* otherwise to use the particular category's variable if set; otherwise to
* use the LANG variable. */
+ bool override_LC_ALL = 0;
+ char * result;
+
if (locale && strEQ(locale, "")) {
# ifdef LC_ALL
locale = PerlEnv_getenv("LC_ALL");
switch (category) {
# ifdef LC_ALL
case LC_ALL:
+ override_LC_ALL = TRUE;
break; /* We already know its variable isn't set */
# endif
# ifdef USE_LOCALE_TIME
# endif
}
- return setlocale(category, locale);
+ result = setlocale(category, locale);
+
+ if (! override_LC_ALL) {
+ return result;
+ }
+
+ /* Here the input locale was LC_ALL, and we have set it to what is in the
+ * LANG variable or the system default if there is no LANG. But these have
+ * lower priority than the other LC_foo variables, so override it for each
+ * one that is set. (If they are set to "", it means to use the same thing
+ * we just set LC_ALL to, so can skip) */
+# ifdef USE_LOCALE_TIME
+ result = PerlEnv_getenv("LC_TIME");
+ if (result && strNE(result, "")) {
+ setlocale(LC_TIME, result);
+ }
+# endif
+# ifdef USE_LOCALE_CTYPE
+ result = PerlEnv_getenv("LC_CTYPE");
+ if (result && strNE(result, "")) {
+ setlocale(LC_CTYPE, result);
+ }
+# endif
+# ifdef USE_LOCALE_COLLATE
+ result = PerlEnv_getenv("LC_COLLATE");
+ if (result && strNE(result, "")) {
+ setlocale(LC_COLLATE, result);
+ }
+# endif
+# ifdef USE_LOCALE_MONETARY
+ result = PerlEnv_getenv("LC_MONETARY");
+ if (result && strNE(result, "")) {
+ setlocale(LC_MONETARY, result);
+ }
+# endif
+# ifdef USE_LOCALE_NUMERIC
+ result = PerlEnv_getenv("LC_NUMERIC");
+ if (result && strNE(result, "")) {
+ setlocale(LC_NUMERIC, result);
+ }
+# endif
+# ifdef USE_LOCALE_MESSAGES
+ result = PerlEnv_getenv("LC_MESSAGES");
+ if (result && strNE(result, "")) {
+ setlocale(LC_MESSAGES, result);
+ }
+# endif
+
+ return setlocale(LC_ALL, NULL);
+
}
#endif
int
Perl_init_i18nl10n(pTHX_ int printwarn)
{
- int ok = 1;
- /* returns
+ /* printwarn is
+ *
+ * 0 if not to output warning when setup locale is bad
+ * 1 if to output warning based on value of PERL_BADLANG
+ * >1 if to output regardless of PERL_BADLANG
+ *
+ * returns
* 1 = set ok or not applicable,
- * 0 = fallback to C locale,
- * -1 = fallback to C locale failed
+ * 0 = fallback to a locale of lower priority
+ * -1 = fallback to all locales failed, not even to the C locale
*/
+ int ok = 1;
+
#if defined(USE_LOCALE)
dVAR;
#ifdef __GLIBC__
char * const language = PerlEnv_getenv("LANGUAGE");
#endif
+
/* NULL uses the existing already set up locale */
const char * const setlocale_init = (PerlEnv_getenv("PERL_SKIP_LOCALE_INIT"))
? NULL
: "";
+ const char* trial_locales[5]; /* 5 = 1 each for "", LC_ALL, LANG, "", C */
+ unsigned int trial_locales_count;
char * const lc_all = PerlEnv_getenv("LC_ALL");
char * const lang = PerlEnv_getenv("LANG");
bool setlocale_failure = FALSE;
+ unsigned int i;
+ char *p;
+ const bool locwarn = (printwarn > 1 ||
+ (printwarn &&
+ (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p))));
+ bool done = FALSE;
+ const char *system_default_locale = NULL;
+
-#ifdef LOCALE_ENVIRON_REQUIRED
+#ifndef LOCALE_ENVIRON_REQUIRED
+ PERL_UNUSED_VAR(done);
+#else
/*
* Ultrix setlocale(..., "") fails if there are no environment
* variables from which to get a locale name.
*/
- bool done = FALSE;
-
# ifdef LC_ALL
if (lang) {
if (my_setlocale(LC_ALL, setlocale_init))
else
curnum = savepv(curnum);
# endif /* USE_LOCALE_NUMERIC */
+# ifdef USE_LOCALE_MESSAGES
+ if (! my_setlocale(LC_MESSAGES,
+ (!done && (lang || PerlEnv_getenv("LC_MESSAGES")))
+ ? setlocale_init : NULL))
+ {
+ setlocale_failure = TRUE;
+ }
+# endif /* USE_LOCALE_MESSAGES */
+# ifdef USE_LOCALE_MONETARY
+ if (! my_setlocale(LC_MONETARY,
+ (!done && (lang || PerlEnv_getenv("LC_MONETARY")))
+ ? setlocale_init : NULL))
+ {
+ setlocale_failure = TRUE;
+ }
+# endif /* USE_LOCALE_MONETARY */
}
# endif /* LC_ALL */
#endif /* !LOCALE_ENVIRON_REQUIRED */
+ /* We try each locale in the list until we get one that works, or exhaust
+ * the list */
+ trial_locales[0] = setlocale_init;
+ trial_locales_count = 1;
+ for (i= 0; i < trial_locales_count; i++) {
+ const char * trial_locale = trial_locales[i];
+
+ if (i > 0) {
+
+ /* XXX This is to preserve old behavior for LOCALE_ENVIRON_REQUIRED
+ * when i==0, but I (khw) don't think that behavior makes much
+ * sense */
+ setlocale_failure = FALSE;
+
+#ifdef WIN32
+
+ /* On Windows machines, an entry of "" after the 0th means to use
+ * the system default locale, which we now proceed to get. */
+ if (strEQ(trial_locale, "")) {
+ unsigned int j;
+
+ /* Note that this may change the locale, but we are going to do
+ * that anyway just below */
+ system_default_locale = setlocale(LC_ALL, "");
+
+ /* Skip if invalid or it's already on the list of locales to
+ * try */
+ if (! system_default_locale) {
+ goto next_iteration;
+ }
+ for (j = 0; j < trial_locales_count; j++) {
+ if (strEQ(system_default_locale, trial_locales[j])) {
+ goto next_iteration;
+ }
+ }
+
+ trial_locale = system_default_locale;
+ }
+#endif
+ }
+
#ifdef LC_ALL
- if (! my_setlocale(LC_ALL, setlocale_init))
- setlocale_failure = TRUE;
+ if (! my_setlocale(LC_ALL, trial_locale)) {
+ setlocale_failure = TRUE;
+ }
+ else {
+ /* Since LC_ALL succeeded, it should have changed all the other
+ * categories it can to its value; so we massage things so that the
+ * setlocales below just return their category's current values.
+ * This adequately handles the case in NetBSD where LC_COLLATE may
+ * not be defined for a locale, and setting it individually will
+ * fail, whereas setting LC_ALL suceeds, leaving LC_COLLATE set to
+ * the POSIX locale. */
+ trial_locale = NULL;
+ }
#endif /* LC_ALL */
- if (!setlocale_failure) {
+ if (!setlocale_failure) {
#ifdef USE_LOCALE_CTYPE
- Safefree(curctype);
- if (! (curctype = my_setlocale(LC_CTYPE, setlocale_init)))
- setlocale_failure = TRUE;
- else
- curctype = savepv(curctype);
+ Safefree(curctype);
+ if (! (curctype = my_setlocale(LC_CTYPE, trial_locale)))
+ setlocale_failure = TRUE;
+ else
+ curctype = savepv(curctype);
#endif /* USE_LOCALE_CTYPE */
#ifdef USE_LOCALE_COLLATE
- Safefree(curcoll);
- if (! (curcoll = my_setlocale(LC_COLLATE, setlocale_init)))
- setlocale_failure = TRUE;
- else
- curcoll = savepv(curcoll);
+ Safefree(curcoll);
+ if (! (curcoll = my_setlocale(LC_COLLATE, trial_locale)))
+ setlocale_failure = TRUE;
+ else
+ curcoll = savepv(curcoll);
#endif /* USE_LOCALE_COLLATE */
#ifdef USE_LOCALE_NUMERIC
- Safefree(curnum);
- if (! (curnum = my_setlocale(LC_NUMERIC, setlocale_init)))
- setlocale_failure = TRUE;
- else
- curnum = savepv(curnum);
+ Safefree(curnum);
+ if (! (curnum = my_setlocale(LC_NUMERIC, trial_locale)))
+ setlocale_failure = TRUE;
+ else
+ curnum = savepv(curnum);
#endif /* USE_LOCALE_NUMERIC */
- }
+#ifdef USE_LOCALE_MESSAGES
+ if (! (my_setlocale(LC_MESSAGES, trial_locale)))
+ setlocale_failure = TRUE;
+#endif /* USE_LOCALE_MESSAGES */
+#ifdef USE_LOCALE_MONETARY
+ if (! (my_setlocale(LC_MONETARY, trial_locale)))
+ setlocale_failure = TRUE;
+#endif /* USE_LOCALE_MONETARY */
+
+ if (! setlocale_failure) { /* Success */
+ break;
+ }
+ }
+
+ /* Here, something failed; will need to try a fallback. */
+ ok = 0;
- if (setlocale_failure) {
- char *p;
- const bool locwarn = (printwarn > 1 ||
- (printwarn &&
- (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p))));
+ if (i == 0) {
+ unsigned int j;
- if (locwarn) {
+ if (locwarn) { /* Output failure info only on the first one */
#ifdef LC_ALL
- PerlIO_printf(Perl_error_log,
- "perl: warning: Setting locale failed.\n");
+ PerlIO_printf(Perl_error_log,
+ "perl: warning: Setting locale failed.\n");
#else /* !LC_ALL */
- PerlIO_printf(Perl_error_log,
- "perl: warning: Setting locale failed for the categories:\n\t");
+ PerlIO_printf(Perl_error_log,
+ "perl: warning: Setting locale failed for the categories:\n\t");
#ifdef USE_LOCALE_CTYPE
- if (! curctype)
- PerlIO_printf(Perl_error_log, "LC_CTYPE ");
+ if (! curctype)
+ PerlIO_printf(Perl_error_log, "LC_CTYPE ");
#endif /* USE_LOCALE_CTYPE */
#ifdef USE_LOCALE_COLLATE
- if (! curcoll)
- PerlIO_printf(Perl_error_log, "LC_COLLATE ");
+ if (! curcoll)
+ PerlIO_printf(Perl_error_log, "LC_COLLATE ");
#endif /* USE_LOCALE_COLLATE */
#ifdef USE_LOCALE_NUMERIC
- if (! curnum)
- PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
+ if (! curnum)
+ PerlIO_printf(Perl_error_log, "LC_NUMERIC ");
#endif /* USE_LOCALE_NUMERIC */
- PerlIO_printf(Perl_error_log, "\n");
+ PerlIO_printf(Perl_error_log, "and possibly others\n");
#endif /* LC_ALL */
- PerlIO_printf(Perl_error_log,
- "perl: warning: Please check that your locale settings:\n");
+ PerlIO_printf(Perl_error_log,
+ "perl: warning: Please check that your locale settings:\n");
#ifdef __GLIBC__
- PerlIO_printf(Perl_error_log,
- "\tLANGUAGE = %c%s%c,\n",
- language ? '"' : '(',
- language ? language : "unset",
- language ? '"' : ')');
+ PerlIO_printf(Perl_error_log,
+ "\tLANGUAGE = %c%s%c,\n",
+ language ? '"' : '(',
+ language ? language : "unset",
+ language ? '"' : ')');
#endif
- PerlIO_printf(Perl_error_log,
- "\tLC_ALL = %c%s%c,\n",
- lc_all ? '"' : '(',
- lc_all ? lc_all : "unset",
- lc_all ? '"' : ')');
+ PerlIO_printf(Perl_error_log,
+ "\tLC_ALL = %c%s%c,\n",
+ lc_all ? '"' : '(',
+ lc_all ? lc_all : "unset",
+ lc_all ? '"' : ')');
#if defined(USE_ENVIRON_ARRAY)
- {
- char **e;
- for (e = environ; *e; e++) {
- if (strnEQ(*e, "LC_", 3)
- && strnNE(*e, "LC_ALL=", 7)
- && (p = strchr(*e, '=')))
- PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
- (int)(p - *e), *e, p + 1);
- }
- }
+ {
+ char **e;
+ for (e = environ; *e; e++) {
+ if (strnEQ(*e, "LC_", 3)
+ && strnNE(*e, "LC_ALL=", 7)
+ && (p = strchr(*e, '=')))
+ PerlIO_printf(Perl_error_log, "\t%.*s = \"%s\",\n",
+ (int)(p - *e), *e, p + 1);
+ }
+ }
#else
- PerlIO_printf(Perl_error_log,
- "\t(possibly more locale environment variables)\n");
+ PerlIO_printf(Perl_error_log,
+ "\t(possibly more locale environment variables)\n");
#endif
- PerlIO_printf(Perl_error_log,
- "\tLANG = %c%s%c\n",
- lang ? '"' : '(',
- lang ? lang : "unset",
- lang ? '"' : ')');
+ PerlIO_printf(Perl_error_log,
+ "\tLANG = %c%s%c\n",
+ lang ? '"' : '(',
+ lang ? lang : "unset",
+ lang ? '"' : ')');
- PerlIO_printf(Perl_error_log,
- " are supported and installed on your system.\n");
- }
+ PerlIO_printf(Perl_error_log,
+ " are supported and installed on your system.\n");
+ }
-#ifdef LC_ALL
+ /* Calculate what fallback locales to try. We have avoided this
+ * until we have to, becuase failure is quite unlikely. This will
+ * usually change the upper bound of the loop we are in.
+ *
+ * Since the system's default way of setting the locale has not
+ * found one that works, We use Perl's defined ordering: LC_ALL,
+ * LANG, and the C locale. We don't try the same locale twice, so
+ * don't add to the list if already there. (On POSIX systems, the
+ * LC_ALL element will likely be a repeat of the 0th element "",
+ * but there's no harm done by doing it explicitly */
+ if (lc_all) {
+ for (j = 0; j < trial_locales_count; j++) {
+ if (strEQ(lc_all, trial_locales[j])) {
+ goto done_lc_all;
+ }
+ }
+ trial_locales[trial_locales_count++] = lc_all;
+ }
+ done_lc_all:
- if (setlocale(LC_ALL, "C")) {
- if (locwarn)
- PerlIO_printf(Perl_error_log,
- "perl: warning: Falling back to the standard locale (\"C\").\n");
- ok = 0;
- }
- else {
- if (locwarn)
- PerlIO_printf(Perl_error_log,
- "perl: warning: Failed to fall back to the standard locale (\"C\").\n");
- ok = -1;
- }
+ if (lang) {
+ for (j = 0; j < trial_locales_count; j++) {
+ if (strEQ(lang, trial_locales[j])) {
+ goto done_lang;
+ }
+ }
+ trial_locales[trial_locales_count++] = lang;
+ }
+ done_lang:
+
+#if defined(WIN32) && defined(LC_ALL)
+ /* For Windows, we also try the system default locale before "C".
+ * (If there exists a Windows without LC_ALL we skip this because
+ * it gets too complicated. For those, the "C" is the next
+ * fallback possibility). The "" is the same as the 0th element of
+ * the array, but the code at the loop above knows to treat it
+ * differently when not the 0th */
+ trial_locales[trial_locales_count++] = "";
+#endif
-#else /* ! LC_ALL */
+ for (j = 0; j < trial_locales_count; j++) {
+ if (strEQ("C", trial_locales[j])) {
+ goto done_C;
+ }
+ }
+ trial_locales[trial_locales_count++] = "C";
- if (0
-#ifdef USE_LOCALE_CTYPE
- || !(curctype || setlocale(LC_CTYPE, "C"))
-#endif /* USE_LOCALE_CTYPE */
-#ifdef USE_LOCALE_COLLATE
- || !(curcoll || setlocale(LC_COLLATE, "C"))
-#endif /* USE_LOCALE_COLLATE */
-#ifdef USE_LOCALE_NUMERIC
- || !(curnum || setlocale(LC_NUMERIC, "C"))
-#endif /* USE_LOCALE_NUMERIC */
- )
- {
- if (locwarn)
- PerlIO_printf(Perl_error_log,
- "perl: warning: Cannot fall back to the standard locale (\"C\").\n");
- ok = -1;
- }
+ done_C: ;
+ } /* end of first time through the loop */
+
+#ifdef WIN32
+ next_iteration: ;
+#endif
+
+ } /* end of looping through the trial locales */
+
+ if (ok < 1) { /* If we tried to fallback */
+ const char* msg;
+ if (! setlocale_failure) { /* fallback succeeded */
+ msg = "Falling back to";
+ }
+ else { /* fallback failed */
+
+ /* We dropped off the end of the loop, so have to decrement i to
+ * get back to the value the last time through */
+ i--;
-#endif /* ! LC_ALL */
+ ok = -1;
+ msg = "Failed to fall back to";
+ /* To continue, we should use whatever values we've got */
#ifdef USE_LOCALE_CTYPE
- Safefree(curctype);
- curctype = savepv(setlocale(LC_CTYPE, NULL));
+ Safefree(curctype);
+ curctype = savepv(setlocale(LC_CTYPE, NULL));
#endif /* USE_LOCALE_CTYPE */
#ifdef USE_LOCALE_COLLATE
- Safefree(curcoll);
- curcoll = savepv(setlocale(LC_COLLATE, NULL));
+ Safefree(curcoll);
+ curcoll = savepv(setlocale(LC_COLLATE, NULL));
#endif /* USE_LOCALE_COLLATE */
#ifdef USE_LOCALE_NUMERIC
- Safefree(curnum);
- curnum = savepv(setlocale(LC_NUMERIC, NULL));
+ Safefree(curnum);
+ curnum = savepv(setlocale(LC_NUMERIC, NULL));
#endif /* USE_LOCALE_NUMERIC */
- }
- else {
+ }
+
+ if (locwarn) {
+ const char * description;
+ const char * name = "";
+ if (strEQ(trial_locales[i], "C")) {
+ description = "the standard locale";
+ name = "C";
+ }
+ else if (strEQ(trial_locales[i], "")) {
+ description = "the system default locale";
+ if (system_default_locale) {
+ name = system_default_locale;
+ }
+ }
+ else {
+ description = "a fallback locale";
+ name = trial_locales[i];
+ }
+ if (name && strNE(name, "")) {
+ PerlIO_printf(Perl_error_log,
+ "perl: warning: %s %s (\"%s\").\n", msg, description, name);
+ }
+ else {
+ PerlIO_printf(Perl_error_log,
+ "perl: warning: %s %s.\n", msg, description);
+ }
+ }
+ } /* End of tried to fallback */
#ifdef USE_LOCALE_CTYPE
new_ctype(curctype);
new_numeric(curnum);
#endif /* USE_LOCALE_NUMERIC */
- }
-
#if defined(USE_PERLIO) && defined(USE_LOCALE_CTYPE)
- {
- /* Set PL_utf8locale to TRUE if using PerlIO _and_
- the current LC_CTYPE locale is UTF-8.
- If PL_utf8locale and PL_unicode (set by -C or by $ENV{PERL_UNICODE})
- are true, perl.c:S_parse_body() will turn on the PerlIO :utf8 layer
- on STDIN, STDOUT, STDERR, _and_ the default open discipline.
- */
- PL_utf8locale = is_cur_LC_category_utf8(LC_CTYPE);
- }
+ /* Set PL_utf8locale to TRUE if using PerlIO _and_ the current LC_CTYPE
+ * locale is UTF-8. If PL_utf8locale and PL_unicode (set by -C or by
+ * $ENV{PERL_UNICODE}) are true, perl.c:S_parse_body() will turn on the
+ * PerlIO :utf8 layer on STDIN, STDOUT, STDERR, _and_ the default open
+ * discipline. */
+ PL_utf8locale = is_cur_LC_category_utf8(LC_CTYPE);
+
/* Set PL_unicode to $ENV{PERL_UNICODE} if using PerlIO.
This is an alternative to using the -C command line switch
(the -C if present will override this). */
return ok;
}
+
#ifdef USE_LOCALE_COLLATE
/*
Safefree(save_input_locale);
return is_utf8;
}
+ Safefree(codeset);
}
# endif
(void) mbtowc(&wc, NULL, 0); /* Reset any shift state */
GCC_DIAG_RESTORE;
errno = 0;
- if (mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8))
+ if ((size_t)mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8))
!= strlen(HYPHEN_UTF8)
|| wc != (wchar_t) 0x2010)
{
&& *(save_input_locale + final_pos - 3) == '5'
&& *(save_input_locale + final_pos - 4) == '6')
{
- Safefree(save_input_locale);
DEBUG_L(PerlIO_printf(Perl_debug_log,
"Locale %s ends with 10056 in name, is UTF-8 locale\n",
save_input_locale));
+ Safefree(save_input_locale);
return TRUE;
}
#endif
/* Other common encodings are the ISO 8859 series, which aren't UTF-8 */
if (instr(save_input_locale, "8859")) {
- Safefree(save_input_locale);
DEBUG_L(PerlIO_printf(Perl_debug_log,
"Locale %s has 8859 in name, not UTF-8 locale\n",
save_input_locale));
+ Safefree(save_input_locale);
return FALSE;
}