src/icu_I18N.h

   1 /* $Id: icu_I18N.h,v 1.11 2007-05-11 10:38:42 marc Exp $
   2    Copyright (c) 2006-2007, Index Data.
   3
   4    This file is part of Pazpar2.
   5
   6    Pazpar2 is free software; you can redistribute it and/or modify it under
   7    the terms of the GNU General Public License as published by the Free
   8    Software Foundation; either version 2, or (at your option) any later
   9    version.
  10
  11    Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
  12    WARRANTY; without even the implied warranty of MERCHANTABILITY or
  13    FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  14    for more details.
  15
  16    You should have received a copy of the GNU General Public License
  17    along with Pazpar2; see the file LICENSE.  If not, write to the
  18    Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
  19    02111-1307, USA.
  20 */
  21
  22 #ifndef ICU_I18NL_H
  23 #define ICU_I18NL_H
  24
  25 #ifdef HAVE_ICU
  26
  27 #include <yaz/nmem.h>
  28
  29
  30 #include <unicode/utypes.h>   /* Basic ICU data types */
  31 #include <unicode/uchar.h>    /* char names           */
  32
  33 //#include <unicode/ustdio.h>
  34 #include <unicode/ucol.h>
  35 //#include <unicode/ucnv.h>     /* C   Converter API    */
  36 //#include <unicode/ustring.h>  /* some more string fcns*/
  37 //#include <unicode/uloc.h>
  38 #include <unicode/ubrk.h>
  39 //#include <unicode/unistr.h>
  40 #include <unicode/utrans.h>
  41
  42
  43
  44 // forward declarations
  45 //struct UBreakIterator;
  46
  47
  48
  49
  50 // declared structs and functions
  51
  52
  53 int icu_check_status (UErrorCode status);
  54
  55 struct icu_buf_utf16
  56 {
  57   UChar * utf16;
  58   int32_t utf16_len;
  59   int32_t utf16_cap;
  60 };
  61
  62 struct icu_buf_utf16 * icu_buf_utf16_create(size_t capacity);
  63 struct icu_buf_utf16 * icu_buf_utf16_resize(struct icu_buf_utf16 * buf16,
  64                                             size_t capacity);
  65 struct icu_buf_utf16 * icu_buf_utf16_copy(struct icu_buf_utf16 * dest16,
  66                                           struct icu_buf_utf16 * src16);
  67 void icu_buf_utf16_destroy(struct icu_buf_utf16 * buf16);
  68
  69
  70
  71 struct icu_buf_utf8
  72 {
  73   uint8_t * utf8;
  74   int32_t utf8_len;
  75   int32_t utf8_cap;
  76 };
  77
  78 struct icu_buf_utf8 * icu_buf_utf8_create(size_t capacity);
  79 struct icu_buf_utf8 * icu_buf_utf8_resize(struct icu_buf_utf8 * buf8,
  80                                           size_t capacity);
  81 void icu_buf_utf8_destroy(struct icu_buf_utf8 * buf8);
  82
  83
  84 UErrorCode icu_utf16_from_utf8(struct icu_buf_utf16 * dest16,
  85                                struct icu_buf_utf8 * src8,
  86                                UErrorCode * status);
  87
  88 UErrorCode icu_utf16_from_utf8_cstr(struct icu_buf_utf16 * dest16,
  89                                     const char * src8cstr,
  90                                     UErrorCode * status);
  91
  92
  93 UErrorCode icu_utf16_to_utf8(struct icu_buf_utf8 * dest8,
  94                              struct icu_buf_utf16 * src16,
  95                              UErrorCode * status);
  96
  97 int icu_utf16_casemap(struct icu_buf_utf16 * dest16,
  98                       struct icu_buf_utf16 * src16,
  99                       const char *locale, char action,
 100                       UErrorCode *status);
 101
 102 UErrorCode icu_sortkey8_from_utf16(UCollator *coll,
 103                                    struct icu_buf_utf8 * dest8,
 104                                    struct icu_buf_utf16 * src16,
 105                                    UErrorCode * status);
 106
 107 struct icu_tokenizer
 108 {
 109   char locale[16];
 110   char action;
 111   UBreakIterator* bi;
 112   struct icu_buf_utf16 * buf16;
 113   int32_t token_count;
 114   int32_t token_id;
 115   int32_t token_start;
 116   int32_t token_end;
 117   // keep always invariant
 118   // 0 <= token_start
 119   //   <= token_end
 120   //   <= buf16->utf16_len
 121   // and invariant
 122   // 0 <= token_id <= token_count
 123 };
 124
 125 struct icu_tokenizer * icu_tokenizer_create(const char *locale, char action,
 126                                             UErrorCode *status);
 127
 128 void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer);
 129
 130 int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
 131                          struct icu_buf_utf16 * src16, UErrorCode *status);
 132
 133 int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
 134                                  struct icu_buf_utf16 * tkn16,
 135                                  UErrorCode *status);
 136
 137 int32_t icu_tokenizer_token_id(struct icu_tokenizer * tokenizer);
 138 int32_t icu_tokenizer_token_start(struct icu_tokenizer * tokenizer);
 139 int32_t icu_tokenizer_token_end(struct icu_tokenizer * tokenizer);
 140 int32_t icu_tokenizer_token_length(struct icu_tokenizer * tokenizer);
 141 int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer);
 142
 143
 144
 145 struct icu_normalizer
 146 {
 147   char action;
 148   struct icu_buf_utf16 * rules16;
 149   UParseError parse_error[256];
 150   UTransliterator * trans;
 151 };
 152
 153 struct icu_normalizer * icu_normalizer_create(const char *rules, char action,
 154                                               UErrorCode *status);
 155
 156
 157 void icu_normalizer_destroy(struct icu_normalizer * normalizer);
 158
 159 int icu_normalizer_normalize(struct icu_normalizer * normalizer,
 160                              struct icu_buf_utf16 * dest16,
 161                              struct icu_buf_utf16 * src16,
 162                              UErrorCode *status);
 163
 164
 165
 166
 167 #endif // HAVE_ICU
 168 #endif // ICU_I18NL_H