JehanneOS/newlib
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Consolidate wctomb/mbtowc calls for POSIX-1.2008

Browse Source 
- Remove charset parameter from low level __foo_wctomb/__foo_mbtowc calls.
- Instead, create array of function for ISO and Windows codepages to point
  to function which does not require to evaluate the charset string on
  each call.  Create matching helper functions.  I.e., __iso_wctomb,
  __iso_mbtowc, __cp_wctomb and __cp_mbtowc are functions returning the
  right function pointer now.
- Create __WCTOMB/__MBTOWC macros utilizing per-reent locale and replace
  calls to __wctomb/__mbtowc with calls to __WCTOMB/__MBTOWC.
- Drop global __wctomb/__mbtowc vars.
- Utilize aforementioned changes in Cygwin to get rid of charset in other,
  calling functions and simplify the code.
- In Cygwin restrict global cygheap locale info to the job performed
  by internal_setlocale.  Use UTF-8 instead of ASCII on the fly in
  internal conversion functions.
- In Cygwin dll_entry, make sure to initialize a TLS area with a NULL
  _REENT->_locale pointer.  Add comment to explain why.

Signed-off by: Corinna Vinschen <corinna@vinschen.de>
This commit is contained in:
Corinna Vinschen
2016-07-20 22:05:59 +02:00
parent 88208d3735
commit d16a56306d
31 changed files with 941 additions and 355 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
winsup/cygwin/cygheap.cc
View File

@@ -28,7 +28,7 @@
static mini_cygheap NO_COPY cygheap_dummy =
{
{__utf8_mbtowc, __utf8_wctomb}
{__utf8_mbtowc}
};
init_cygheap NO_COPY *cygheap = (init_cygheap *) &cygheap_dummy;
@@ -245,8 +245,6 @@ cygheap_init ()
cygheap->bucket_val[b] = sz[b & 1];
/* Default locale settings. */
cygheap->locale.mbtowc = __utf8_mbtowc;
cygheap->locale.wctomb = __utf8_wctomb;
strcpy (cygheap->locale.charset, "UTF-8");
/* Set umask to a sane default. */
cygheap->umask = 022;
cygheap->rlim_core = RLIM_INFINITY;

2
winsup/cygwin/cygheap.h
View File

@@ -346,8 +346,6 @@ struct cygheap_debug
struct cygheap_locale
{
mbtowc_p mbtowc;
wctomb_p wctomb;
char charset[ENCODING_LEN + 1];
};
struct user_heap_info

2
winsup/cygwin/fhandler.h
View File

@@ -1355,7 +1355,7 @@ class dev_console
inline UINT get_console_cp ();
DWORD con_to_str (char *d, int dlen, WCHAR w);
DWORD str_to_con (mbtowc_p, const char *, PWCHAR d, const char *s, DWORD sz);
DWORD str_to_con (mbtowc_p, PWCHAR d, const char *s, DWORD sz);
void set_color (HANDLE);
void set_default_attr ();
int set_cl_x (cltype);

35
winsup/cygwin/fhandler_console.cc
View File

@@ -225,10 +225,9 @@ dev_console::get_console_cp ()
}
inline DWORD
dev_console::str_to_con (mbtowc_p f_mbtowc, const char *charset,
PWCHAR d, const char *s, DWORD sz)
dev_console::str_to_con (mbtowc_p f_mbtowc, PWCHAR d, const char *s, DWORD sz)
{
return sys_cp_mbstowcs (f_mbtowc, charset, d, CONVERT_LIMIT, s, sz);
return sys_cp_mbstowcs (f_mbtowc, d, CONVERT_LIMIT, s, sz);
}
bool
@@ -2002,21 +2001,10 @@ fhandler_console::write_normal (const unsigned char *src,
const unsigned char *found = src;
size_t ret;
mbstate_t ps;
UINT cp = con.get_console_cp ();
const char *charset;
mbtowc_p f_mbtowc;
if (cp)
{
/* The alternate charset is always 437, just as in the Linux console. */
f_mbtowc = __cp_mbtowc;
charset = "CP437";
}
else
{
f_mbtowc = cygheap->locale.mbtowc;
charset = cygheap->locale.charset;
}
/* The alternate charset is always 437, just as in the Linux console. */
f_mbtowc = con.get_console_cp () ? __cp_mbtowc (437) : __MBTOWC;
/* First check if we have cached lead bytes of a former try to write
a truncated multibyte sequence. If so, process it. */
@@ -2027,7 +2015,7 @@ fhandler_console::write_normal (const unsigned char *src,
memcpy (trunc_buf.buf + trunc_buf.len, src, cp_len);
memset (&ps, 0, sizeof ps);
switch (ret = f_mbtowc (_REENT, NULL, (const char *) trunc_buf.buf,
trunc_buf.len + cp_len, charset, &ps))
trunc_buf.len + cp_len, &ps))
{
case -2:
/* Still truncated multibyte sequence? Keep in trunc_buf. */
@@ -2052,9 +2040,9 @@ fhandler_console::write_normal (const unsigned char *src,
/* Valid multibyte sequence? Process. */
if (nfound)
{
buf_len = con.str_to_con (f_mbtowc, charset, write_buf,
(const char *) trunc_buf.buf,
nfound - trunc_buf.buf);
buf_len = con.str_to_con (f_mbtowc, write_buf,
(const char *) trunc_buf.buf,
nfound - trunc_buf.buf);
if (!write_console (write_buf, buf_len, done))
{
debug_printf ("multibyte sequence write failed, handle %p", get_output_handle ());
@@ -2075,7 +2063,7 @@ fhandler_console::write_normal (const unsigned char *src,
&& base_chars[*found] == NOR)
{
switch (ret = f_mbtowc (_REENT, NULL, (const char *) found,
end - found, charset, &ps))
end - found, &ps))
{
case -2: /* Truncated multibyte sequence. Store for next write. */
trunc_buf.len = end - found;
@@ -2098,8 +2086,7 @@ do_print:
if (found != src)
{
DWORD len = found - src;
buf_len = con.str_to_con (f_mbtowc, charset, write_buf,
(const char *) src, len);
buf_len = con.str_to_con (f_mbtowc, write_buf, (const char *) src, len);
if (!buf_len)
{
debug_printf ("conversion error, handle %p",
@@ -2178,7 +2165,7 @@ do_print:
if (found + 1 < end)
{
ret = __utf8_mbtowc (_REENT, NULL, (const char *) found + 1,
end - found - 1, NULL, &ps);
end - found - 1, &ps);
if (ret != (size_t) -1)
while (ret-- > 0)
{

9
winsup/cygwin/init.cc
View File

@@ -83,6 +83,15 @@ dll_entry (HANDLE h, DWORD reason, void *static_load)
cygwin_hmodule = (HMODULE) h;
dynamically_loaded = (static_load == NULL);
/* Starting with adding the POSIX-1.2008 per-thread locale functionality,
we need an initalized _REENT area even for the functions called from
dll_crt0_0. In fact, we only need the _REENT->_locale pointer
initialized to NULL, so subsequent calls to locale-specific functions
will always fall back to __global_locale, rather then crash due to
_REENT->_locale having an arbitrary value. */
(void) alloca (CYGTLS_PADSIZE);
_REENT->_locale = NULL;
dll_crt0_0 ();
_my_oldfunc = TlsAlloc ();
dll_finished_loading = true;

95
winsup/cygwin/nlsfuncs.cc
View File

@@ -16,8 +16,6 @@ details. */
#include "dtable.h"
#include "cygheap.h"
#include "tls_pbuf.h"
/* Internal headers from newlib */
#include "../locale/setlocale.h"
#include "lc_msg.h"
#include "lc_era.h"
@@ -31,8 +29,7 @@ details. */
__eval_datetimefmt(lcid,(type),(flags),&lc_time_ptr,\
lc_time_end-lc_time_ptr)
#define charfromwchar(category,in) \
__charfromwchar (_##category##_locale->in,_LC(category),\
f_wctomb,charset)
__charfromwchar (_##category##_locale->in,_LC(category),f_wctomb)
#define has_modifier(x) ((x)[0] && !strcmp (modifier, (x)))
@@ -159,8 +156,7 @@ __get_lcid_from_locale (const char *name)
is set, s==NULL returns -1 since then it's used to recognize invalid strings
in the used charset. */
static size_t
lc_wcstombs (wctomb_p f_wctomb, const char *charset,
char *s, const wchar_t *pwcs, size_t n,
lc_wcstombs (wctomb_p f_wctomb, char *s, const wchar_t *pwcs, size_t n,
bool return_invalid = false)
{
char *ptr = s;
@@ -175,7 +171,7 @@ lc_wcstombs (wctomb_p f_wctomb, const char *charset,
size_t num_bytes = 0;
while (*pwcs != 0)
{
bytes = f_wctomb (_REENT, buf, *pwcs++, charset, &state);
bytes = f_wctomb (_REENT, buf, *pwcs++, &state);
if (bytes != (size_t) -1)
num_bytes += bytes;
else if (return_invalid)
@@ -185,7 +181,7 @@ lc_wcstombs (wctomb_p f_wctomb, const char *charset,
}
while (n > 0)
{
bytes = f_wctomb (_REENT, buf, *pwcs, charset, &state);
bytes = f_wctomb (_REENT, buf, *pwcs, &state);
if (bytes == (size_t) -1)
{
memset (&state, 0, sizeof state);
@@ -207,8 +203,7 @@ lc_wcstombs (wctomb_p f_wctomb, const char *charset,
/* Never returns -1. Invalid sequences are translated to replacement
wide-chars. */
static size_t
lc_mbstowcs (mbtowc_p f_mbtowc, const char *charset,
wchar_t *pwcs, const char *s, size_t n)
lc_mbstowcs (mbtowc_p f_mbtowc, wchar_t *pwcs, const char *s, size_t n)
{
size_t ret = 0;
char *t = (char *) s;
@@ -220,8 +215,7 @@ lc_mbstowcs (mbtowc_p f_mbtowc, const char *charset,
n = 1;
while (n > 0)
{
bytes = f_mbtowc (_REENT, pwcs, t, 6 /* fake, always enough */,
charset, &state);
bytes = f_mbtowc (_REENT, pwcs, t, 6 /* fake, always enough */, &state);
if (bytes == (size_t) -1)
{
state.__count = 0;
@@ -294,13 +288,12 @@ __setlocaleinfo (char **ptr, size_t size, wchar_t val)
}
static char *
__charfromwchar (const wchar_t *in, char **ptr, size_t size,
wctomb_p f_wctomb, const char *charset)
__charfromwchar (const wchar_t *in, char **ptr, size_t size, wctomb_p f_wctomb)
{
size_t num;
char *ret;
num = lc_wcstombs (f_wctomb, charset, ret = *ptr, in, size);
num = lc_wcstombs (f_wctomb, ret = *ptr, in, size);
*ptr += num + 1;
return ret;
}
@@ -600,11 +593,11 @@ __set_lc_time_from_win (const char *name,
/* Evaluate string length in target charset. Characters invalid in the
target charset are simply ignored, as on Linux. */
len = 0;
len += lc_wcstombs (f_wctomb, charset, NULL, era->era, 0) + 1;
len += lc_wcstombs (f_wctomb, charset, NULL, era->era_d_fmt, 0) + 1;
len += lc_wcstombs (f_wctomb, charset, NULL, era->era_d_t_fmt, 0) + 1;
len += lc_wcstombs (f_wctomb, charset, NULL, era->era_t_fmt, 0) + 1;
len += lc_wcstombs (f_wctomb, charset, NULL, era->alt_digits, 0) + 1;
len += lc_wcstombs (f_wctomb, NULL, era->era, 0) + 1;
len += lc_wcstombs (f_wctomb, NULL, era->era_d_fmt, 0) + 1;
len += lc_wcstombs (f_wctomb, NULL, era->era_d_t_fmt, 0) + 1;
len += lc_wcstombs (f_wctomb, NULL, era->era_t_fmt, 0) + 1;
len += lc_wcstombs (f_wctomb, NULL, era->alt_digits, 0) + 1;
len += (wcslen (era->era) + 1) * sizeof (wchar_t);
len += (wcslen (era->era_d_fmt) + 1) * sizeof (wchar_t);
len += (wcslen (era->era_d_t_fmt) + 1) * sizeof (wchar_t);
@@ -742,8 +735,7 @@ __set_lc_ctype_from_win (const char *name,
lc_ctype_ptr = (char *) woutdig;
_ctype_locale->outdigits[i] = lc_ctype_ptr;
memset (&state, 0, sizeof state);
lc_ctype_ptr += f_wctomb (_REENT, lc_ctype_ptr, digits[i], charset,
&state);
lc_ctype_ptr += f_wctomb (_REENT, lc_ctype_ptr, digits[i], &state);
*lc_ctype_ptr++ = '\0';
}
}
@@ -885,8 +877,7 @@ __set_lc_monetary_from_win (const char *name,
LOCALE_SCURRENCY);
/* As on Linux: If the currency_symbol can't be represented in the
given charset, use int_curr_symbol. */
if (lc_wcstombs (f_wctomb, charset, NULL,
_monetary_locale->wcurrency_symbol,
if (lc_wcstombs (f_wctomb, NULL, _monetary_locale->wcurrency_symbol,
0, true) == (size_t) -1)
_monetary_locale->currency_symbol = _monetary_locale->int_curr_symbol;
else
@@ -1026,10 +1017,10 @@ __set_lc_messages_from_win (const char *name,
len += (strlen (charset) + 1);
if (lcid)
{
len += lc_wcstombs (f_wctomb, charset, NULL, msg->yesexpr, 0) + 1;
len += lc_wcstombs (f_wctomb, charset, NULL, msg->noexpr, 0) + 1;
len += lc_wcstombs (f_wctomb, charset, NULL, msg->yesstr, 0) + 1;
len += lc_wcstombs (f_wctomb, charset, NULL, msg->nostr, 0) + 1;
len += lc_wcstombs (f_wctomb, NULL, msg->yesexpr, 0) + 1;
len += lc_wcstombs (f_wctomb, NULL, msg->noexpr, 0) + 1;
len += lc_wcstombs (f_wctomb, NULL, msg->yesstr, 0) + 1;
len += lc_wcstombs (f_wctomb, NULL, msg->nostr, 0) + 1;
len += (wcslen (msg->yesexpr) + 1) * sizeof (wchar_t);
len += (wcslen (msg->noexpr) + 1) * sizeof (wchar_t);
len += (wcslen (msg->yesstr) + 1) * sizeof (wchar_t);
@@ -1051,13 +1042,13 @@ __set_lc_messages_from_win (const char *name,
if (lcid)
{
_messages_locale->yesexpr = (const char *) c;
len = lc_wcstombs (f_wctomb, charset, c, msg->yesexpr, lc_messages_end - c);
len = lc_wcstombs (f_wctomb, c, msg->yesexpr, lc_messages_end - c);
_messages_locale->noexpr = (const char *) (c += len + 1);
len = lc_wcstombs (f_wctomb, charset, c, msg->noexpr, lc_messages_end - c);
len = lc_wcstombs (f_wctomb, c, msg->noexpr, lc_messages_end - c);
_messages_locale->yesstr = (const char *) (c += len + 1);
len = lc_wcstombs (f_wctomb, charset, c, msg->yesstr, lc_messages_end - c);
len = lc_wcstombs (f_wctomb, c, msg->yesstr, lc_messages_end - c);
_messages_locale->nostr = (const char *) (c += len + 1);
len = lc_wcstombs (f_wctomb, charset, c, msg->nostr, lc_messages_end - c);
len = lc_wcstombs (f_wctomb, c, msg->nostr, lc_messages_end - c);
c += len + 1;
if ((uintptr_t) c % 1)
++c;
@@ -1149,15 +1140,14 @@ strcoll (const char *__restrict s1, const char *__restrict s2)
/* The ANSI version of CompareString uses the default charset of the lcid,
so we must use the Unicode version. */
mbtowc_p collate_mbtowc = __get_current_collate_locale ()->mbtowc;
const char *collate_charset = __get_current_collate_locale ()->codeset;
n1 = lc_mbstowcs (collate_mbtowc, collate_charset, NULL, s1, 0) + 1;
n1 = lc_mbstowcs (collate_mbtowc, NULL, s1, 0) + 1;
ws1 = (n1 > NT_MAX_PATH ? (wchar_t *) malloc (n1 * sizeof (wchar_t))
: tp.w_get ());
lc_mbstowcs (collate_mbtowc, collate_charset, ws1, s1, n1);
n2 = lc_mbstowcs (collate_mbtowc, collate_charset, NULL, s2, 0) + 1;
lc_mbstowcs (collate_mbtowc, ws1, s1, n1);
n2 = lc_mbstowcs (collate_mbtowc, NULL, s2, 0) + 1;
ws2 = (n2 > NT_MAX_PATH ? (wchar_t *) malloc (n2 * sizeof (wchar_t))
: tp.w_get ());
lc_mbstowcs (collate_mbtowc, collate_charset, ws2, s2, n2);
lc_mbstowcs (collate_mbtowc, ws2, s2, n2);
ret = CompareStringW (collate_lcid, 0, ws1, -1, ws2, -1);
if (n1 > NT_MAX_PATH)
free (ws1);
@@ -1226,13 +1216,12 @@ strxfrm (char *__restrict s1, const char *__restrict s2, size_t sn)
/* The ANSI version of LCMapString uses the default charset of the lcid,
so we must use the Unicode version. */
mbtowc_p collate_mbtowc = __get_current_collate_locale ()->mbtowc;
const char *collate_charset = __get_current_collate_locale ()->codeset;
n2 = lc_mbstowcs (collate_mbtowc, collate_charset, NULL, s2, 0) + 1;
n2 = lc_mbstowcs (collate_mbtowc, NULL, s2, 0) + 1;
ws2 = (n2 > NT_MAX_PATH ? (wchar_t *) malloc (n2 * sizeof (wchar_t))
: tp.w_get ());
if (ws2)
{
lc_mbstowcs (collate_mbtowc, collate_charset, ws2, s2, n2);
lc_mbstowcs (collate_mbtowc, ws2, s2, n2);
/* The sort key is a NUL-terminated byte string. */
ret = LCMapStringW (collate_lcid, LCMAP_SORTKEY, ws2, -1,
(PWCHAR) s1, sn);
@@ -1474,7 +1463,7 @@ __set_locale_from_locale_alias (const char *locale, char *new_locale)
if (strlen (replace) > ENCODING_LEN)
continue;
/* The file is latin1 encoded */
lc_mbstowcs (__iso_mbtowc, "ISO-8859-1", walias, alias, ENCODING_LEN + 1);
lc_mbstowcs (__iso_mbtowc (1), walias, alias, ENCODING_LEN + 1);
walias[ENCODING_LEN] = L'\0';
if (!wcscmp (wlocale, walias))
{
@@ -1503,33 +1492,25 @@ internal_setlocale ()
wchar_t *w_path = NULL, *w_cwd;
/* Don't do anything if the charset hasn't actually changed. */
if (strcmp (cygheap->locale.charset, __locale_charset ()) == 0)
if (cygheap->locale.mbtowc == __global_locale.mbtowc)
return;
debug_printf ("Cygwin charset changed from %s to %s",
cygheap->locale.charset, __locale_charset ());
debug_printf ("Cygwin charset chang to %s", __locale_charset ());
/* Fetch PATH and CWD and convert to wchar_t in previous charset. */
path = getenv ("PATH");
if (path && *path) /* $PATH can be potentially unset. */
{
w_path = tp.w_get ();
sys_mbstowcs (w_path, 32768, path);
sys_cp_mbstowcs (cygheap->locale.mbtowc, w_path, 32768, path);
}
w_cwd = tp.w_get ();
cwdstuff::cwd_lock.acquire ();
sys_mbstowcs (w_cwd, 32768, cygheap->cwd.get_posix ());
sys_cp_mbstowcs (cygheap->locale.mbtowc, w_cwd, 32768,
cygheap->cwd.get_posix ());
/* Set charset for internal conversion functions. */
if (*__locale_charset () == 'A'/*SCII*/)
{
cygheap->locale.mbtowc = __utf8_mbtowc;
cygheap->locale.wctomb = __utf8_wctomb;
}
else
{
cygheap->locale.mbtowc = __mbtowc;
cygheap->locale.wctomb = __wctomb;
}
strcpy (cygheap->locale.charset, __locale_charset ());
cygheap->locale.mbtowc = __global_locale.mbtowc;
if (cygheap->locale.mbtowc == __ascii_mbtowc)
cygheap->locale.mbtowc = __utf8_mbtowc;
/* Restore CWD and PATH in new charset. */
cygheap->cwd.reset_posix (w_cwd);
cwdstuff::cwd_lock.release ();

60
winsup/cygwin/strfuncs.cc
View File

@@ -140,15 +140,13 @@ __db_wctomb (struct _reent *r, char *s, wchar_t wchar, UINT cp)
}
extern "C" int
__sjis_wctomb (struct _reent *r, char *s, wchar_t wchar, const char *charset,
mbstate_t *state)
__sjis_wctomb (struct _reent *r, char *s, wchar_t wchar, mbstate_t *state)
{
return __db_wctomb (r,s, wchar, 932);
}
extern "C" int
__eucjp_wctomb (struct _reent *r, char *s, wchar_t wchar, const char *charset,
mbstate_t *state)
__eucjp_wctomb (struct _reent *r, char *s, wchar_t wchar, mbstate_t *state)
{
/* Unfortunately, the Windows eucJP codepage 20932 is not really 100%
compatible to eucJP. It's a cute approximation which makes it a
@@ -192,22 +190,19 @@ __eucjp_wctomb (struct _reent *r, char *s, wchar_t wchar, const char *charset,
}
extern "C" int
__gbk_wctomb (struct _reent *r, char *s, wchar_t wchar, const char *charset,
mbstate_t *state)
__gbk_wctomb (struct _reent *r, char *s, wchar_t wchar, mbstate_t *state)
{
return __db_wctomb (r,s, wchar, 936);
}
extern "C" int
__kr_wctomb (struct _reent *r, char *s, wchar_t wchar, const char *charset,
mbstate_t *state)
__kr_wctomb (struct _reent *r, char *s, wchar_t wchar, mbstate_t *state)
{
return __db_wctomb (r,s, wchar, 949);
}
extern "C" int
__big5_wctomb (struct _reent *r, char *s, wchar_t wchar, const char *charset,
mbstate_t *state)
__big5_wctomb (struct _reent *r, char *s, wchar_t wchar, mbstate_t *state)
{
return __db_wctomb (r,s, wchar, 950);
}
@@ -268,14 +263,14 @@ __db_mbtowc (struct _reent *r, wchar_t *pwc, const char *s, size_t n, UINT cp,
extern "C" int
__sjis_mbtowc (struct _reent *r, wchar_t *pwc, const char *s, size_t n,
const char *charset, mbstate_t *state)
mbstate_t *state)
{
return __db_mbtowc (r, pwc, s, n, 932, state);
}
extern "C" int
__eucjp_mbtowc (struct _reent *r, wchar_t *pwc, const char *s, size_t n,
const char *charset, mbstate_t *state)
mbstate_t *state)
{
/* See comment in __eucjp_wctomb above. */
wchar_t dummy;
@@ -352,21 +347,21 @@ jis_x_0212:
extern "C" int
__gbk_mbtowc (struct _reent *r, wchar_t *pwc, const char *s, size_t n,
const char *charset, mbstate_t *state)
mbstate_t *state)
{
return __db_mbtowc (r, pwc, s, n, 936, state);
}
extern "C" int
__kr_mbtowc (struct _reent *r, wchar_t *pwc, const char *s, size_t n,
const char *charset, mbstate_t *state)
mbstate_t *state)
{
return __db_mbtowc (r, pwc, s, n, 949, state);
}
extern "C" int
__big5_mbtowc (struct _reent *r, wchar_t *pwc, const char *s, size_t n,
const char *charset, mbstate_t *state)
mbstate_t *state)
{
return __db_mbtowc (r, pwc, s, n, 950, state);
}
@@ -408,7 +403,7 @@ __big5_mbtowc (struct _reent *r, wchar_t *pwc, const char *s, size_t n,
*/
static size_t __reg3
sys_wcstombs (char *dst, size_t len, const wchar_t *src, size_t nwc,
bool is_path)
bool is_path)
{
char buf[10];
char *ptr = dst;
@@ -416,9 +411,10 @@ sys_wcstombs (char *dst, size_t len, const wchar_t *src, size_t nwc,
size_t n = 0;
mbstate_t ps;
save_errno save;
wctomb_p f_wctomb = cygheap->locale.wctomb;
const char *charset = cygheap->locale.charset;
wctomb_p f_wctomb = __WCTOMB;
if (f_wctomb == __ascii_wctomb)
f_wctomb = __utf8_wctomb;
memset (&ps, 0, sizeof ps);
if (dst == NULL)
len = (size_t) -1;
@@ -441,13 +437,13 @@ sys_wcstombs (char *dst, size_t len, const wchar_t *src, size_t nwc,
}
else
{
bytes = f_wctomb (_REENT, buf, pw, charset, &ps);
if (bytes == -1 && *charset != 'U'/*TF-8*/)
bytes = f_wctomb (_REENT, buf, pw, &ps);
if (bytes == -1 && f_wctomb != __utf8_wctomb)
{
/* Convert chars invalid in the current codepage to a sequence
ASCII CAN; UTF-8 representation of invalid char. */
buf[0] = 0x18; /* ASCII CAN */
bytes = __utf8_wctomb (_REENT, buf + 1, pw, charset, &ps);
bytes = __utf8_wctomb (_REENT, buf + 1, pw, &ps);
if (bytes == -1)
{
++pwcs;
@@ -465,8 +461,7 @@ sys_wcstombs (char *dst, size_t len, const wchar_t *src, size_t nwc,
ps.__count = 0;
continue;
}
bytes += __utf8_wctomb (_REENT, buf + bytes, *pwcs, charset,
&ps);
bytes += __utf8_wctomb (_REENT, buf + bytes, *pwcs, &ps);
nwc--;
}
}
@@ -557,8 +552,8 @@ sys_wcstombs_alloc_no_path (char **dst_p, int type, const wchar_t *src,
charset, which is the charset returned by GetConsoleCP (). Most of the
time this is used for box and line drawing characters. */
size_t __reg3
sys_cp_mbstowcs (mbtowc_p f_mbtowc, const char *charset, wchar_t *dst,
size_t dlen, const char *src, size_t nms)
sys_cp_mbstowcs (mbtowc_p f_mbtowc, wchar_t *dst, size_t dlen,
const char *src, size_t nms)
{
wchar_t *ptr = dst;
unsigned const char *pmbs = (unsigned const char *) src;
@@ -581,10 +576,11 @@ sys_cp_mbstowcs (mbtowc_p f_mbtowc, const char *charset, wchar_t *dst,
next byte must be a valid UTF-8 start byte. If the charset
isn't UTF-8 anyway, try to convert the following bytes as UTF-8
sequence. */
if (nms > 2 && pmbs[1] >= 0xc2 && pmbs[1] <= 0xf4 && *charset != 'U'/*TF-8*/)
if (nms > 2 && pmbs[1] >= 0xc2 && pmbs[1] <= 0xf4
&& f_mbtowc != __utf8_mbtowc)
{
bytes = __utf8_mbtowc (_REENT, ptr, (const char *) pmbs + 1,
nms - 1, charset, &ps);
nms - 1, &ps);
if (bytes < 0)
{
/* Invalid UTF-8 sequence? Treat the ASCII CAN character as
@@ -603,7 +599,7 @@ sys_cp_mbstowcs (mbtowc_p f_mbtowc, const char *charset, wchar_t *dst,
wchar_t *ptr2 = dst ? ptr + 1 : NULL;
int bytes2 = __utf8_mbtowc (_REENT, ptr2,
(const char *) pmbs + bytes,
nms - bytes, charset, &ps);
nms - bytes, &ps);
if (bytes2 < 0)
memset (&ps, 0, sizeof ps);
else
@@ -625,7 +621,7 @@ sys_cp_mbstowcs (mbtowc_p f_mbtowc, const char *charset, wchar_t *dst,
}
}
else if ((bytes = f_mbtowc (_REENT, ptr, (const char *) pmbs, nms,
charset, &ps)) < 0)
&ps)) < 0)
{
/* The technique is based on a discussion here:
http://www.mail-archive.com/linux-utf8@nl.linux.org/msg00080.html
@@ -668,8 +664,10 @@ sys_cp_mbstowcs (mbtowc_p f_mbtowc, const char *charset, wchar_t *dst,
size_t __reg3
sys_mbstowcs (wchar_t * dst, size_t dlen, const char *src, size_t nms)
{
return sys_cp_mbstowcs (cygheap->locale.mbtowc, cygheap->locale.charset,
dst, dlen, src, nms);
mbtowc_p f_mbtowc = __MBTOWC;
if (f_mbtowc == __ascii_mbtowc)
f_mbtowc = __utf8_mbtowc;
return sys_cp_mbstowcs (f_mbtowc, dst, dlen, src, nms);
}
/* Same as sys_wcstombs_alloc, just backwards. */

40
winsup/cygwin/wchar.h
View File

@@ -11,6 +11,9 @@ details. */
#include_next <wchar.h>
/* Internal headers from newlib */
#include "../locale/setlocale.h"
#define ENCODING_LEN 31
#ifdef __cplusplus
@@ -18,29 +21,23 @@ extern "C" {
#endif
typedef int mbtowc_f (struct _reent *, wchar_t *, const char *, size_t,
const char *, mbstate_t *);
mbstate_t *);
typedef mbtowc_f *mbtowc_p;
extern mbtowc_p __mbtowc;
extern mbtowc_f __ascii_mbtowc;
extern mbtowc_f __utf8_mbtowc;
extern mbtowc_f __iso_mbtowc;
extern mbtowc_f __cp_mbtowc;
extern mbtowc_f __sjis_mbtowc;
extern mbtowc_f __eucjp_mbtowc;
extern mbtowc_f __gbk_mbtowc;
extern mbtowc_f __kr_mbtowc;
extern mbtowc_f __big5_mbtowc;
extern mbtowc_p __iso_mbtowc (int);
extern mbtowc_p __cp_mbtowc (int);
typedef int wctomb_f (struct _reent *, char *, wchar_t, const char *,
mbstate_t *);
#define __MBTOWC (__get_current_locale ()->mbtowc)
typedef int wctomb_f (struct _reent *, char *, wchar_t, mbstate_t *);
typedef wctomb_f *wctomb_p;
extern wctomb_p __wctomb;
extern wctomb_f __ascii_wctomb;
extern wctomb_f __utf8_wctomb;
extern char *__locale_charset ();
#define __WCTOMB (__get_current_locale ()->wctomb)
#ifdef __cplusplus
}
@@ -49,20 +46,21 @@ extern char *__locale_charset ();
#ifdef __INSIDE_CYGWIN__
#ifdef __cplusplus
size_t __reg3 sys_wcstombs (char *dst, size_t len, const wchar_t * src,
size_t nwc = (size_t) -1);
size_t nwc = (size_t) -1);
size_t __reg3 sys_wcstombs_no_path (char *dst, size_t len,
const wchar_t * src, size_t nwc = (size_t) -1);
const wchar_t * src,
size_t nwc = (size_t) -1);
size_t __reg3 sys_wcstombs_alloc (char **, int, const wchar_t *,
size_t = (size_t) -1);
size_t = (size_t) -1);
size_t __reg3 sys_wcstombs_alloc_no_path (char **, int, const wchar_t *,
size_t = (size_t) -1);
size_t = (size_t) -1);
size_t __reg3 sys_cp_mbstowcs (mbtowc_p, const char *, wchar_t *, size_t,
const char *, size_t = (size_t) -1);
size_t __reg3 sys_cp_mbstowcs (mbtowc_p, wchar_t *, size_t, const char *,
size_t = (size_t) -1);
size_t __reg3 sys_mbstowcs (wchar_t * dst, size_t dlen, const char *src,
size_t nms = (size_t) -1);
size_t nms = (size_t) -1);
size_t __reg3 sys_mbstowcs_alloc (wchar_t **, int, const char *,
size_t = (size_t) -1);
size_t = (size_t) -1);
#endif /* __cplusplus */
#endif /* __INSIDE_CYGWIN__ */