IO patches

simplify error handling
use get and inject
use wide support in OS
be stricter in checkin streams and arguments
This commit is contained in:
Vítor Santos Costa
2015-10-08 02:23:45 +01:00
parent b788dc131d
commit b3cc23ce64
30 changed files with 1224 additions and 1381 deletions

View File

@@ -19,9 +19,6 @@
#define ENCODING_H 1
#include "Yap.h"
typedef enum {
ENC_OCTET = 0, /// binary files
ENC_ISO_LATIN1 = 1, /// US+West Europe
@@ -33,12 +30,56 @@ typedef enum {
ENC_ISO_UTF32_BE = 64, /// nobody
ENC_ISO_UTF32_LE = 128, /// yes, nobody
} encoding_t;
/// read the initial encoding from the Operating System's environment;
encoding_t Yap_InitialEncoding( void );
#if defined(__BIG_ENDIAN__)
#define ENC_WCHAR ENC_ISO_UTF32_BE
#else
#define ENC_WCHAR ENC_ISO_UTF32_LE
#endif
#ifdef YAP_H
/// read the current environment, as set by the user or as Initial
encoding_t Yap_DefaultEncoding( void );
void Yap_SetDefaultEncoding(encoding_t new_encoding);
#if HAVE_XLOCALE_H
typedef enum {
SEQ_ENC_OCTET, /// binary files
SEQ_ENC_ISO_LATIN1, /// US+West Europe
SEQ_ENC_ISO_ASCII , /// US only
SEQ_ENC_ISO_ANSI , /// Who cares
SEQ_ENC_ISO_UTF8 , /// Most everyone nowadays
SEQ_ENC_UTF16_BE, /// People who made a mistake
SEQ_ENC_UTF16_LE, /// People who made the same mistake
v\ SEQ_ENC_ISO_UTF32_BE, /// nobody
SEQ_ENC_ISO_UTF32_LE /// yes, nobody
} seq_encoding_t;
/// convert from unary to binary representation.
static inline seq_encoding_t seq_encoding(encoding_t inp) {
#if HAVE__BUILTIN_FFSLL
return __builtin_ffsll(inp);
#elif HAVE_FFSLL
return ffsll(inp);
#else
unsigned int out;
// supports max 16 different encodings.
if (inp==0)
return 0L;
// if (inp & ((CELL)0xffffL << 16)) {inp >>= 16; out += 16;}
if (inp & ((CELL)0xffL << 8)) {inp >>= 8; out += 8;}
if (inp & ((CELL)0xfL << 4)) {inp >>= 4; out += 4;}
if (inp & ((CELL)0x3L << 2)) {inp >>= 2; out += 2;}
if (inp & ((CELL)0x1 << 1)) out++;
#endif
return out;
}
extern xlocale enc_locales[SEQ_ENC_ISO_UTF32_LE+1];
#endif
static inline const char *enc_name(encoding_t enc)
{
switch(enc)
@@ -59,23 +100,24 @@ static inline
encoding_t enc_id(char *s)
{
{
if (!strcmp(s, "octet")) return ENC_OCTET;
if (!strcmp(s, "iso_utf8")) return ENC_ISO_UTF8;
if (!strcmp(s, "utf16_be")) return ENC_UTF16_BE;
if (!strcmp(s, "utf16_le")) return ENC_UTF16_LE;
if (!strcmp(s, "octet")) return ENC_OCTET;
if (!strcmp(s, "iso_latin_1")) return ENC_ISO_LATIN1;
if (!strcmp(s, "iso_ascii")) return ENC_ISO_ASCII;
if (!strcmp(s, "iso_ansi")) return ENC_ISO_ANSI;
if (!strcmp(s, "iso_utf8")) return ENC_ISO_UTF8;
if (!strcmp(s, "utf16_be")) return ENC_UTF16_BE;
if (!strcmp(s, "utf16_le")) return ENC_UTF16_LE;
if (!strcmp(s, "utf32_be")) return ENC_ISO_UTF32_BE;
if (!strcmp(s, "utf32_le")) return ENC_ISO_UTF32_LE;
if (!strcmp(s, "default")) return Yap_DefaultEncoding();
else {
Yap_Error(DOMAIN_ERROR_OUT_OF_RANGE, 0, "bad encoding %s", s);
Yap_Error(DOMAIN_ERROR_OUT_OF_RANGE, 0, "bad encoding %s", s);
return ENC_OCTET;
}
}
}
#endif
#endif