include/yaz/icu_I18N.h

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2010 Index Data.
   3  * All rights reserved.
   4  * Redistribution and use in source and binary forms, with or without
   5  * modification, are permitted provided that the following conditions are met:
   6  *
   7  *     * Redistributions of source code must retain the above copyright
   8  *       notice, this list of conditions and the following disclaimer.
   9  *     * Redistributions in binary form must reproduce the above copyright
  10  *       notice, this list of conditions and the following disclaimer in the
  11  *       documentation and/or other materials provided with the distribution.
  12  *     * Neither the name of Index Data nor the names of its contributors
  13  *       may be used to endorse or promote products derived from this
  14  *       software without specific prior written permission.
  15  *
  16  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND ANY
  17  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  18  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
  19  * DISCLAIMED. IN NO EVENT SHALL THE REGENTS AND CONTRIBUTORS BE LIABLE FOR ANY
  20  * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
  21  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  22  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
  23  * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  24  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
  25  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  26  */
  27
  28 /** \file
  29     \brief Internal header for ICU utilities
  30
  31     These functions, while non-static, are considered unstable and internal
  32     and may be renamed for each YAZ release.
  33 */
  34
  35 #ifndef ICU_I18NL_H
  36 #define ICU_I18NL_H
  37
  38 #include <yaz/yconfig.h>
  39
  40 #include <unicode/utypes.h>   /* Basic ICU data types */
  41 #include <unicode/uchar.h>    /* char names           */
  42
  43 #include <unicode/ucol.h>
  44 #include <unicode/ubrk.h>
  45
  46 #include <yaz/icu.h>
  47
  48 /* declared structs and functions */
  49
  50 int icu_check_status (UErrorCode status);
  51
  52 struct icu_buf_utf16
  53 {
  54     UChar * utf16;
  55     int32_t utf16_len;
  56     int32_t utf16_cap;
  57 };
  58
  59 struct icu_buf_utf16 * icu_buf_utf16_create(size_t capacity);
  60
  61 struct icu_buf_utf16 * icu_buf_utf16_clear(struct icu_buf_utf16 * buf16);
  62
  63 struct icu_buf_utf16 * icu_buf_utf16_resize(struct icu_buf_utf16 * buf16,
  64                                             size_t capacity);
  65
  66 struct icu_buf_utf16 *icu_buf_utf16_copy(struct icu_buf_utf16 * dest16,
  67                                          const struct icu_buf_utf16 * src16);
  68
  69 void icu_buf_utf16_destroy(struct icu_buf_utf16 * buf16);
  70
  71 struct icu_buf_utf8;
  72
  73 struct icu_buf_utf8
  74 {
  75     uint8_t * utf8;
  76     int32_t utf8_len;
  77     int32_t utf8_cap;
  78 };
  79
  80 struct icu_buf_utf8 * icu_buf_utf8_create(size_t capacity);
  81
  82 struct icu_buf_utf8 * icu_buf_utf8_clear(struct icu_buf_utf8 * buf8);
  83
  84 struct icu_buf_utf8 * icu_buf_utf8_resize(struct icu_buf_utf8 * buf8,
  85                                           size_t capacity);
  86
  87 void icu_buf_utf8_destroy(struct icu_buf_utf8 * buf8);
  88
  89
  90 UErrorCode icu_utf16_from_utf8_cstr(struct icu_buf_utf16 * dest16,
  91                                     const char * src8cstr,
  92                                     UErrorCode * status);
  93
  94 const char *icu_buf_utf8_to_cstr(struct icu_buf_utf8 *src8);
  95
  96
  97 UErrorCode icu_utf16_to_utf8(struct icu_buf_utf8 *dest8,
  98                              const struct icu_buf_utf16 *src16,
  99                              UErrorCode * status);
 100
 101 struct icu_casemap;
 102
 103 struct icu_casemap * icu_casemap_create(char action, UErrorCode *status);
 104
 105 void icu_casemap_destroy(struct icu_casemap * casemap);
 106
 107 int icu_casemap_casemap(struct icu_casemap * casemap,
 108                         struct icu_buf_utf16 * dest16,
 109                         struct icu_buf_utf16 * src16,
 110                         UErrorCode *status,
 111                         const char *locale);
 112
 113 int icu_utf16_casemap(struct icu_buf_utf16 * dest16,
 114                       struct icu_buf_utf16 * src16,
 115                       const char *locale, char action,
 116                       UErrorCode *status);
 117
 118 void icu_sortkey8_from_utf16(UCollator *coll,
 119                              struct icu_buf_utf8 * dest8,
 120                              struct icu_buf_utf16 * src16,
 121                              UErrorCode * status);
 122
 123 struct icu_tokenizer
 124 {
 125     char action;
 126     UBreakIterator* bi;
 127     struct icu_buf_utf16 * buf16;
 128     int32_t token_count;
 129     int32_t token_id;
 130     int32_t token_start;
 131     int32_t token_end;
 132 /*
 133   keep always invariant
 134   0 <= token_start
 135   <= token_end
 136   <= buf16->utf16_len
 137   and invariant
 138   0 <= token_id <= token_count
 139 */
 140 };
 141
 142 struct icu_tokenizer * icu_tokenizer_create(const char *locale, char action,
 143                                             UErrorCode *status);
 144
 145 void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer);
 146
 147 int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
 148                          struct icu_buf_utf16 * src16, UErrorCode *status);
 149
 150 int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
 151                                  struct icu_buf_utf16 * tkn16,
 152                                  UErrorCode *status);
 153
 154 int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer);
 155
 156 struct icu_transform;
 157
 158 struct icu_transform * icu_transform_create(const char *id, char action,
 159                                             const char *rules,
 160                                             UErrorCode *status);
 161
 162 void icu_transform_destroy(struct icu_transform * transform);
 163
 164 int icu_transform_trans(struct icu_transform * transform,
 165                         struct icu_buf_utf16 * dest16,
 166                         const struct icu_buf_utf16 * src16,
 167                         UErrorCode *status);
 168
 169 struct icu_chain_step;
 170
 171 int icu_chain_token_number(yaz_icu_chain_t chain);
 172
 173 yaz_icu_chain_t icu_chain_create(const char * locale,
 174                                  int sort,
 175                                  UErrorCode * status);
 176
 177
 178 struct icu_iter;
 179 struct icu_iter *icu_iter_create(struct icu_chain *chain,
 180                                  const char *src8cstr);
 181 void icu_iter_destroy(struct icu_iter *iter);
 182 int icu_iter_next(struct icu_iter *iter, struct icu_buf_utf8 *result);
 183 const char *icu_iter_get_sortkey(struct icu_iter *iter);
 184 const char *icu_iter_get_display(struct icu_iter *iter);
 185
 186 #endif /* ICU_I18NL_H */
 187
 188 /*
 189  * Local variables:
 190  * c-basic-offset: 4
 191  * c-file-style: "Stroustrup"
 192  * indent-tabs-mode: nil
 193  * End:
 194  * vim: shiftwidth=4 tabstop=8 expandtab
 195  */
 196