Added test checks and optional find-ls.gz file input

Author: mickeyn

bin/author.pl

@@ -51,10 +51,11 @@
 };
 
 # args
-my ( $pauseid, $whois_file );
+my ( $findls_file, $pauseid, $whois_file );
 GetOptions(
-    "pauseid=s"    => \$pauseid,
-    "whois_file=s" => \$whois_file,
+    "findls_file=s" => \$findls_file,
+    "pauseid=s"     => \$pauseid,
+    "whois_file=s"  => \$whois_file,
 );
 
 # setup
@@ -198,7 +199,7 @@ ($id)
     return undef
         unless $dir->is_dir;
 
-    my $cpan_file_map     = cpan_file_map();
+    my $cpan_file_map     = cpan_file_map( $findls_file );
     my $author_cpan_files = $cpan_file_map->{$id}
         or return undef;
 

bin/backpan.pl

@@ -9,15 +9,16 @@
 use MetaCPAN::Ingest qw< cpan_file_map >;
 
 # args
-my ( $distribution, $files_only, $undo );
+my ( $distribution, $files_only, $findls_file, $undo );
 GetOptions(
     "distribution=s" => \$distribution,
     "files_only"     => \$files_only,
+    "findls_file=s"  => \$findls_file,
     "undo"           => \$undo,
 );
 
 # setup
-my $cpan_file_map = cpan_file_map();
+my $cpan_file_map = cpan_file_map( $findls_file );
 my $es_release    = MetaCPAN::ES->new( index => "release" );
 my $es_file       = MetaCPAN::ES->new( index => "file" );
 

lib/MetaCPAN/ES.pm

@@ -37,6 +37,14 @@ sub new ( $class, %args ) {
     }, $class;
 }
 
+sub test ( $self ) {
+    return !!(
+        ref($self) eq __PACKAGE__
+        and ref($self->{es})
+        and ref($self->{es}) =~ /^Search::Elasticsearch/
+    );
+}
+
 sub index ( $self, %args ) {
     $self->{es}->index(
         index => $self->{index},

lib/MetaCPAN/Ingest.pm

@@ -248,13 +248,13 @@ sub read_url ( $url ) {
     return $resp->decoded_content;
 }
 
-sub cpan_file_map () {
-    my $cpan = cpan_dir();
-
-    my $ls   = $cpan->child(qw< indices find-ls.gz >);
-    if ( !-e $ls ) {
-        warn "File $ls does not exist";
-        return {};
+sub cpan_file_map ( $ls = undef ) {
+    if (!$ls) {
+        my $cpan = cpan_dir();
+        $ls = $cpan->child(qw< indices find-ls.gz >);
+        if ( !-e $ls ) {
+            die "File $ls does not exist";
+        }
     }
 
     log_info {"Reading $ls"};

lib/MetaCPAN/Mapper.pm

@@ -35,6 +35,14 @@ sub new ( $class, %args ) {
     }, $class;
 }
 
+sub test ( $self ) {
+    return !!(
+        ref($self) eq __PACKAGE__
+        and ref($self->{es})
+        and ref($self->{es}) =~ /^Search::Elasticsearch/
+    );
+}
+
 sub index_exists ( $self, $index ) {
     $self->{es}->indices->exists( index => $index );
 }

t/00_setup.t

@@ -8,29 +8,28 @@ use MetaCPAN::ES;
 use MetaCPAN::Mapper;
 use MetaCPAN::Ingest qw< home >;
 
-my $es     = MetaCPAN::ES->new();
+my $es = MetaCPAN::ES->new();
+ok( $es->test, 'es is valid' );
+
 my $mapper = MetaCPAN::Mapper->new();
+ok( $mapper->test, 'mapper is valid' );
 
 my $home       = home();
 my $d_bin      = $home->child('bin');
 my $d_test     = $home->child('test_data');
 my $d_fakecpan = $d_test->child('fakecpan');
 my $d_authors  = $d_fakecpan->child('authors');
 my $d_modules  = $d_fakecpan->child('modules');
+my $d_indices  = $d_fakecpan->child('indices');
 
 # === Files
 
-my @files = qw<
-    00whois.xml
-    02packages.details.txt.gz
-    06perms.txt
->;
-
 subtest 'Check Files' => sub {
     ok( $d_authors->child('00whois.xml'), "Found 00whois.xml" );
     ok( $d_modules->child('02packages.details.txt.gz'),
         "Found 02packages.details.txt.gz" );
     ok( $d_modules->child('06perms.txt'), "Found 06perms.txt" );
+    ok( $d_indices->child('find-ls.gz'), "Found find-ls.gz" );
 };
 
 my @packages = qw<
@@ -73,9 +72,10 @@ subtest 'Check Index' => sub {
 subtest 'Author Indexing' => sub {
     my $author_script = $d_bin->child('author.pl');
     my $whois_file    = $d_authors->child('00whois.xml');
+    my $findls_file   = $d_indices->child('find-ls.gz');
 
     # run the author indexing script
-    `perl $author_script -whois_file $whois_file`;
+    `perl $author_script -whois_file $whois_file -findls_file $findls_file`;
 
     my $es_author = MetaCPAN::ES->new( index => 'author' );
     ok( $es_author->exists( index => 'author', id => 'OALDERS' ),