%% options
copyright owner = Dirk Krause
copyright year = 2015-xxxx
license = bsd
%% header
/** @file
String recoding from ASCII to UTF-8.
CRT on Windows: Not used.
*/
#ifndef DK4CONF_H_INCLUDED
#include "dk4conf.h"
#endif
#ifndef DK4TYPES_H_INCLUDED
#include "dk4types.h"
#endif
#ifndef DK4ERROR_H_INCLUDED
#include "dk4error.h"
#endif
#ifdef __cplusplus
extern "C" {
#endif
/** Recode string from ASCII to UTF-8.
@param dstb Destination buffer.
@param szdstb Size of destination buffer (number of bytes).
@param src Source string to convert.
@param erp Error report, may be NULL.
@return 1 on success, 0 on error.
Error codes:
- DK4_E_INVALID_ARGUMENTS
if src or dstb is NULL or szdstb is 0,
- DK4_E_BUFFER_TOO_SMALL
if dstb is too small,
- DK4_E_SYNTAX
with the number of successfully recoded characters in nelem if a
non-recodable character was found.
*/
int
dk4recode_ascii_to_utf8(
char *dstb, size_t szdstb, const char *src, dk4_er_t *erp
);
/** Calculate required buffer size (including final 0x00 byte)
for ASCII to UTF-8 conversion.
@param src ASCII encoded string.
@param erp Error report, may be NULL.
@return Positive value on success, 0 on error.
Error codes:
- DK4_E_INVALID_ARGUMENTS
if src is NULL,
- DK4_E_SYNTAX
if there are encoding errors,
- DK4_E_MATH_OVERFLOW
if a numeric overflow occured in size calculation.
*/
size_t
dk4recode_size_ascii_to_utf8(const char *src, dk4_er_t *erp);
#ifdef __cplusplus
}
#endif
%% module
#include "dk4rec03.h"
#include "dk4mem.h"
#include "dk4utf8.h"
size_t
dk4recode_size_ascii_to_utf8(const char *src, dk4_er_t *erp)
{
dk4_er_t er; /* Internal error report */
unsigned char buf[8]; /* current character UTF-8 encoded */
size_t sz; /* Size of current UTF-8 encoded character */
size_t rdb = 0; /* Number of bytes read from input */
size_t back = 0; /* Function result */
unsigned char uc; /* Currenct character */
if (NULL != src) {
dk4error_init(&er);
while ('\0' != *src) {
uc = (unsigned char)(*(src++));
rdb++;
sz = sizeof(buf);
if (0 != dk4utf8_encode(buf, &sz, (dk4_c32_t)uc, &er)) {
if ((SIZE_MAX - sz) < back) {
dk4error_set_simple_error_code(erp, DK4_E_MATH_OVERFLOW);
}
back += sz;
} else {
dk4error_set_elsize_nelem(&er, DK4_E_SYNTAX, 1, rdb);
}
}
if (SIZE_MAX == back) {
dk4error_set_simple_error_code(erp, DK4_E_MATH_OVERFLOW);
}
back++;
if (DK4_E_NONE != er.ec) {
dk4error_copy(erp, &er);
back = 0;
}
} else {
dk4error_set_simple_error_code(erp, DK4_E_INVALID_ARGUMENTS);
}
return back;
}
int
dk4recode_ascii_to_utf8(
char *dstb, size_t szdstb, const char *src, dk4_er_t *erp
)
{
unsigned char buf[8];
size_t sz;
size_t rdb = 0;
size_t used = 0;
int back = 0;
unsigned char uc = '\0';
if ((NULL != dstb) && (NULL != src) && (0 < szdstb)) {
back = 1;
while(('\0' != *src) && (1 == back) && (used < szdstb)) {
uc = (unsigned char)(*(src++));
sz = sizeof(buf);
if (0 != dk4utf8_encode(buf, &sz, (dk4_c32_t)uc, erp)) {
if ((szdstb - used) > sz) {
DK4_MEMCPY(&(dstb[used]),buf,sz);
used += sz;
} else {
back = 0;
dk4error_set_simple_error_code(erp, DK4_E_BUFFER_TOO_SMALL);
}
} else {
back = 0;
dk4error_set_elsize_nelem(erp, DK4_E_SYNTAX, 1, rdb);
}
if (0 != back) { rdb++; }
}
if (used < szdstb) {
dstb[used] = '\0';
} else {
dstb[szdstb - 1] = '\0';
back = 0;
dk4error_set_simple_error_code(erp, DK4_E_BUFFER_TOO_SMALL);
}
} else {
dk4error_set_simple_error_code(erp, DK4_E_INVALID_ARGUMENTS);
}
return back;
}