parse_result.pl code move to DatacheckTapToJson module

Ensembl · Jun 14, 2020 · 7915339 · 7915339
1 parent 494caee
commit 7915339
Show file tree

Hide file tree

Showing 3 changed files with 60 additions and 86 deletions.
diff --git a/lib/Bio/EnsEMBL/DataCheck/Pipeline/DataCheckTapToJson.pm b/lib/Bio/EnsEMBL/DataCheck/Pipeline/DataCheckTapToJson.pm
@@ -34,40 +34,54 @@ use TAP::Parser;
 
 
 use base ('Bio::EnsEMBL::Hive::Process');
+
 sub run {
 
   my $self = shift;
 
   my $output_dir   = $self->param('output_dir');
-  my @tap_files = map { $_->stringify } path($output_dir)->children;
-
+  my $to_json   = 1;
+
+  parse_datachecks($output_dir, $output_dir, 1, 1, $to_json);
+}
+
+sub parse_datachecks {
+
+  my ($tap, $output_file, $by_species, $passed, $to_json) = @_;
+
+  my @tap_files;
+  if (-d $tap) {
+    @tap_files = map { $_->stringify } path($tap)->children;
+  } else {
+    push @tap_files, $tap;
+  }
+
   my %results;
   my $datacheck;
   my $species;
   my $test;
   my %tests;
-  my $passed = 1;
-  my $by_species = 1;
 
   foreach my $tap_file (@tap_files) {
     my $tap = path($tap_file)->slurp;
     my $parser = TAP::Parser->new( { tap => $tap } );
+
     while (my $result = $parser->next) {
       if ($result->is_comment) {
-         if ($result->as_string =~ /^# Subtest: (.+)/) {
-           $datacheck = $1;
-         }
+        if ($result->as_string =~ /^# Subtest: (.+)/) {
+          $datacheck = $1;
+        }
       } elsif ($result->is_unknown) {
-       if ($result->as_string =~ /^\s+# Subtest: (.+)/) {
+        if ($result->as_string =~ /^\s+# Subtest: (.+)/) {
           $species = $1;
           %tests = ();
-       } elsif ($result->as_string =~ /^\s{8}((?:not ok|# No tests run).*)/) {
+        } elsif ($result->as_string =~ /^\s{8}((?:not ok|# No tests run).*)/) {
           $test = $1;
           $tests{$test} = [];
-       } elsif ($result->as_string =~ /^\s{8}((?:ok|.* # SKIP).*)/ && $passed) {
+        } elsif ($result->as_string =~ /^\s{8}((?:ok|.* # SKIP).*)/ && $passed) {
           $test = $1;
           $tests{$test} = [];
-       } elsif ($result->as_string =~ /^\s{8}#\s(\s*.*)/) {
+        } elsif ($result->as_string =~ /^\s{8}#\s(\s*.*)/) {
           if (defined $test) {
             push @{$tests{$test}}, $1;
           } else {
@@ -93,12 +107,31 @@ sub run {
   }
 
   my $json = JSON->new->canonical->pretty->encode(\%results);
-  my $basename = path($output_dir)->basename;
-  my $output_file = $output_dir . '/' . $basename . '.json';
-  path($output_file)->parent->mkpath;
-  path($output_file)->spew($json)
+
+  if ($output_file) {
+    if($to_json){  
+      my $basename = path($output_file)->basename;
+      $output_file = $output_file . '/' . $basename . '.json';
+    } 
+      path($output_file)->parent->mkpath;
+      path($output_file)->spew($json)
+
+  } else {
+    print($json);
+  }
+
+
+
+
+
+
+
+
+
+
 
 }
 
+
 1;
 
diff --git a/lib/Bio/EnsEMBL/DataCheck/Pipeline/DbDataChecks_conf.pm b/lib/Bio/EnsEMBL/DataCheck/Pipeline/DbDataChecks_conf.pm
@@ -71,6 +71,7 @@ sub default_options {
     email         => undef,
     report_per_db => 0,
     report_all    => 0,
+
   };
 }
 
@@ -301,14 +302,17 @@ sub pipeline_analyses {
       -max_retry_count   => 0,
       -rc_name           => 'default',
       -flow_into         => {
-                              '1' => ['?table_name=result', 'DataCheckTapToJson'],
+                              '1' =>  ['?table_name=result', 'DataCheckTapToJson']
                             },
     },
     {
       -logic_name        => 'DataCheckTapToJson',
       -module            => 'Bio::EnsEMBL::DataCheck::Pipeline::DataCheckTapToJson',
       -analysis_capacity => 10,
       -max_retry_count   => 0,
+      -parameters        => {
+                                output_dir => $self->o('output_dir'),
+                            },
       -rc_name           => 'default',
     },
 

diff --git a/scripts/parse_results.pl b/scripts/parse_results.pl
@@ -60,90 +60,27 @@ =head1 OPTIONS
 use Path::Tiny;
 use Pod::Usage;
 use TAP::Parser;
+use Bio::EnsEMBL::DataCheck::Pipeline::DataCheckTapToJson;
 
-my ($help, $tap, $output_file, $by_species, $passed);
+my ($help, $tap); 
+my $output_file = '';
+my $by_species = 0;
+my $passed = 0;
 
 GetOptions(
   "help!",         \$help,
   "tap:s",         \$tap,
   "output_file:s", \$output_file,
   "by_species!",   \$by_species,
-  "passed!",       \$passed,
+  "passed!",       \$passed
 );
 
 pod2usage(1) if $help;
-
 if (! defined $tap) {
   die "Need a source of TAP data";
 } elsif (! -e $tap) {
   die "TAP source does not exist: $tap";
 }
 
-my @tap_files;
-if (-d $tap) {
-  @tap_files = map { $_->stringify } path($tap)->children;
-} else {
-  push @tap_files, $tap;
-}
-
-my %results;
-my $datacheck;
-my $species;
-my $test;
-my %tests;
-
-foreach my $tap_file (@tap_files) {
-  my $tap = path($tap_file)->slurp;
-  my $parser = TAP::Parser->new( { tap => $tap } );
-
-  # to-do: extract test number, use that as key with test message and diag messages as lists.
-  while (my $result = $parser->next) {
-    if ($result->is_comment) {
-      # Unindented 'Subtest' comment is the name of the datacheck,
-      # indented 'Subtest's are the species/database.
-      if ($result->as_string =~ /^# Subtest: (.+)/) {
-        $datacheck = $1;
-      }
-    } elsif ($result->is_unknown) {
-      if ($result->as_string =~ /^\s+# Subtest: (.+)/) {
-        $species = $1;
-        %tests = ();
-      } elsif ($result->as_string =~ /^\s{8}((?:not ok|# No tests run).*)/) {
-        $test = $1;
-        $tests{$test} = [];
-      } elsif ($result->as_string =~ /^\s{8}((?:ok|.* # SKIP).*)/ && $passed) {
-        $test = $1;
-        $tests{$test} = [];
-      } elsif ($result->as_string =~ /^\s{8}#\s(\s*.*)/) {
-        if (defined $test) {
-          push @{$tests{$test}}, $1;
-        } else {
-          warn "Premature diagnostication: diagnostics incomplete ".
-               "for $species because they cannot be linked to a test";
-        }
-      } elsif ($result->as_string =~ /^\s{4}((?:ok|not ok))/) {
-        my $ok = $1 eq 'ok' ? 1 : 0;
-        if (!$ok || $passed) {
-          my %datacheck_tests = %tests;
-          if ($by_species) {
-            $results{$species}{$datacheck}{'ok'} = $ok;
-            $results{$species}{$datacheck}{'tests'} = \%datacheck_tests;
-          } else {
-            $results{$datacheck}{$species}{'ok'} = $ok;
-            $results{$datacheck}{$species}{'tests'} = \%datacheck_tests;
-          }
-        }
-        $test = undef;
-      }  
-    }
-  }
-}
-
-my $json = JSON->new->canonical->pretty->encode(\%results);
+Bio::EnsEMBL::DataCheck::Pipeline::DataCheckTapToJson::parse_datachecks($tap, $output_file, $by_species, $passed, 0);
 
-if ($output_file) {
-  path($output_file)->parent->mkpath;
-  path($output_file)->spew($json)
-} else {
-  say $json;
-}