c);
 114 /**
 115  * Calculate the number of units (8 or 16-bit, depending on the
 116  * destination charset), that would be needed to convert the input
 117  * string which is expected to be in in src_charset encoding to the
 118  * destination charset (which should be a unicode charset).
 119  */
 120 size_t strlen_m_ext_handle(struct smb_iconv_handle *ic,
 121                            const char *s, charset_t src_charset, charset_t dst_charset);
 122 size_t strlen_m_ext(const char *s, charset_t src_charset, charset_t dst_charset);
 123 size_t strlen_m_ext_term(const char *s, charset_t src_charset,
 124                          charset_t dst_charset);
 125 size_t strlen_m_term(const char *s);
 126 size_t strlen_m_term_null(const char *s);
 127 size_t strlen_m(const char *s);
 128 char *alpha_strcpy(char *dest, const char *src, const char *other_safe_chars, size_t maxlength);
 129 void string_replace_m(char *s, char oldc, char newc);
 130 bool strcsequal(const char *s1,const char *s2);
 131 bool strequal_m(const char *s1, const char *s2);
 132 int strncasecmp_m(const char *s1, const char *s2, size_t n);
 133 int strncasecmp_m_handle(struct smb_iconv_handle *iconv_handle,
 134                          const char *s1, const char *s2, size_t n);
 135 bool next_token(const char **ptr,char *buff, const char *sep, size_t bufsize);
 136 int strcasecmp_m_handle(struct smb_iconv_handle *iconv_handle,
 137                         const char *s1, const char *s2);
 138 int strcasecmp_m(const char *s1, const char *s2);
 139 size_t count_chars_m(const char *s, char c);
 140 void strupper_m(char *s);
 141 void strlower_m(char *s);
 142 char *strupper_talloc(TALLOC_CTX *ctx, const char *src);
 143 char *talloc_strdup_upper(TALLOC_CTX *ctx, const char *src);
 144 char *strupper_talloc_n_handle(struct smb_iconv_handle *iconv_handle,
 145                                 TALLOC_CTX *ctx, const char *src, size_t n);
 146 char *strupper_talloc_n(TALLOC_CTX *ctx, const char *src, size_t n);
 147  char *strlower_talloc_handle(struct smb_iconv_handle *iconv_handle,
 148                               TALLOC_CTX *ctx, const char *src);
 149 char *strlower_talloc(TALLOC_CTX *ctx, const char *src);
 150 bool strhasupper(const char *string);
 151 bool strhaslower_handle(struct smb_iconv_handle *ic,
 152                         const char *string);
 153 bool strhaslower(const char *string);
 154 bool strhasupper_handle(struct smb_iconv_handle *ic,
 155                         const char *string);
 156 char *strrchr_m(const char *s, char c);
 157 char *strchr_m(const char *s, char c);
 158 char *strstr_m(const char *src, const char *findstr);
 159
 160 bool push_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 161 bool push_ucs2_talloc(TALLOC_CTX *ctx, smb_ucs2_t **dest, const char *src, size_t *converted_size);
 162 bool push_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 163 bool pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 164 bool pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const smb_ucs2_t *src, size_t *converted_size);
 165 bool pull_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
 166 ssize_t push_string(void *dest, const char *src, size_t dest_len, int flags);
 167 ssize_t pull_string(char *dest, const void *src, size_t dest_len, size_t src_len, int flags);
 168
 169 bool convert_string_talloc(TALLOC_CTX *ctx,
 170                            charset_t from, charset_t to,
 171                            void const *src, size_t srclen,
 172                            void *dest, size_t *converted_size);
 173
 174 bool convert_string(charset_t from, charset_t to,
 175                       void const *src, size_t srclen,
 176                       void *dest, size_t destlen,
 177                       size_t *converted_size);
 178 bool convert_string_error(charset_t from, charset_t to,
 179                           void const *src, size_t srclen,
 180                           void *dest, size_t destlen,
 181                           size_t *converted_size);
 182
 183 ssize_t iconv_talloc(TALLOC_CTX *mem_ctx,
 184                                        smb_iconv_t cd,
 185                                        void const *src, size_t srclen,
 186                                        void *dest);
 187
 188 extern struct smb_iconv_handle *global_iconv_handle;
 189 struct smb_iconv_handle *get_iconv_handle(void);
 190 struct smb_iconv_handle *get_iconv_testing_handle(TALLOC_CTX *mem_ctx,
 191                                                   const char *dos_charset,
 192                                                   const char *unix_charset,
 193                                                   const char *display_charset);
 194 smb_iconv_t get_conv_handle(struct smb_iconv_handle *ic,
 195                             charset_t from, charset_t to);
 196 const char *charset_name(struct smb_iconv_handle *ic, charset_t ch);
 197
 198 codepoint_t next_codepoint_ext(const char *str, charset_t src_charset,
 199                                size_t *size);
 200 codepoint_t next_codepoint(const char *str, size_t *size);
 201 ssize_t push_codepoint(char *str, codepoint_t c);
 202
 203 /* codepoints */
 204 codepoint_t next_codepoint_handle_ext(struct smb_iconv_handle *ic,
 205                             const char *str, charset_t src_charset,
 206                             size_t *size);
 207 codepoint_t next_codepoint_handle(struct smb_iconv_handle *ic,
 208                             const char *str, size_t *size);
 209 ssize_t push_codepoint_handle(struct smb_iconv_handle *ic,
 210                                 char *str, codepoint_t c);
 211
 212 codepoint_t toupper_m(codepoint_t val);
 213 codepoint_t tolower_m(codepoint_t val);
 214 bool islower_m(codepoint_t val);
 215 bool isupper_m(codepoint_t val);
 216 int codepoint_cmpi(codepoint_t c1, codepoint_t c2);
 217
 218 /* Iconv convenience functions */
 219 struct smb_iconv_handle *smb_iconv_handle_reinit(TALLOC_CTX *mem_ctx,
 220                                                            const char *dos_charset,
 221                                                            const char *unix_charset,
 222                                                            const char *display_charset,
 223                                                            bool native_iconv,
 224                                                            struct smb_iconv_handle *old_ic);
 225
 226 bool convert_string_handle(struct smb_iconv_handle *ic,
 227                                 charset_t from, charset_t to,
 228                                 void const *src, size_t srclen,
 229                                 void *dest, size_t destlen, size_t *converted_size);
 230 bool convert_string_error_handle(struct smb_iconv_handle *ic,
 231                                  charset_t from, charset_t to,
 232                                  void const *src, size_t srclen,
 233                                  void *dest, size_t destlen,
 234                                  size_t *converted_size);
 235
 236 bool convert_string_talloc_handle(TALLOC_CTX *ctx,
 237                                        struct smb_iconv_handle *ic,
 238                                        charset_t from, charset_t to,
 239                                        void const *src, size_t srclen,
 240                                        void *dest, size_t *converted_size);
 241 /* iconv */
 242 smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode);
 243 int smb_iconv_close(smb_iconv_t cd);
 244 size_t smb_iconv(smb_iconv_t cd,
 245                  const char **inbuf, size_t *inbytesleft,
 246                  char **outbuf, size_t *outbytesleft);
 247 smb_iconv_t smb_iconv_open_ex(TALLOC_CTX *mem_ctx, const char *tocode,
 248                               const char *fromcode, bool native_iconv);
 249
 250 void load_case_tables(void);
 251 void load_case_tables_library(void);
 252 bool smb_register_charset(const struct charset_functions *funcs_in);
 253
 254 /* The following definitions come from util_unistr_w.c  */
 255
 256 size_t strlen_w(const smb_ucs2_t *src);
 257 size_t strnlen_w(const smb_ucs2_t *src, size_t max);
 258 smb_ucs2_t *strchr_w(const smb_ucs2_t *s, smb_ucs2_t c);
 259 smb_ucs2_t *strchr_wa(const smb_ucs2_t *s, char c);
 260 smb_ucs2_t *strrchr_w(const smb_ucs2_t *s, smb_ucs2_t c);
 261 smb_ucs2_t *strnrchr_w(const smb_ucs2_t *s, smb_ucs2_t c, unsigned int n);
 262 smb_ucs2_t *strstr_w(const smb_ucs2_t *s, const smb_ucs2_t *ins);
 263 bool strlower_w(smb_ucs2_t *s);
 264 bool strupper_w(smb_ucs2_t *s);
 265 int strcmp_w(const smb_ucs2_t *a, const smb_ucs2_t *b);
 266 int strcasecmp_w(const smb_ucs2_t *a, const smb_ucs2_t *b);
 267 int strncasecmp_w(const smb_ucs2_t *a, const smb_ucs2_t *b, size_t len);
 268 int strcmp_wa(const smb_ucs2_t *a, const char *b);
 269 int toupper_ascii(int c);
 270 int tolower_ascii(int c);
 271 int isupper_ascii(int c);
 272 int islower_ascii(int c);
 273
 274 /*
 275  *   Define stub for charset module which implements 8-bit encoding with gaps.
 276  *   Encoding tables for such module should be produced from glibc's CHARMAPs
 277  *   using script source/script/gen-8bit-gap.sh
 278  *   CHARSETNAME is CAPITALIZED charset name
 279  *
 280  *     */
 281 #define SMB_GENERATE_CHARSET_MODULE_8_BIT_GAP(CHARSETNAME)                                      \
 282 static size_t CHARSETNAME ## _push(void *cd, const char **inbuf, size_t *inbytesleft,                   \
 283                          char **outbuf, size_t *outbytesleft)                                   \
 284 {                                                                                               \
 285         while (*inbytesleft >= 2 && *outbytesleft >= 1) {                                       \
 286                 int i;                                                                          \
 287                 int done = 0;                                                                   \
 288                                                                                                 \
 289                 uint16 ch = SVAL(*inbuf,0);                                                     \
 290                                                                                                 \
 291                 for (i=0; from_idx[i].start != 0xffff; i++) {                                   \
 292                         if ((from_idx[i].start <= ch) && (from_idx[i].end >= ch)) {             \
 293                                 ((unsigned char*)(*outbuf))[0] = from_ucs2[from_idx[i].idx+ch]; \
 294                                 (*inbytesleft) -= 2;                                            \
 295                                 (*outbytesleft) -= 1;                                           \
 296                                 (*inbuf)  += 2;                                                 \
 297                                 (*outbuf) += 1;                                                 \
 298                                 done = 1;                                                       \
 299                                 break;                                                          \
 300                         }                                                                       \
 301                 }                                                                               \
 302                 if (!done) {                                                                    \
 303                         errno = EINVAL;                                                         \
 304                         return -1;                                                              \
 305                 }                                                                               \
 306                                                                                                 \
 307         }                                                                                       \
 308                                                                                                 \
 309         if (*inbytesleft == 1) {                                                                \
 310                 errno = EINVAL;                                                                 \
 311                 return -1;                                                                      \
 312         }                                                                                       \
 313                                                                                                 \
 314         if (*inbytesleft > 1) {                                                                 \
 315                 errno = E2BIG;                                                                  \
 316                 return -1;                                                                      \
 317         }                                                                                       \
 318                                                                                                 \
 319         return 0;                                                                               \
 320 }                                                                                               \
 321                                                                                                 \
 322 static size_t CHARSETNAME ## _pull(void *cd, const char **inbuf, size_t *inbytesleft,                           \
 323                          char **outbuf, size_t *outbytesleft)                                   \
 324 {                                                                                               \
 325         while (*inbytesleft >= 1 && *outbytesleft >= 2) {                                       \
 326                 SSVAL(*outbuf, 0, to_ucs2[((unsigned char*)(*inbuf))[0]]);                      \
 327                 (*inbytesleft)  -= 1;                                                           \
 328                 (*outbytesleft) -= 2;                                                           \
 329                 (*inbuf)  += 1;                                                                 \
 330                 (*outbuf) += 2;                                                                 \
 331         }                                                                                       \
 332                                                                                                 \
 333         if (*inbytesleft > 0) {                                                                 \
 334                 errno = E2BIG;                                                                  \
 335                 return -1;                                                                      \
 336         }                                                                                       \
 337                                                                                                 \
 338         return 0;                                                                               \
 339 }                                                                                               \
 340                                                                                                 \
 341 struct charset_functions CHARSETNAME ## _functions =                                            \
 342                 {#CHARSETNAME, CHARSETNAME ## _pull, CHARSETNAME ## _push};                     \
 343                                                                                                 \
 344 NTSTATUS charset_ ## CHARSETNAME ## _init(void);                                                        \
 345 NTSTATUS charset_ ## CHARSETNAME ## _init(void)                                                 \
 346 {                                                                                               \
 347         if (!smb_register_charset(& CHARSETNAME ## _functions)) {       \
 348                 return NT_STATUS_INTERNAL_ERROR;                        \
 349         }                                                               \
 350         return NT_STATUS_OK; \
 351 }                                               \
 352
 353
 354 #endif /* __CHARSET_H__ */