sort untranslated_keys output again

[rails.git] / script / locale / diff

diff --git a/script/locale/diff b/script/locale/diff
index 1bb07d8189905ce00290a294376a32df67fa7b50..d0dce719a5974bed0b8fa2fb1ee0de56cb59184f 100755 (executable)
--- a/script/locale/diff
+++ b/script/locale/diff
@@ -1,8 +1,7 @@
 #!/usr/bin/env perl
 use strict;
 use warnings;
-use YAML::Syck qw(LoadFile);
-use Data::Walk;
+use YAML::Syck qw(Load LoadFile);
 use Test::Differences;
 use Pod::Usage ();
 use Getopt::Long ();
@@ -13,8 +12,17 @@ locale-diff - Compare two YAML files and print how their datastructures differ
 
 =head1 SYNOPSIS
 
-    locale-diff en.yml is.yml
-    locale-diff en.yml is.yml | grep '*'
+    # --keys is the default
+    diff en.yml is.yml
+    diff --keys en.yml is.yml
+
+    # --untranslated-values compares prints keys whose values don't differ
+    diff --untranslated-values-all en.yml is.yml
+
+    # --untranslated-values-all compares prints keys whose values
+    # don't differ. Ignoring the blacklist which prunes things
+    # unlikley to be translated
+    diff --untranslated-values-all en.yml is.yml
 
 =head1 DESCRIPTION
 
@@ -31,6 +39,24 @@ translated files when F<en.yml> changes.
 
 Print this help message.
 
+=item --keys
+
+Show the hash keys that differ between the two files, useful merging
+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.
+
+This helps to find untranslated values.
+
+=item --untranslated-values-all
+
+Like C<--untranslated-values> but ignores blacklists.
+
 =back
 
 =head1 AUTHOR
@@ -44,8 +70,14 @@ Getopt::Long::Parser->new(
     config => [ qw< bundling no_ignore_case no_require_order pass_through > ],
 )->getoptions(
     'h|help' => \my $help,
+    'keys' => \my $keys,
+    'untranslated-values' => \my $untranslated_values,
+    'untranslated-values-all' => \my $untranslated_values_all,
 ) or help();
 
+# --keys is the default
+$keys = 1 if not $untranslated_values_all and not $untranslated_values;
+
 # On --help
 help() if $help;
 
@@ -57,31 +89,95 @@ my ($from, $to) = @ARGV;
 my $from_data = LoadFile($from);
 my $to_data   = LoadFile($to);
 
-# Normalize the two to have the same root element
-my ($from_key) = keys %$from_data;
-$from_data = $from_data->{$from_key};
+my $from_parsed = { iterate($from_data->{basename($from)}) };
+my $to_parsed = { iterate($to_data->{basename($to)}) };
 
-my ($to_key) = keys %$to_data;
-$to_data = $to_data->{$to_key};
+# Since this used to be the default, support that...
+if ($keys)
+{
+    print_key_differences();
+}
+elsif ($untranslated_values or $untranslated_values_all)
+{
+    my @untranslated = untranslated_keys($from_parsed, $to_parsed);
 
-# Delete hash values
-walkdepth \&delete_hash_values, $_ for $from_data, $to_data;
+    # Prune according to blacklist
+    if ($untranslated_values) {
+        @untranslated = prune_untranslated_with_blacklist(basename($to), @untranslated);
+    }
 
-# Hack around Test::Differences wanting a Test::* module loaded
-$INC{"Test.pm"} = 1;
-sub Test::ok { print shift }
+    print $_, "\n" for @untranslated;
+}
 
-# Diff the tree
-eq_or_diff($from_data, $to_data);
+exit 0;
 
-sub delete_hash_values
+sub print_key_differences
 {
-    return unless defined $Data::Walk::type and $Data::Walk::type eq 'HASH';
+    # Hack around Test::Differences wanting a Test::* module loaded
+    $INC{"Test.pm"} = 1;
+    sub Test::ok { print shift }
 
-    # We totally need Perl 6's $OUTER::_ to make this prettier
-    my $hash = $_;
+    # Diff the tree
+    eq_or_diff([ sort keys %$from_parsed ], [ sort keys %$to_parsed ]);
+}
 
-    @$hash{grep { not ref $hash->{$_} } keys %$hash} = ();
+sub untranslated_keys
+{
+    my ($from_parsed, $to_parsed) = @_;
+    sort grep { not exists $to_parsed->{$_} or $from_parsed->{$_} eq $to_parsed->{$_} } keys %$from_parsed;
+}
+
+sub prune_untranslated_with_blacklist
+{
+    my ($language, @keys) = @_;
+    my %keys;
+    @keys{@keys} = ();
+
+    my $end_yaml = Load(join '', <DATA>);
+    my $untranslated_values = $end_yaml->{untranslated_values};
+    my $default = $untranslated_values->{default};
+    my $this_language = $untranslated_values->{$language} || {};
+
+    my %bw_list = (%$default, %$this_language);
+    
+    use feature ':5.10';
+    use Data::Dump 'dump';
+    say STDERR dump \%bw_list;
+
+    while (my ($key, $blacklisted) = each %bw_list)
+    {
+        # FIXME: Does syck actually support true/false booleans in yaml?
+        delete $keys{$key} if $blacklisted eq 'true'
+    }
+
+    sort keys %keys;
+}
+
+sub iterate
+{
+    my ($hash, @path) = @_;
+    my @ret;
+        
+    while (my ($k, $v) = each %$hash)
+    {
+        if (ref $v eq 'HASH')
+        {
+             push @ret => iterate($v, @path, $k);
+        }
+        else
+        {
+            push @ret => join(".",@path, $k), $v;
+        }
+    }
+
+    return @ret;
+}
+
+sub basename
+{
+    my $name = shift;
+    $name =~ s[\..*?$][];
+    $name;
 }
 
 sub help
@@ -93,3 +189,20 @@ sub help
         -exitval => $arg{ exitval } || 0,
     );
 }
+
+__DATA__
+untranslated_values:
+
+  # Default/Per language blacklist/whitelist for the
+  # --untranslated-values switch. "true" as a value indicates that the
+  # key is to be blacklisted, and "false" that it's to be
+  # whitelisted. "false" is only required to whitelist a key
+  # blacklisted by default on a per-language basis.
+
+  default:
+    html.dir: true
+    layouts.intro_3_bytemark: true
+    layouts.intro_3_ucl: true
+  de:
+    layouts.export: true
+