X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/97f4de96034f2588862da2cb99fabf16f8009223..334afb3ef43aa91a8701c4ca7e753a5ec8e71f88:/locale.c diff --git a/locale.c b/locale.c index e4e6650..2e68b23 100644 --- a/locale.c +++ b/locale.c @@ -159,7 +159,7 @@ Perl_new_numeric(pTHX_ const char *newnum) * dot. * * This sets several interpreter-level variables: - * PL_numeric_name The default locale's name: a copy of 'newnum' + * PL_numeric_name The underlying locale's name: a copy of 'newnum' * PL_numeric_local A boolean indicating if the toggled state is such * that the current locale is the program's underlying * locale @@ -527,7 +527,8 @@ Perl_init_i18nl10n(pTHX_ int printwarn) char *p; const bool locwarn = (printwarn > 1 || (printwarn && - (!(p = PerlEnv_getenv("PERL_BADLANG")) || atoi(p)))); + (!(p = PerlEnv_getenv("PERL_BADLANG")) || + grok_atou(p, NULL)))); bool done = FALSE; #ifdef WIN32 /* In some systems you can find out the system default locale @@ -1010,8 +1011,10 @@ Perl__is_cur_LC_category_utf8(pTHX_ int category) /* Returns TRUE if the current locale for 'category' is UTF-8; FALSE * otherwise. 'category' may not be LC_ALL. If the platform doesn't have * nl_langinfo(), nor MB_CUR_MAX, this employs a heuristic, which hence - * could give the wrong result. It errs on the side of not being a UTF-8 - * locale. */ + * could give the wrong result. The result will very likely be correct for + * languages that have commonly used non-ASCII characters, but for notably + * English, it comes down to if the locale's name ends in something like + * "UTF-8". It errs on the side of not being a UTF-8 locale. */ char *save_input_locale = NULL; STRLEN final_pos; @@ -1048,12 +1051,13 @@ Perl__is_cur_LC_category_utf8(pTHX_ int category) if (category != LC_CTYPE) { /* These work only on LC_CTYPE */ /* Get the current LC_CTYPE locale */ - save_ctype_locale = stdize_locale(savepv(setlocale(LC_CTYPE, NULL))); + save_ctype_locale = setlocale(LC_CTYPE, NULL); if (! save_ctype_locale) { DEBUG_L(PerlIO_printf(Perl_debug_log, "Could not find current locale for LC_CTYPE\n")); goto cant_use_nllanginfo; } + save_ctype_locale = stdize_locale(savepv(save_ctype_locale)); /* If LC_CTYPE and the desired category use the same locale, this * means that finding the value for LC_CTYPE is the same as finding @@ -1081,8 +1085,9 @@ Perl__is_cur_LC_category_utf8(pTHX_ int category) # if defined(HAS_NL_LANGINFO) && defined(CODESET) { - char *codeset = savepv(nl_langinfo(CODESET)); + char *codeset = nl_langinfo(CODESET); if (codeset && strNE(codeset, "")) { + codeset = savepv(codeset); /* If we switched LC_CTYPE, switch back */ if (save_ctype_locale) { @@ -1100,7 +1105,6 @@ Perl__is_cur_LC_category_utf8(pTHX_ int category) Safefree(save_input_locale); return is_utf8; } - Safefree(codeset); } # endif @@ -1129,9 +1133,7 @@ Perl__is_cur_LC_category_utf8(pTHX_ int category) * result */ if (is_utf8) { wchar_t wc; - GCC_DIAG_IGNORE(-Wunused-result); - (void) mbtowc(&wc, NULL, 0); /* Reset any shift state */ - GCC_DIAG_RESTORE; + PERL_UNUSED_RESULT(mbtowc(&wc, NULL, 0));/* Reset any shift state */ errno = 0; if ((size_t)mbtowc(&wc, HYPHEN_UTF8, strlen(HYPHEN_UTF8)) != strlen(HYPHEN_UTF8) @@ -1158,60 +1160,64 @@ Perl__is_cur_LC_category_utf8(pTHX_ int category) cant_use_nllanginfo: -#endif /* HAS_NL_LANGINFO etc */ +#else /* nl_langinfo should work if available, so don't bother compiling this + fallback code. The final fallback of looking at the name is + compiled, and will be executed if nl_langinfo fails */ /* nl_langinfo not available or failed somehow. Next try looking at the * currency symbol to see if it disambiguates things. Often that will be * in the native script, and if the symbol isn't in UTF-8, we know that the * locale isn't. If it is non-ASCII UTF-8, we infer that the locale is - * too. */ + * too, as the odds of a non-UTF8 string being valid UTF-8 are quite small + * */ #ifdef HAS_LOCALECONV # ifdef USE_LOCALE_MONETARY { char *save_monetary_locale = NULL; - bool illegal_utf8 = FALSE; bool only_ascii = FALSE; - const struct lconv* const lc = localeconv(); + bool is_utf8 = FALSE; + struct lconv* lc; /* Like above for LC_CTYPE, we first set LC_MONETARY to the locale of * the desired category, if it isn't that locale already */ if (category != LC_MONETARY) { - save_monetary_locale = stdize_locale(savepv(setlocale(LC_MONETARY, - NULL))); + save_monetary_locale = setlocale(LC_MONETARY, NULL); if (! save_monetary_locale) { DEBUG_L(PerlIO_printf(Perl_debug_log, "Could not find current locale for LC_MONETARY\n")); goto cant_use_monetary; } + save_monetary_locale = stdize_locale(savepv(save_monetary_locale)); - if (strNE(save_monetary_locale, save_input_locale)) { - if (! setlocale(LC_MONETARY, save_input_locale)) { - DEBUG_L(PerlIO_printf(Perl_debug_log, - "Could not change LC_MONETARY locale to %s\n", - save_input_locale)); - Safefree(save_monetary_locale); - goto cant_use_monetary; - } + if (strEQ(save_monetary_locale, save_input_locale)) { + Safefree(save_monetary_locale); + save_monetary_locale = NULL; + } + else if (! setlocale(LC_MONETARY, save_input_locale)) { + DEBUG_L(PerlIO_printf(Perl_debug_log, + "Could not change LC_MONETARY locale to %s\n", + save_input_locale)); + Safefree(save_monetary_locale); + goto cant_use_monetary; } } /* Here the current LC_MONETARY is set to the locale of the category * whose information is desired. */ - if (lc && lc->currency_symbol) { - if (! is_utf8_string((U8 *) lc->currency_symbol, 0)) { - DEBUG_L(PerlIO_printf(Perl_debug_log, - "Currency symbol for %s is not legal UTF-8\n", - save_input_locale)); - illegal_utf8 = TRUE; - } - else if (is_ascii_string((U8 *) lc->currency_symbol, 0)) { - DEBUG_L(PerlIO_printf(Perl_debug_log, "Currency symbol for %s contains only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale)); - only_ascii = TRUE; - } + lc = localeconv(); + if (! lc + || ! lc->currency_symbol + || is_ascii_string((U8 *) lc->currency_symbol, 0)) + { + DEBUG_L(PerlIO_printf(Perl_debug_log, "Couldn't get currency symbol for %s, or contains only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale)); + only_ascii = TRUE; + } + else { + is_utf8 = is_utf8_string((U8 *) lc->currency_symbol, 0); } /* If we changed it, restore LC_MONETARY to its original locale */ @@ -1220,85 +1226,179 @@ Perl__is_cur_LC_category_utf8(pTHX_ int category) Safefree(save_monetary_locale); } - if (only_ascii) { - goto cant_use_monetary; - } - - Safefree(save_input_locale); + if (! only_ascii) { - /* It isn't a UTF-8 locale if the symbol is not legal UTF-8; otherwise - * assume the locale is UTF-8 if and only if the symbol is non-ascii - * UTF-8. (We can't really tell if the locale is UTF-8 or not if the - * symbol is just a '$', so we err on the side of it not being UTF-8) - * */ - DEBUG_L(PerlIO_printf(Perl_debug_log, "\tis_utf8=%d\n", ! illegal_utf8)); - return ! illegal_utf8; + /* It isn't a UTF-8 locale if the symbol is not legal UTF-8; + * otherwise assume the locale is UTF-8 if and only if the symbol + * is non-ascii UTF-8. */ + DEBUG_L(PerlIO_printf(Perl_debug_log, "\t?Currency symbol for %s is UTF-8=%d\n", + save_input_locale, is_utf8)); + Safefree(save_input_locale); + return is_utf8; + } } cant_use_monetary: # endif /* USE_LOCALE_MONETARY */ #endif /* HAS_LOCALECONV */ -#if 0 && defined(HAS_STRERROR) && defined(USE_LOCALE_MESSAGES) +#if defined(HAS_STRFTIME) && defined(USE_LOCALE_TIME) + +/* Still haven't found a non-ASCII string to disambiguate UTF-8 or not. Try + * the names of the months and weekdays, timezone, and am/pm indicator */ + { + char *save_time_locale = NULL; + int hour = 10; + bool is_dst = FALSE; + int dom = 1; + int month = 0; + int i; + char * formatted_time; + + + /* Like above for LC_MONETARY, we set LC_TIME to the locale of the + * desired category, if it isn't that locale already */ + + if (category != LC_TIME) { + + save_time_locale = setlocale(LC_TIME, NULL); + if (! save_time_locale) { + DEBUG_L(PerlIO_printf(Perl_debug_log, + "Could not find current locale for LC_TIME\n")); + goto cant_use_time; + } + save_time_locale = stdize_locale(savepv(save_time_locale)); + + if (strEQ(save_time_locale, save_input_locale)) { + Safefree(save_time_locale); + save_time_locale = NULL; + } + else if (! setlocale(LC_TIME, save_input_locale)) { + DEBUG_L(PerlIO_printf(Perl_debug_log, + "Could not change LC_TIME locale to %s\n", + save_input_locale)); + Safefree(save_time_locale); + goto cant_use_time; + } + } + + /* Here the current LC_TIME is set to the locale of the category + * whose information is desired. Look at all the days of the week and + * month names, and the timezone and am/pm indicator for non-ASCII + * characters. The first such a one found will tell us if the locale + * is UTF-8 or not */ + + for (i = 0; i < 7 + 12; i++) { /* 7 days; 12 months */ + formatted_time = my_strftime("%A %B %Z %p", + 0, 0, hour, dom, month, 112, 0, 0, is_dst); + if (! formatted_time || is_ascii_string((U8 *) formatted_time, 0)) { + + /* Here, we didn't find a non-ASCII. Try the next time through + * with the complemented dst and am/pm, and try with the next + * weekday. After we have gotten all weekdays, try the next + * month */ + is_dst = ! is_dst; + hour = (hour + 12) % 24; + dom++; + if (i > 6) { + month++; + } + continue; + } + + /* Here, we have a non-ASCII. Return TRUE is it is valid UTF8; + * false otherwise. But first, restore LC_TIME to its original + * locale if we changed it */ + if (save_time_locale) { + setlocale(LC_TIME, save_time_locale); + Safefree(save_time_locale); + } + + DEBUG_L(PerlIO_printf(Perl_debug_log, "\t?time-related strings for %s are UTF-8=%d\n", + save_input_locale, + is_utf8_string((U8 *) formatted_time, 0))); + Safefree(save_input_locale); + return is_utf8_string((U8 *) formatted_time, 0); + } + + /* Falling off the end of the loop indicates all the names were just + * ASCII. Go on to the next test. If we changed it, restore LC_TIME + * to its original locale */ + if (save_time_locale) { + setlocale(LC_TIME, save_time_locale); + Safefree(save_time_locale); + } + DEBUG_L(PerlIO_printf(Perl_debug_log, "All time-related words for %s contain only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale)); + } + cant_use_time: + +#endif + +#if 0 && defined(USE_LOCALE_MESSAGES) && defined(HAS_SYS_ERRLIST) /* This code is ifdefd out because it was found to not be necessary in testing - * on our dromedary test machine, which has over 700 locales. There, looking - * at just the currency symbol gave essentially the same results as doing this - * extra work. Executing this also caused segfaults in miniperl. I left it in - * so as to avoid rewriting it if real-world experience indicates that - * dromedary is an outlier. Essentially, instead of returning abpve if we + * on our dromedary test machine, which has over 700 locales. There, this + * added no value to looking at the currency symbol and the time strings. I + * left it in so as to avoid rewriting it if real-world experience indicates + * that dromedary is an outlier. Essentially, instead of returning abpve if we * haven't found illegal utf8, we continue on and examine all the strerror() * messages on the platform for utf8ness. If all are ASCII, we still don't * know the answer; but otherwise we have a pretty good indication of the - * utf8ness. The reason this doesn't necessarily help much is that the - * messages may not have been translated into the locale. The currency symbol - * is much more likely to have been translated. The code below would need to - * be altered somewhat to just be a continuation of testing the currency - * symbol. */ + * utf8ness. The reason this doesn't help much is that the messages may not + * have been translated into the locale. The currency symbol and time strings + * are much more likely to have been translated. */ + { int e; - unsigned int failures = 0, non_ascii = 0; + bool is_utf8 = FALSE; + bool non_ascii = FALSE; char *save_messages_locale = NULL; + const char * errmsg = NULL; - /* Like above for LC_CTYPE, we set LC_MESSAGES to the locale of the - * desired category, if it isn't that locale already */ + /* Like above, we set LC_MESSAGES to the locale of the desired + * category, if it isn't that locale already */ if (category != LC_MESSAGES) { - save_messages_locale = stdize_locale(savepv(setlocale(LC_MESSAGES, - NULL))); + save_messages_locale = setlocale(LC_MESSAGES, NULL); if (! save_messages_locale) { + DEBUG_L(PerlIO_printf(Perl_debug_log, + "Could not find current locale for LC_MESSAGES\n")); goto cant_use_messages; } + save_messages_locale = stdize_locale(savepv(save_messages_locale)); if (strEQ(save_messages_locale, save_input_locale)) { - Safefree(save_input_locale); + Safefree(save_messages_locale); + save_messages_locale = NULL; } else if (! setlocale(LC_MESSAGES, save_input_locale)) { + DEBUG_L(PerlIO_printf(Perl_debug_log, + "Could not change LC_MESSAGES locale to %s\n", + save_input_locale)); Safefree(save_messages_locale); goto cant_use_messages; } } /* Here the current LC_MESSAGES is set to the locale of the category - * whose information is desired. Look through all the messages */ + * whose information is desired. Look through all the messages. We + * can't use Strerror() here because it may expand to code that + * segfaults in miniperl */ - for (e = 0; -#ifdef HAS_SYS_ERRLIST - e <= sys_nerr -#endif - ; e++) - { - const U8* const errmsg = (U8 *) Strerror(e) ; - if (!errmsg) - break; - if (! is_utf8_string(errmsg, 0)) { - failures++; + for (e = 0; e <= sys_nerr; e++) { + errno = 0; + errmsg = sys_errlist[e]; + if (errno || !errmsg) { break; } - else if (! is_ascii_string(errmsg, 0)) { - non_ascii++; + errmsg = savepv(errmsg); + if (! is_ascii_string((U8 *) errmsg, 0)) { + non_ascii = TRUE; + is_utf8 = is_utf8_string((U8 *) errmsg, 0); + break; } } + Safefree(errmsg); /* And, if we changed it, restore LC_MESSAGES to its original locale */ if (save_messages_locale) { @@ -1306,15 +1406,25 @@ Perl__is_cur_LC_category_utf8(pTHX_ int category) Safefree(save_messages_locale); } - /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid, - * any non-ascii means it is one; otherwise we assume it isn't */ - return (failures) ? FALSE : non_ascii; + if (non_ascii) { + + /* Any non-UTF-8 message means not a UTF-8 locale; if all are valid, + * any non-ascii means it is one; otherwise we assume it isn't */ + DEBUG_L(PerlIO_printf(Perl_debug_log, "\t?error messages for %s are UTF-8=%d\n", + save_input_locale, + is_utf8)); + Safefree(save_input_locale); + return is_utf8; + } + DEBUG_L(PerlIO_printf(Perl_debug_log, "All error messages for %s contain only ASCII; can't use for determining if UTF-8 locale\n", save_input_locale)); } cant_use_messages: #endif +#endif /* the code that is compiled when no nl_langinfo */ + /* As a last resort, look at the locale name to see if it matches * qr/UTF -? * 8 /ix, or some other common locale names. This "name", the * return of setlocale(), is actually defined to be opaque, so we can't @@ -1330,8 +1440,8 @@ Perl__is_cur_LC_category_utf8(pTHX_ int category) while ((name += strcspn(name, "Uu") + 1) <= save_input_locale + final_pos - 2) { - if (toFOLD(*(name)) != 't' - || toFOLD(*(name + 1)) != 'f') + if (!isALPHA_FOLD_NE(*name, 't') + || isALPHA_FOLD_NE(*(name + 1), 'f')) { continue; } @@ -1343,10 +1453,10 @@ Perl__is_cur_LC_category_utf8(pTHX_ int category) name++; } if (*(name) == '8') { - Safefree(save_input_locale); DEBUG_L(PerlIO_printf(Perl_debug_log, "Locale %s ends with UTF-8 in name\n", save_input_locale)); + Safefree(save_input_locale); return TRUE; } } @@ -1448,6 +1558,41 @@ Perl_my_strerror(pTHX_ const int errnum) { } /* + +=head1 Locale-related functions and macros + +=for apidoc sync_locale + +Changing the program's locale should be avoided by XS code. Nevertheless, +certain non-Perl libraries called from XS, such as C do so. When this +happens, Perl needs to be told that the locale has changed. Use this function +to do so, before returning to Perl. + +=cut +*/ + +void +Perl_sync_locale(pTHX) +{ + +#ifdef USE_LOCALE_CTYPE + new_ctype(setlocale(LC_CTYPE, NULL)); +#endif /* USE_LOCALE_CTYPE */ + +#ifdef USE_LOCALE_COLLATE + new_collate(setlocale(LC_COLLATE, NULL)); +#endif + +#ifdef USE_LOCALE_NUMERIC + set_numeric_local(); /* Switch from "C" to underlying LC_NUMERIC */ + new_numeric(setlocale(LC_NUMERIC, NULL)); +#endif /* USE_LOCALE_NUMERIC */ + +} + + + +/* * Local variables: * c-indentation-style: bsd * c-basic-offset: 4