Add script to update config/languages.yml from translatewiki language list.

[rails.git] / script / locale / diff

diff --git a/script/locale/diff b/script/locale/diff
index 8ce35ad5c64bc90c9b4a34965aed1f4fb18b03a9..bfc63694d72efc1eca3ba7341bbf13496578936b 100755 (executable)
--- a/script/locale/diff
+++ b/script/locale/diff
@@ -2,7 +2,7 @@
 use feature ':5.10';
 use strict;
 use warnings;
-use YAML::Syck qw(Load LoadFile);
+use YAML::Syck qw(Dump LoadFile);
 use Test::Differences;
 use Pod::Usage ();
 use Getopt::Long ();
@@ -18,7 +18,7 @@ locale-diff - Compare two YAML files and print how their datastructures differ
     diff --keys en.yml is.yml
 
     # --untranslated-values compares prints keys whose values don't differ
-    diff --untranslated-values-all en.yml is.yml
+    diff --untranslated-values en.yml is.yml
 
     # --untranslated-values-all compares prints keys whose values
     # don't differ. Ignoring the blacklist which prunes things
@@ -50,10 +50,12 @@ new entries from F<en.yml> to a local file.
 
 =item --untranslated-values
 
-Show keys whose values are either exactly the same between the two
-files, or don't exist in the target file (the latter file
-specified). The values are pruned according to global and language
-specific blacklists found in the C<__DATA__> section of this script.
+Show keys that B<exist in both the compared files> and whose values
+are exactly the same. Use C<--keys> to a list of values that hasn't
+been merged.
+
+The values are pruned according to global and language specific
+blacklists found in the C<__DATA__> section of this script.
 
 This helps to find untranslated values.
 
@@ -66,11 +68,16 @@ Like C<--untranslated-values> but ignores blacklists.
 Check that interpolated Ruby i18n variables (C<{{foo}}> and
 C<[[foo]]>) are equivalent in the two provided files.
 
+=item --dump-flat
+
+Dump a flat version of the translation hash in YAML format,
+i.e. "foo.bar" instead of "{foo}->{bar}".
+
 =back
 
 =head1 AUTHOR
 
-E<AElig>var ArnfjE<ouml>rE<eth> Bjarmason <avar@f-prot.com>
+E<AElig>var ArnfjE<ouml>rE<eth> Bjarmason <avarab@gmail.com>
 
 =cut
 
@@ -80,26 +87,50 @@ Getopt::Long::Parser->new(
 )->getoptions(
     'h|help' => \my $help,
     'keys' => \my $keys,
+    'dump-flat' => \my $dump_flat,
     'untranslated-values' => \my $untranslated_values,
     'untranslated-values-all' => \my $untranslated_values_all,
     'validate-variables' => \my $validate_variables,
+    'reconstruct' => \my $reconstruct,
 ) or help();
 
 # --keys is the default
-$keys = 1 if not $untranslated_values_all and not $untranslated_values and not $validate_variables;
+$keys = 1 if not $untranslated_values_all and not $untranslated_values and not $validate_variables and not $dump_flat;
 
 # On --help
 help() if $help;
 
 # If we're not given two .yml files
-help() if @ARGV != 2 or (!-f $ARGV[0] or !-f $ARGV[1]);
+help() if (@ARGV != 2 or (!-f $ARGV[0] or !-f $ARGV[1])) and not $dump_flat || $reconstruct;
 
 my ($from, $to) = @ARGV;
 
 my $from_data = LoadFile($from);
-my $to_data   = LoadFile($to);
-
 my $from_parsed = { iterate($from_data->{basename($from)}) };
+
+if ($dump_flat)
+{
+    mark_utf8($from_parsed);
+
+    print Dump $from_parsed;
+
+    exit 0;
+}
+
+if ($reconstruct) {
+    mark_utf8($from_parsed);
+
+    my %out;
+    while (my ($k, $v) = each %$from_parsed) {
+        insert_string_deep(\%out, $k, $v);
+    }
+
+    print Dump { basename($from) => \%out };
+
+    exit 0;
+}
+
+my $to_data   = LoadFile($to);
 my $to_parsed = { iterate($to_data->{basename($to)}) };
 
 if ($keys)
@@ -138,7 +169,7 @@ sub print_key_differences
 sub untranslated_keys
 {
     my ($from_parsed, $to_parsed) = @_;
-    sort grep { not exists $to_parsed->{$_} or $from_parsed->{$_} eq $to_parsed->{$_} } keys %$from_parsed;
+    sort grep { exists $to_parsed->{$_} and $from_parsed->{$_} eq $to_parsed->{$_} } keys %$from_parsed;
 }
 
 sub prune_untranslated_with_blacklist
@@ -147,7 +178,7 @@ sub prune_untranslated_with_blacklist
     my %keys;
     @keys{@keys} = ();
 
-    my $end_yaml = Load(join '', <DATA>);
+    my $end_yaml = LoadFile(*DATA);
     my $untranslated_values = $end_yaml->{untranslated_values};
     my $default = $untranslated_values->{default};
     my $this_language = $untranslated_values->{$language} || {};
@@ -215,13 +246,52 @@ sub iterate
     return @ret;
 }
 
+# $s = 'foo.bar.baz.spam.eggs.ham'; $h = \%h; $h = $h->{$_} = {} for split /\./, $s; \%h
+# ==> {foo => {bar => {baz => {spam => {eggs => {ham => {}}}}}}}
+sub insert_string_deep {
+    my ($h, $ks, $v) = @_;
+    my $p = \$h; $p = \$$p->{$_} for split /\./, $ks;
+    $$p = $v;
+}
+
+# sub insert_string_deep
+# {
+#     my ($hash, $key, $value) = @_;
+#
+#     my @key = split /\./, $key;
+#     my $h = $hash;
+#
+#     my $i = 0;
+#     for my $k (@key) {
+#         $i ++;
+#         if ($i == @key) {
+#             $h->{$k} = $value;
+#         } else {
+#             if (ref $h->{$k}) {
+#                 $h = $h->{$k};
+#             } else {
+#                 $h = $h->{$k} = {};
+#             }
+#         }
+#     }
+# }
+
 sub basename
 {
     my $name = shift;
     $name =~ s[\..*?$][];
+    $name =~ s[.*/][];
     $name;
 }
 
+sub mark_utf8
+{
+    my ($hash) = @_;
+
+    # Mark as UTF-8
+    map { if (ref $_ eq 'ARRAY') { map { utf8::decode($_) } @$_ } else {  utf8::decode($_) } } values %$hash;
+}
+
 sub help
 {
     my %arg = @_;
@@ -248,6 +318,11 @@ untranslated_values:
     layouts.project_name.h1: true
     layouts.project_name.title: true
     site.index.license.project_url: true
+    browse.relation_member.entry: true
+
+    # #{{id}}
+    changeset.changeset.id: true
+
   de:
     activerecord.attributes.message.sender: true
     activerecord.attributes.trace.name: true
@@ -284,3 +359,24 @@ untranslated_values:
     trace.view.tags: true
     user.account.public editing.enabled link: true
 
+  is:
+    # ({{link}})
+    site.edit.anon_edits: true
+
+    # Creative Commons Attribution-Share Alike 2.0
+    site.index.license.license_name: true
+
+    # http://creativecommons.org/licenses/by-sa/2.0/
+    site.index.license.license_url: true
+
+    # {{id}}
+    printable_name.with_id: true
+    
+    # {{name}} ({{id}})
+    printable_name.with_name: true
+
+    # {{type}} 
+    geocoder.search_osm_namefinder.prefix: true
+
+    # {{suffix}}, {{parentname}}
+    geocoder.search_osm_namefinder.suffix_suburb: true