This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
t/loc_tools.pl: Nits in comments, white-space only
[perl5.git] / t / loc_tools.pl
CommitLineData
73fc293b 1# Common tools for test files files to find the locales which exist on the
abfe20b2
KW
2# system. Caller should have defined ok() for the unlikely event that setup
3# here fails, and should have verified that this isn't miniperl before calling
4# the functions.
73fc293b
KW
5
6# Note that it's okay that some languages have their native names
7# capitalized here even though that's not "right". They are lowercased
8# anyway later during the scanning process (and besides, some clueless
9# vendor might have them capitalized erroneously anyway).
10
6475ad77
KW
11# Some of the locales on the system may not play well with Perl. Since, we
12# may be trying every possible locale, we don't want to be warned about the
13# weird ones.
14no warnings 'locale';
15
73fc293b 16sub _trylocale { # Adds the locale given by the first parameter to the list
9b0711ee
KW
17 # given by the 3rd iff the platform supports the locale in
18 # each of the categories given by the 2nd parameter, which
19 # is either a single category or a reference to a list of
20 # categories
73fc293b 21 my $locale = shift;
9b0711ee 22 my $categories = shift;
73fc293b 23 my $list = shift;
162eb2b8 24 return if ! $locale || grep { $locale eq $_ } @$list;
9b0711ee
KW
25
26 $categories = [ $categories ] unless ref $categories;
27
28 foreach my $category (@$categories) {
29 return unless setlocale($category, $locale);
30 }
31
73fc293b
KW
32 my $badutf8;
33 {
34 local $SIG{__WARN__} = sub {
35 $badutf8 = $_[0] =~ /Malformed UTF-8/;
36 };
37 }
38
39 if ($badutf8) {
40 ok(0, "Verify locale name doesn't contain malformed utf8");
41 return;
42 }
43 push @$list, $locale;
44}
45
46sub _decode_encodings {
47 my @enc;
48
49 foreach (split(/ /, shift)) {
50 if (/^(\d+)$/) {
51 push @enc, "ISO8859-$1";
52 push @enc, "iso8859$1"; # HP
53 if ($1 eq '1') {
54 push @enc, "roman8"; # HP
55 }
56 push @enc, $_;
57 push @enc, "$_.UTF-8";
d646bffe
KW
58 push @enc, "$_.65001"; # Windows UTF-8
59 push @enc, "$_.ACP"; # Windows ANSI code page
60 push @enc, "$_.OCP"; # Windows OEM code page
73fc293b
KW
61 }
62 }
63 if ($^O eq 'os390') {
64 push @enc, qw(IBM-037 IBM-819 IBM-1047);
65 }
66 push @enc, "UTF-8";
d646bffe 67 push @enc, "65001"; # Windows UTF-8
73fc293b
KW
68
69 return @enc;
70}
71
9b0711ee
KW
72sub find_locales ($) { # Returns an array of all the locales we found on the
73 # system. The parameter is either a single locale
74 # category or a reference to a list of categories to
b94d11d6 75 # find valid locales for it (or in the case of
c1d1e689
KW
76 # multiple) for all of them. Note that currently the
77 # array includes even those locales that don't play
78 # well with Perl
9b0711ee 79 my $categories = shift;
73fc293b 80
abfe20b2
KW
81 use Config;;
82 my $have_setlocale = $Config{d_setlocale};
d369fd5b
KW
83
84 # Visual C's CRT goes silly on strings of the form "en_US.ISO8859-1"
85 # and mingw32 uses said silly CRT
86 # This doesn't seem to be an issue any more, at least on Windows XP,
87 # so re-enable the tests for Windows XP onwards.
88 my $winxp = ($^O eq 'MSWin32' && defined &Win32::GetOSVersion &&
89 join('.', (Win32::GetOSVersion())[1..2]) >= 5.1);
90 $have_setlocale = 0 if ((($^O eq 'MSWin32' && !$winxp) || $^O eq 'NetWare') &&
91 $Config{cc} =~ /^(cl|gcc|g\+\+|ici)/i);
92
93 # UWIN seems to loop after taint tests, just skip for now
94 $have_setlocale = 0 if ($^O =~ /^uwin/);
95
96 return unless $have_setlocale;
97
abfe20b2
KW
98 # Done this way in case this is 'required' in the caller before seeing if
99 # this is miniperl.
395e4985
JH
100 eval { require POSIX; import POSIX 'locale_h'; };
101 unless (defined &POSIX::LC_CTYPE) {
102 return;
103 }
abfe20b2 104
9b0711ee
KW
105 _trylocale("C", $categories, \@Locale);
106 _trylocale("POSIX", $categories, \@Locale);
73fc293b 107 foreach (0..15) {
9b0711ee
KW
108 _trylocale("ISO8859-$_", $categories, \@Locale);
109 _trylocale("iso8859$_", $categories, \@Locale);
110 _trylocale("iso8859-$_", $categories, \@Locale);
111 _trylocale("iso_8859_$_", $categories, \@Locale);
112 _trylocale("isolatin$_", $categories, \@Locale);
113 _trylocale("isolatin-$_", $categories, \@Locale);
114 _trylocale("iso_latin_$_", $categories, \@Locale);
73fc293b
KW
115 }
116
117 # Sanitize the environment so that we can run the external 'locale'
118 # program without the taint mode getting grumpy.
119
120 # $ENV{PATH} is special in VMS.
8b76dc1d 121 delete local $ENV{PATH} if $^O ne 'VMS' or $Config{d_setenv};
73fc293b
KW
122
123 # Other subversive stuff.
8b76dc1d 124 delete local @ENV{qw(IFS CDPATH ENV BASH_ENV)};
73fc293b
KW
125
126 if (-x "/usr/bin/locale"
127 && open(LOCALES, "/usr/bin/locale -a 2>/dev/null|"))
128 {
129 while (<LOCALES>) {
130 # It seems that /usr/bin/locale steadfastly outputs 8 bit data, which
131 # ain't great when we're running this testPERL_UNICODE= so that utf8
132 # locales will cause all IO hadles to default to (assume) utf8
133 next unless utf8::valid($_);
134 chomp;
9b0711ee 135 _trylocale($_, $categories, \@Locale);
73fc293b
KW
136 }
137 close(LOCALES);
138 } elsif ($^O eq 'VMS'
139 && defined($ENV{'SYS$I18N_LOCALE'})
140 && -d 'SYS$I18N_LOCALE')
141 {
142 # The SYS$I18N_LOCALE logical name search list was not present on
143 # VAX VMS V5.5-12, but was on AXP && VAX VMS V6.2 as well as later versions.
144 opendir(LOCALES, "SYS\$I18N_LOCALE:");
145 while ($_ = readdir(LOCALES)) {
146 chomp;
9b0711ee 147 _trylocale($_, $categories, \@Locale);
73fc293b
KW
148 }
149 close(LOCALES);
150 } elsif (($^O eq 'openbsd' || $^O eq 'bitrig' ) && -e '/usr/share/locale') {
151
152 # OpenBSD doesn't have a locale executable, so reading /usr/share/locale
153 # is much easier and faster than the last resort method.
154
155 opendir(LOCALES, '/usr/share/locale');
156 while ($_ = readdir(LOCALES)) {
157 chomp;
9b0711ee 158 _trylocale($_, $categories, \@Locale);
73fc293b
KW
159 }
160 close(LOCALES);
161 } else { # Final fallback. Try our list of locales hard-coded here
162
163 # This is going to be slow.
164 my @Data;
165
73fc293b
KW
166 # Locales whose name differs if the utf8 bit is on are stored in these two
167 # files with appropriate encodings.
168 if ($^H & 0x08 || (${^OPEN} || "") =~ /:utf8/) {
169 @Data = do "lib/locale/utf8";
170 } else {
171 @Data = do "lib/locale/latin1";
172 }
173
174 # The rest of the locales are in this file.
175 push @Data, <DATA>;
176
177 foreach my $line (@Data) {
178 my ($locale_name, $language_codes, $country_codes, $encodings) =
179 split /:/, $line;
180 my @enc = _decode_encodings($encodings);
181 foreach my $loc (split(/ /, $locale_name)) {
9b0711ee 182 _trylocale($loc, $categories, \@Locale);
73fc293b 183 foreach my $enc (@enc) {
9b0711ee 184 _trylocale("$loc.$enc", $categories, \@Locale);
73fc293b
KW
185 }
186 $loc = lc $loc;
187 foreach my $enc (@enc) {
9b0711ee 188 _trylocale("$loc.$enc", $categories, \@Locale);
73fc293b
KW
189 }
190 }
191 foreach my $lang (split(/ /, $language_codes)) {
9b0711ee 192 _trylocale($lang, $categories, \@Locale);
73fc293b
KW
193 foreach my $country (split(/ /, $country_codes)) {
194 my $lc = "${lang}_${country}";
9b0711ee 195 _trylocale($lc, $categories, \@Locale);
73fc293b 196 foreach my $enc (@enc) {
9b0711ee 197 _trylocale("$lc.$enc", $categories, \@Locale);
73fc293b
KW
198 }
199 my $lC = "${lang}_\U${country}";
9b0711ee 200 _trylocale($lC, $categories, \@Locale);
73fc293b 201 foreach my $enc (@enc) {
9b0711ee 202 _trylocale("$lC.$enc", $categories, \@Locale);
73fc293b
KW
203 }
204 }
205 }
206 }
207 }
208
209 @Locale = sort @Locale;
210
211 return @Locale;
73fc293b
KW
212}
213
ab8b8bcc 214sub is_locale_utf8 ($) { # Return a boolean as to if core Perl thinks the input
9b0711ee 215 # is a UTF-8 locale
92c0a900
KW
216
217 # On z/OS, even locales marked as UTF-8 aren't.
218 return 0 if ord "A" != 65;
219
ab8b8bcc
KW
220 my $locale = shift;
221
31f05a37
KW
222 use locale;
223
224 my $save_locale = setlocale(&POSIX::LC_CTYPE());
225 if (! $save_locale) {
226 ok(0, "Verify could save previous locale");
227 return 0;
228 }
229
230 if (! setlocale(&POSIX::LC_CTYPE(), $locale)) {
231 ok(0, "Verify could setlocale to $locale");
232 return 0;
233 }
234
235 my $ret = 0;
236
237 # Use an op that gives different results for UTF-8 than any other locale.
238 # If a platform has UTF-8 locales, there should be at least one locale on
239 # most platforms with UTF-8 in its name, so if there is a bug in the op
240 # giving a false negative, we should get a failure for those locales as we
241 # go through testing all the locales on the platform.
242 if (CORE::fc(chr utf8::unicode_to_native(0xdf)) ne "ss") {
243 if ($locale =~ /UTF-?8/i) {
244 ok (0, "Verify $locale with UTF-8 in name is a UTF-8 locale");
245 }
246 }
247 else {
248 $ret = 1;
249 }
250
251 die "Couldn't restore locale '$save_locale'"
252 unless setlocale(&POSIX::LC_CTYPE(), $save_locale);
253
254 return $ret;
ab8b8bcc
KW
255}
256
c1d1e689 257sub find_utf8_ctype_locale (;$) { # Return the name of a locale that core Perl
9b0711ee
KW
258 # thinks is a UTF-8 LC_CTYPE locale.
259 # Optional parameter is a reference to a
260 # list of locales to try; if omitted, this
261 # tries all locales it can find on the
262 # platform
ab8b8bcc 263 my $locales_ref = shift;
46f4bdef 264 return if !defined &POSIX::LC_CTYPE;
ab8b8bcc 265 if (! defined $locales_ref) {
9b0711ee 266 my @locales = find_locales(&POSIX::LC_CTYPE());
ab8b8bcc
KW
267 $locales_ref = \@locales;
268 }
269
270 foreach my $locale (@$locales_ref) {
271 return $locale if is_locale_utf8($locale);
272 }
273
274 return;
275}
276
73fc293b
KW
2771
278
279# Format of data is: locale_name, language_codes, country_codes, encodings
280__DATA__
281Afrikaans:af:za:1 15
282Arabic:ar:dz eg sa:6 arabic8
283Brezhoneg Breton:br:fr:1 15
284Bulgarski Bulgarian:bg:bg:5
285Chinese:zh:cn tw:cn.EUC eucCN eucTW euc.CN euc.TW Big5 GB2312 tw.EUC
286Hrvatski Croatian:hr:hr:2
287Cymraeg Welsh:cy:cy:1 14 15
288Czech:cs:cz:2
289Dansk Danish:da:dk:1 15
290Nederlands Dutch:nl:be nl:1 15
291English American British:en:au ca gb ie nz us uk zw:1 15 cp850
292Esperanto:eo:eo:3
293Eesti Estonian:et:ee:4 6 13
294Suomi Finnish:fi:fi:1 15
295Flamish::fl:1 15
296Deutsch German:de:at be ch de lu:1 15
297Euskaraz Basque:eu:es fr:1 15
298Galego Galician:gl:es:1 15
299Ellada Greek:el:gr:7 g8
300Frysk:fy:nl:1 15
301Greenlandic:kl:gl:4 6
302Hebrew:iw:il:8 hebrew8
303Hungarian:hu:hu:2
304Indonesian:id:id:1 15
305Gaeilge Irish:ga:IE:1 14 15
306Italiano Italian:it:ch it:1 15
307Nihongo Japanese:ja:jp:euc eucJP jp.EUC sjis
308Korean:ko:kr:
309Latine Latin:la:va:1 15
310Latvian:lv:lv:4 6 13
311Lithuanian:lt:lt:4 6 13
312Macedonian:mk:mk:1 15
313Maltese:mt:mt:3
314Moldovan:mo:mo:2
315Norsk Norwegian:no no\@nynorsk nb nn:no:1 15
316Occitan:oc:es:1 15
317Polski Polish:pl:pl:2
318Rumanian:ro:ro:2
319Russki Russian:ru:ru su ua:5 koi8 koi8r KOI8-R koi8u cp1251 cp866
320Serbski Serbian:sr:yu:5
321Slovak:sk:sk:2
322Slovene Slovenian:sl:si:2
323Sqhip Albanian:sq:sq:1 15
324Svenska Swedish:sv:fi se:1 15
325Thai:th:th:11 tis620
326Turkish:tr:tr:9 turkish8
327Yiddish:yi::1 15