src/icu_utf8.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2013 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_len = 0;
  35     buf8->utf8_cap = capacity;
  36     if (capacity > 0)
  37     {
  38         buf8->utf8 = (uint8_t *) xmalloc(sizeof(uint8_t) * capacity);
  39         buf8->utf8[0] = (uint8_t) 0;
  40     }
  41     else
  42         buf8->utf8 = 0;
  43     return buf8;
  44 }
  45
  46 struct icu_buf_utf8 *icu_buf_utf8_clear(struct icu_buf_utf8 *buf8)
  47 {
  48     if (buf8)
  49     {
  50         if (buf8->utf8)
  51             buf8->utf8[0] = (uint8_t) 0;
  52         buf8->utf8_len = 0;
  53     }
  54     return buf8;
  55 }
  56
  57 struct icu_buf_utf8 *icu_buf_utf8_resize(struct icu_buf_utf8 *buf8,
  58                                          size_t capacity)
  59 {
  60     if (!buf8)
  61         return 0;
  62
  63     if (capacity > 0)
  64     {
  65         if (0 == buf8->utf8)
  66             buf8->utf8 = (uint8_t *) xmalloc(sizeof(uint8_t) * capacity);
  67         else
  68             buf8->utf8
  69                 = (uint8_t *) xrealloc(buf8->utf8, sizeof(uint8_t) * capacity);
  70
  71         buf8->utf8_cap = capacity;
  72     }
  73     else
  74     {
  75         xfree(buf8->utf8);
  76         buf8->utf8 = 0;
  77         buf8->utf8_cap = 0;
  78     }
  79
  80     return buf8;
  81 }
  82
  83 const char *icu_buf_utf8_to_cstr(struct icu_buf_utf8 *src8)
  84 {
  85     if (!src8 || src8->utf8_len == 0)
  86         return "";
  87
  88     if (src8->utf8_len == src8->utf8_cap)
  89         src8 = icu_buf_utf8_resize(src8, src8->utf8_len * 2 + 1);
  90
  91     src8->utf8[src8->utf8_len] = '\0';
  92
  93     return (const char *) src8->utf8;
  94 }
  95
  96 void icu_buf_utf8_destroy(struct icu_buf_utf8 *buf8)
  97 {
  98     if (buf8)
  99         xfree(buf8->utf8);
 100     xfree(buf8);
 101 }
 102
 103 UErrorCode icu_utf16_from_utf8_cstr(struct icu_buf_utf16 *dest16,
 104                                     const char *src8cstr,
 105                                     UErrorCode *status)
 106 {
 107     size_t src8cstr_len = 0;
 108     int32_t utf16_len = 0;
 109
 110     *status = U_ZERO_ERROR;
 111     src8cstr_len = strlen(src8cstr);
 112
 113     u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 114                   &utf16_len,
 115                   src8cstr, src8cstr_len, status);
 116
 117     /* check for buffer overflow, resize and retry */
 118     if (*status == U_BUFFER_OVERFLOW_ERROR)
 119     {
 120         icu_buf_utf16_resize(dest16, utf16_len * 2);
 121         *status = U_ZERO_ERROR;
 122         u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 123                       &utf16_len,
 124                       src8cstr, src8cstr_len, status);
 125     }
 126
 127     if (U_SUCCESS(*status) && utf16_len <= dest16->utf16_cap)
 128         dest16->utf16_len = utf16_len;
 129     else
 130         icu_buf_utf16_clear(dest16);
 131
 132     return *status;
 133 }
 134
 135 UErrorCode icu_utf16_to_utf8(struct icu_buf_utf8 *dest8,
 136                              const struct icu_buf_utf16 *src16,
 137                              UErrorCode *status)
 138 {
 139     int32_t utf8_len = 0;
 140
 141     u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 142                 &utf8_len,
 143                 src16->utf16, src16->utf16_len, status);
 144
 145     /* check for buffer overflow, resize and retry */
 146     if (*status == U_BUFFER_OVERFLOW_ERROR)
 147     {
 148         icu_buf_utf8_resize(dest8, utf8_len * 2);
 149         *status = U_ZERO_ERROR;
 150         u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 151                     &utf8_len,
 152                     src16->utf16, src16->utf16_len, status);
 153     }
 154
 155     if (U_SUCCESS(*status) && utf8_len <= dest8->utf8_cap)
 156         dest8->utf8_len = utf8_len;
 157     else
 158         icu_buf_utf8_clear(dest8);
 159
 160     return *status;
 161 }
 162
 163 #endif /* YAZ_HAVE_ICU */
 164
 165 /*
 166  * Local variables:
 167  * c-basic-offset: 4
 168  * c-file-style: "Stroustrup"
 169  * indent-tabs-mode: nil
 170  * End:
 171  * vim: shiftwidth=4 tabstop=8 expandtab
 172  */
 173