src/icu_utf8.c

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