%% options copyright owner = Dirk Krause copyright year = 2015-xxxx license = bsd %% header /** @file String recoding from UTF-8 to ASCII. 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 UTF-8 to ASCII @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_utf8_to_ascii( char *dstb, size_t szdstb, const char *src, dk4_er_t *erp ); /** Calculate required buffer size (including final 0x00 byte) to convert from UTF-8 to ASCII. @param src UTF-8 encoded source string. @param erp Error report, may be NULL. @return Positive required buffer size on success, 0 on error. Error codes: - DK4_E_INVALID_ARGUMENTS
if src is NULL, - DK4_E_MATH_OVERFLOW
if the required size calculation results in mathematical overflow, - DK4_E_SYNTAX
if there are errors while decoding the source string. */ size_t dk4recode_size_utf8_to_ascii(const char *src, dk4_er_t *erp); #ifdef __cplusplus } #endif %% module #include "dk4rec04.h" #include "dk4utf8.h" size_t dk4recode_size_utf8_to_ascii(const char *src, dk4_er_t *erp) { dk4_er_t er; dk4_utf8_decoder_t dec; size_t rdb = 0; size_t back = 0; dk4_c32_t c32; if (NULL != src) { /* Initialize data structures. */ dk4error_init(&er); dk4utf8_init(&dec); /* Process source string. */ while ('\0' != *src) { rdb++; switch(dk4utf8_add(&dec, (unsigned char)(*(src++)))) { case DK4_EDSTM_FINISHED: { c32 = dk4utf8_get(&dec); dk4utf8_init(&dec); if (dkC32(0x00000100) > c32) { if (SIZE_MAX == back) { dk4error_set_simple_error_code(&er, DK4_E_MATH_OVERFLOW); } else { back++; } } else { dk4error_set_elsize_nelem(&er, DK4_E_SYNTAX, 1, rdb); } } break; case DK4_EDSTM_ERROR: { dk4error_set_elsize_nelem(&er, DK4_E_SYNTAX, 1, rdb); } break; } } if (0 == dk4utf8_is_empty(&dec)) { dk4error_set_elsize_nelem(&er, DK4_E_SYNTAX, 1, rdb); } /* Add one for the finalizer. */ if (SIZE_MAX == back) { dk4error_set_simple_error_code(&er, DK4_E_MATH_OVERFLOW); } else { back++; } /* On error return 0 and pass error code. */ if (DK4_E_NONE != er.ec) { back = 0; dk4error_copy(erp, &er); } } else { dk4error_set_simple_error_code(erp, DK4_E_INVALID_ARGUMENTS); } return back; } int dk4recode_utf8_to_ascii( char *dstb, size_t szdstb, const char *src, dk4_er_t *erp ) { dk4_utf8_decoder_t dec; size_t used = 0; size_t rdb = 0; dk4_c32_t c32 = dkC32(0); int back = 0; if ((NULL != dstb) && (NULL != src) && (0 < szdstb)) { back = 1; dk4utf8_init(&dec); while(('\0' != *src) && (1 == back) && (used < szdstb)) { switch(dk4utf8_add(&dec, (unsigned char)(*(src++)))) { case DK4_EDSTM_FINISHED: { c32 = dk4utf8_get(&dec); dk4utf8_init(&dec); if (dkC32(0x00000100) > c32) { dstb[used++] = (char)c32; } else { back = 0; dk4error_set_elsize_nelem(erp, DK4_E_SYNTAX, 1, rdb); } } break; case DK4_EDSTM_ERROR: { back = 0; dk4error_set_elsize_nelem(erp, DK4_E_SYNTAX, 1, rdb); } break; } if(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); } if (0 == dk4utf8_is_empty(&dec)) { back = 0; dk4error_set_elsize_nelem(erp, DK4_E_SYNTAX, 1, rdb); } } else { dk4error_set_simple_error_code(erp, DK4_E_INVALID_ARGUMENTS); } return back; }