src/icu_utf8.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2012 Index Data
   3  * See the file LICENSE for details.
   4  */
   5
   6 /**
   7  * \file
   8  * \brief UTF-8 string utilities for ICU
   9  */
  10
  11 #if HAVE_CONFIG_H
  12 #include "config.h"
  13 #endif
  14
  15 #if YAZ_HAVE_ICU
  16 #include <yaz/xmalloc.h>
  17
  18 #include <yaz/icu_I18N.h>
  19
  20 #include <yaz/log.h>
  21
  22 #include <string.h>
  23 #include <stdlib.h>
  24 #include <stdio.h>
  25
  26 #include <unicode/ustring.h>  /* some more string fcns*/
  27 #include <unicode/uchar.h>    /* char names           */
  28
  29 struct icu_buf_utf8 *icu_buf_utf8_create(size_t capacity)
  30 {
  31     struct icu_buf_utf8 * buf8
  32         = (struct icu_buf_utf8 *) xmalloc(sizeof(struct icu_buf_utf8));
  33
  34     buf8->utf8 = 0;
  35     buf8->utf8_len = 0;
  36     buf8->utf8_cap = 0;
  37
  38     if (capacity > 0)
  39     {
  40         buf8->utf8 = (uint8_t *) xmalloc(sizeof(uint8_t) * capacity);
  41         buf8->utf8[0] = (uint8_t) 0;
  42         buf8->utf8_cap = capacity;
  43     }
  44     return buf8;
  45 }
  46
  47 struct icu_buf_utf8 * icu_buf_utf8_clear(struct icu_buf_utf8 * buf8)
  48 {
  49     if (buf8)
  50     {
  51         if (buf8->utf8)
  52             buf8->utf8[0] = (uint8_t) 0;
  53         buf8->utf8_len = 0;
  54     }
  55     return buf8;
  56 }
  57
  58 struct icu_buf_utf8 * icu_buf_utf8_resize(struct icu_buf_utf8 * buf8,
  59                                           size_t capacity)
  60 {
  61     if (!buf8)
  62         return 0;
  63
  64     if (capacity > 0)
  65     {
  66         if (0 == buf8->utf8)
  67             buf8->utf8 = (uint8_t *) xmalloc(sizeof(uint8_t) * capacity);
  68         else
  69             buf8->utf8
  70                 = (uint8_t *) xrealloc(buf8->utf8, sizeof(uint8_t) * capacity);
  71
  72         buf8->utf8_cap = capacity;
  73     }
  74     else
  75     {
  76         xfree(buf8->utf8);
  77         buf8->utf8 = 0;
  78         buf8->utf8_len = 0;
  79         buf8->utf8_cap = 0;
  80     }
  81
  82     return buf8;
  83 }
  84
  85 const char *icu_buf_utf8_to_cstr(struct icu_buf_utf8 *src8)
  86 {
  87     if (!src8 || src8->utf8_len == 0)
  88         return "";
  89
  90     if (src8->utf8_len == src8->utf8_cap)
  91         src8 = icu_buf_utf8_resize(src8, src8->utf8_len * 2 + 1);
  92
  93     src8->utf8[src8->utf8_len] = '\0';
  94
  95     return (const char *) src8->utf8;
  96 }
  97
  98 void icu_buf_utf8_destroy(struct icu_buf_utf8 * buf8)
  99 {
 100     if (buf8)
 101         xfree(buf8->utf8);
 102     xfree(buf8);
 103 }
 104
 105 UErrorCode icu_utf16_from_utf8_cstr(struct icu_buf_utf16 * dest16,
 106                                     const char * src8cstr,
 107                                     UErrorCode * status)
 108 {
 109     size_t src8cstr_len = 0;
 110     int32_t utf16_len = 0;
 111
 112     *status = U_ZERO_ERROR;
 113     src8cstr_len = strlen(src8cstr);
 114
 115     u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 116                   &utf16_len,
 117                   src8cstr, src8cstr_len, status);
 118
 119     /* check for buffer overflow, resize and retry */
 120     if (*status == U_BUFFER_OVERFLOW_ERROR)
 121     {
 122         icu_buf_utf16_resize(dest16, utf16_len * 2);
 123         *status = U_ZERO_ERROR;
 124         u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 125                       &utf16_len,
 126                       src8cstr, src8cstr_len, status);
 127     }
 128
 129     if (U_SUCCESS(*status) && utf16_len <= dest16->utf16_cap)
 130         dest16->utf16_len = utf16_len;
 131     else
 132         icu_buf_utf16_clear(dest16);
 133
 134     return *status;
 135 }
 136
 137 UErrorCode icu_utf16_to_utf8(struct icu_buf_utf8 *dest8,
 138                              const struct icu_buf_utf16 *src16,
 139                              UErrorCode * status)
 140 {
 141     int32_t utf8_len = 0;
 142
 143     u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 144                 &utf8_len,
 145                 src16->utf16, src16->utf16_len, status);
 146
 147     /* check for buffer overflow, resize and retry */
 148     if (*status == U_BUFFER_OVERFLOW_ERROR)
 149     {
 150         icu_buf_utf8_resize(dest8, utf8_len * 2);
 151         *status = U_ZERO_ERROR;
 152         u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 153                     &utf8_len,
 154                     src16->utf16, src16->utf16_len, status);
 155     }
 156
 157     if (U_SUCCESS(*status) && utf8_len <= dest8->utf8_cap)
 158         dest8->utf8_len = utf8_len;
 159     else
 160         icu_buf_utf8_clear(dest8);
 161
 162     return *status;
 163 }
 164
 165 #endif /* YAZ_HAVE_ICU */
 166
 167 /*
 168  * Local variables:
 169  * c-basic-offset: 4
 170  * c-file-style: "Stroustrup"
 171  * indent-tabs-mode: nil
 172  * End:
 173  * vim: shiftwidth=4 tabstop=8 expandtab
 174  */
 175