add files_by_category end point

The existing interesting_files end point had a long list of files to
include, with inconsistent rules for what file extensions and
capitalization would be allowed.  It also would just give the files back
in a single list.  The code for finding change logs partly reproduced
this.

Unify the logic for the files we search for, with consistent rules for
file extensions, capitalization, and handling of special dists like
"perl".  Keep the different types of files categorized so that
interesting_files can both filter the files it is querying for, and
adding a new end point files_by_category to easily find the files of a
specific type, like was previously possible for change logs.  Install
files and contributing docs are now listed on the front end, so this
should make it easier for them to find the files.

Author: haarg

lib/MetaCPAN/Document/File/Set.pm

@@ -17,7 +17,13 @@ has query_file => (
     isa     => 'MetaCPAN::Query::File',
     lazy    => 1,
     builder => '_build_query_file',
-    handles => [qw< dir interesting_files >],
+    handles => [
+        qw(
+            dir
+            interesting_files
+            files_by_category
+            )
+    ],
 );
 
 sub _build_query_file {
@@ -654,75 +660,10 @@ sub autocomplete_suggester {
 
 sub find_changes_files {
     my ( $self, $author, $release ) = @_;
-
-    # find the most likely file
-    # TODO: should we do this when the release is indexed
-    # and store the result as { 'changes_file' => $name }
-
-    my @candidates = qw(
-        CHANGELOG
-        ChangeLog
-        Changelog
-        ChangeLog.pm
-        changelog.pm
-        ChangeLog.pod
-        CHANGES
-        Changes
-        CHANGES.md
-        CHANGES.markdown
-        CHANGES.pm
-        Changes.pm
-        CHANGES.pod
-        Changes.pod
-        NEWS
-    );
-
-    # use $c->model b/c we can't let any filters apply here
-    my $file = $self->raw->filter(
-        {
-            and => [
-                { term => { release => $release } },
-                { term => { author  => $author } },
-                {
-                    or => [
-
-                        # if it's a perl release, get perldelta
-                        {
-                            and => [
-                                { term => { distribution => 'perl' } },
-                                {
-                                    term => {
-                                        'name' => 'perldelta.pod'
-                                    }
-                                },
-                            ]
-                        },
-
-                      # otherwise look for one of these candidates in the root
-                        {
-                            and => [
-                                { term => { level     => 0 } },
-                                { term => { directory => 0 } },
-                                {
-                                    or => [
-                                        map { { term => { 'name' => $_ } } }
-                                            @candidates
-                                    ]
-                                }
-                            ]
-                        }
-                    ],
-                }
-            ]
-        }
-    )->size(1)
-
-        # HACK: Sort by level/desc to put pod/perldeta.pod first (if found)
-        # otherwise sort root files by name and select the first.
-        ->sort( [ { level => 'desc' }, { name => 'asc' } ] )->first;
-
-    return unless is_hashref($file);
-    return $file->{_source};
+    my $result = $self->files_by_category( $author, $release, ['changelog'],
+        { fields => \1 } );
+    my ($file) = @{ $result->{categories}{changelog} || [] };
+    return $file;
 }
 
 __PACKAGE__->meta->make_immutable;