Support blacklists/whitelists for --untranslated-values output

diff --git a/script/locale/diff b/script/locale/diff
index 2afffb828a645da512cf8dfb963c8080af0a7183..7ddfb3eb3f5ca99692462703f4abbf02a11cdc28 100755 (executable)
--- a/script/locale/diff
+++ b/script/locale/diff
@@ -1,7 +1,7 @@
 #!/usr/bin/env perl
 use strict;
 use warnings;
-use YAML::Syck qw(LoadFile);
+use YAML::Syck qw(Load LoadFile);
 use Test::Differences;
 use Pod::Usage ();
 use Getopt::Long ();
@@ -12,11 +12,17 @@ locale-diff - Compare two YAML files and print how their datastructures differ
 
 =head1 SYNOPSIS
 
-    diff en.yml is.yml
     # --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 en.yml is.yml
+    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
 
@@ -41,10 +47,16 @@ 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).
+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
@@ -60,8 +72,12 @@ Getopt::Long::Parser->new(
     '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;
 
@@ -77,14 +93,19 @@ my $from_parsed = { iterate($from_data->{basename($from)}) };
 my $to_parsed = { iterate($to_data->{basename($to)}) };
 
 # Since this used to be the default, support that...
-if ((not $untranslated_values and not $keys) or $keys)
+if ($keys)
 {
     print_key_differences();
 }
-elsif ($untranslated_values)
+elsif ($untranslated_values or $untranslated_values_all)
 {
     my @untranslated = untranslated_keys($from_parsed, $to_parsed);
 
+    # Prune according to blacklist
+    if ($untranslated_values) {
+        @untranslated = prune_untranslated_with_blacklist(basename($to), @untranslated);
+    }
+
     print $_, "\n" for @untranslated;
 }
 
@@ -103,7 +124,33 @@ 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;
+    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
@@ -142,3 +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
+