Merge pull request #100 from sanger-pathogens/fix_validate_sample_man…

…ifest

Fix validate sample manifest

Dockerfile

@@ -0,0 +1,56 @@
+FROM ubuntu:18.04
+MAINTAINER = [email protected]
+
+# dependency versions -- COULD specify these at build time
+ARG VR_CODEBASE_VERSION=0.04
+ARG CONFIG_GENERAL_VERSION=2.52
+
+RUN apt-get update -qq -y
+RUN apt-get upgrade -qq -y
+RUN apt-get install -qq -y \
+    default-jdk \
+    build-essential \
+    git \
+    file \
+    wget \
+    curl \
+    libxml2-dev \
+    libexpat1-dev \
+    libgd-dev \
+    libssl-dev \
+    libdb-dev \
+    libmysqlclient-dev \
+    cpanminus \
+    locales \
+    genometools
+
+RUN   sed -i -e 's/# \(en_GB\.UTF-8 .*\)/\1/' /etc/locale.gen && \
+      touch /usr/share/locale/locale.alias && \
+      locale-gen
+ENV   LANG     en_GB.UTF-8
+ENV   LANGUAGE en_GB:en
+ENV   LC_ALL   en_GB.UTF-8
+
+# dzil
+RUN cpanm --notest \
+    Dist::Zilla \
+    Config::General@${CONFIG_GENERAL_VERSION}
+
+# vr-codebase
+RUN cd /opt \
+    && wget -q https://github.com/sanger-pathogens/vr-codebase/archive/v${VR_CODEBASE_VERSION}.tar.gz \
+    && tar xf v${VR_CODEBASE_VERSION}.tar.gz \
+    && rm v${VR_CODEBASE_VERSION}.tar.gz
+ENV PERL5LIB /opt/vr-codebase-${VR_CODEBASE_VERSION}/modules:$PERL5LIB
+
+# bio-ena-datasubmission
+RUN mkdir -p /opt/Bio-ENA-DataSubmission
+COPY . /opt/Bio-ENA-DataSubmission
+ENV PATH /opt/Bio-ENA-DataSubmission/bin:$PATH
+ENV PERL5LIB /opt/Bio-ENA-DataSubmission/lib:$PERL5LIB
+ENV ENA_SUBMISSIONS_DATA /opt/Bio-ENA-DataSubmission/data
+RUN cd /opt/Bio-ENA-DataSubmission && dzil authordeps --missing | cpanm --notest
+RUN cd /opt/Bio-ENA-DataSubmission && dzil listdeps --missing | grep -v 'VRTrack::Lane' | cpanm --notest
+
+RUN   cd /opt/Bio-ENA-DataSubmission && \
+      dzil test

install-travis-dependencies.sh

@@ -10,25 +10,27 @@ cpanm --notest Moose
 cpanm --notest YAML::XS
 cpanm --notest DBD::mysql
 
-
-VRCODEBASE_GIT_URL='https://github.com/sanger-pathogens/vr-codebase.git'
+VR_CODEBASE_VERSION=0.04
+VRCODEBASE_GIT_URL="https://github.com/sanger-pathogens/vr-codebase/archive/v${VR_CODEBASE_VERSION}.tar.gz"
 
 # Make an install location
 if [ ! -d 'git_repos' ]; then
   mkdir git_repos
 fi
 cd git_repos
 
-git clone $VRCODEBASE_GIT_URL
+wget -O- ${VRCODEBASE_GIT_URL} | tar xzvf -
 
 #Add locations to PERL5LIB
-VRCODEBASE_LIB=${start_dir}'/git_repos/vr-codebase/modules'
+VRCODEBASE_LIB="${start_dir}/git_repos/vr-codebase-${VR_CODEBASE_VERSION}/modules"
 
 export PERL5LIB=${VRCODEBASE_LIB}:$PERL5LIB
 
 cd $start_dir
 dzil authordeps --missing | cpanm --notest
 dzil listdeps --missing | grep -v 'VRTrack::Lane' | cpanm --notest
 
+dzil test
+
 set +eu
-set +x
+set +x

lib/Bio/ENA/DataSubmission/CommandLine/ValidateManifest.pm

@@ -149,17 +149,21 @@ sub run {
     my %taxon_id_to_ncbi_taxon_name;
 
     my @errors_found;
-    foreach my $r (0 .. $#manifest) {
-        next unless (defined $manifest[$r]);
+    my $row_num = 0;
+    MANIFEST_ROW: foreach my $r (0 .. $#manifest) {
+        next MANIFEST_ROW unless (defined $manifest[$r]);
         my @row = @{$manifest[$r]};
+        ++$row_num;
         my $missing_accessions_error = Bio::ENA::DataSubmission::Validator::Error::MandatorySampleAccession->new(
             row       => \@row,
             mandatory => 0,
             rownumber => $r
         )->validate;
-        push(@errors_found, $missing_accessions_error) if ($missing_accessions_error->triggered);
-        next unless $#errors_found == -1;
-
+        if($missing_accessions_error->triggered) {
+          push(@errors_found, $missing_accessions_error);
+          warn "Manifest row $row_num has a missing accession number: NO FURTHER CHECKS will be done on this row";
+          next MANIFEST_ROW;
+        }
         my $acc = $row[0];
 
         # validate all generally
@@ -175,7 +179,7 @@ sub run {
             }
         }
 
-        # validate more specific cells separately
+        # validate mo re specific cells separately
 
         # mandatory cells
         my $mandatory = [ 4, 5, 14, 15, 16, 17, 19, 25, 28 ];

t/Bio/ENA/DataSubmission/WEBINCli.t

@@ -188,11 +188,15 @@ sub test_directory_not_a_directory {
 
 # Check manifest validation when file is not readable
 {
-    my $args = { %full_args };
-    $args->{'manifest'} = $unreadable_manifest;
-    my $under_test = Bio::ENA::DataSubmission::WEBINCli->new(%$args);
-    throws_ok {$under_test->run()} 'Bio::ENA::DataSubmission::Exception::CannotReadFile', "dies if manifest is not a readable file";
-
+   SKIP: {
+      skip "running as root, so disabled the test \"Check manifest validation when file is not readable\"", 1
+         if ('root' eq scalar(getpwuid $>));
+
+      my $args = { %full_args };
+      $args->{'manifest'} = $unreadable_manifest;
+      my $under_test = Bio::ENA::DataSubmission::WEBINCli->new(%$args);
+      throws_ok {$under_test->run()} 'Bio::ENA::DataSubmission::Exception::CannotReadFile', "dies if manifest is not a readable file";
+   }
 }
 
 remove_tree($temp_input_dir_name, $temp_output_dir_name, $filename, $manifest_filename, $temp_dir_name, $unreadable_manifest);