return grep {length $_} map {split $splitter} @t;
};
-my $ditch_empty_space = sub {return &{$ditch_empty}(' ',@_)};
+our $sort_and_unique = sub {
+ my @v;
+ my %u;
+ my $all_numeric = 1;
+ for my $v (@_) {
+ if ($all_numeric and $v =~ /\D/) {
+ $all_numeric = 0;
+ }
+ next if exists $u{$v};
+ $u{$v} = 1;
+ push @v, $v;
+ }
+ if ($all_numeric) {
+ return sort {$a <=> $b} @v;
+ } else {
+ return sort @v;
+ }
+};
+
+my $ditch_space_unique_and_sort = sub {return &{$sort_and_unique}(&{$ditch_empty}(' ',@_))};
my %split_fields =
(package => \&splitpackages,
affects => \&splitpackages,
- blocks => $ditch_empty_space,
- blockedby => $ditch_empty_space,
+ blocks => $ditch_space_unique_and_sort,
+ blockedby => $ditch_space_unique_and_sort,
# this isn't strictly correct, but we'll split both of them for
# the time being until we ditch all use of keywords everywhere
# from the code
- keywords => $ditch_empty_space,
- tags => $ditch_empty_space,
- found_versions => $ditch_empty_space,
- fixed_versions => $ditch_empty_space,
- mergedwith => $ditch_empty_space,
+ keywords => $ditch_space_unique_and_sort,
+ tags => $ditch_space_unique_and_sort,
+ found_versions => $ditch_space_unique_and_sort,
+ fixed_versions => $ditch_space_unique_and_sort,
+ mergedwith => $ditch_space_unique_and_sort,
);
sub split_status_fields {