Merge pull request #1617 from seek4science/related-sample-datafile-at…

…tribute

Datafile attribute relates samples and datafiles

app/models/data_file.rb

@@ -20,6 +20,8 @@ class DataFile < ApplicationRecord
 
   belongs_to :file_template
   has_many :extracted_samples, class_name: 'Sample', foreign_key: :originating_data_file_id
+  has_many :sample_resource_links, -> { where(resource_type: 'DataFile') }, class_name: 'SampleResourceLink', foreign_key: :resource_id
+  has_many :linked_samples, through: :sample_resource_links, source: :sample
 
   has_many :workflow_data_files, dependent: :destroy, autosave: true
   has_many :workflows, ->{ distinct }, through: :workflow_data_files
@@ -122,7 +124,7 @@ def possible_sample_types(user = User.current_user)
   end
 
   def related_samples
-    extracted_samples
+    extracted_samples + linked_samples
   end
 
   # Extracts samples using the given sample_type

app/models/sample.rb

@@ -30,6 +30,8 @@ class Sample < ApplicationRecord
   has_many :linked_samples, through: :sample_resource_links, source: :resource, source_type: 'Sample'
   has_many :linking_samples, through: :reverse_sample_resource_links, source: :sample
 
+  has_many :linked_data_files, through: :sample_resource_links, source: :resource, source_type: 'DataFile'
+
   validates :projects, presence: true, projects: { self: true }
   validates :title, :sample_type, presence: true
 
@@ -60,8 +62,8 @@ def self.can_create?
     User.logged_in_and_member? && Seek::Config.samples_enabled
   end
 
-  def related_data_file
-    originating_data_file
+  def related_data_files
+    [originating_data_file].compact + linked_data_files
   end
 
   def related_samples
@@ -81,6 +83,10 @@ def referenced_resources
     end.flatten.compact
   end
 
+  def referenced_data_files
+    referenced_resources.select { |r| r.is_a?(DataFile) }
+  end
+
   def referenced_strains
     referenced_resources.select { |r| r.is_a?(Strain) }
   end
@@ -198,6 +204,7 @@ def update_sample_resource_links
     return unless sample_type.present?
     self.strains = referenced_strains
     self.linked_samples = referenced_samples
+    self.linked_data_files = referenced_data_files
   end
 
   def attribute_class

lib/tasks/seek_upgrades.rake

@@ -10,6 +10,7 @@ namespace :seek do
     environment
     decouple_extracted_samples_policies
     decouple_extracted_samples_projects
+    link_sample_datafile_attributes
   ]
 
   # these are the tasks that are executes for each upgrade as standard, and rarely change
@@ -84,6 +85,22 @@ namespace :seek do
     puts " ... finished copying project ids of #{decoupled.to_s} extracted samples"
   end
 
+  task(link_sample_datafile_attributes: [:environment]) do
+    puts '... updating sample_resource_links for samples with data_file attributes...'
+    samples_updated = 0
+    disable_authorization_checks do
+      df_attrs = SampleAttribute.joins(:sample_attribute_type).where('sample_attribute_types.base_type' => Seek::Samples::BaseType::SEEK_DATA_FILE).pluck(:id)
+      samples = Sample.joins(sample_type: :sample_attributes).where('sample_attributes.id' => df_attrs)
+      samples.each do |sample|
+        if sample.sample_resource_links.where(resource_type: 'DataFile').empty?
+          sample.send(:update_sample_resource_links)
+          samples_updated += 1
+        end
+      end
+    end
+    puts " ... finished updating sample_resource_links of #{samples_updated.to_s} samples with data_file attributes"
+  end
+
   private
 
   ##

test/unit/data_file_test.rb

@@ -489,4 +489,25 @@ class DataFileTest < ActiveSupport::TestCase
     assert data_file.respond_to?(:data_format_annotations)
     assert data_file.respond_to?(:data_type_annotations)
   end
+
+  test 'related samples include extracted samples and samples linked in attributes' do
+    project = FactoryBot.create(:project)
+    df_no_samples = FactoryBot.create(:data_file) # No samples
+    df_extracted = FactoryBot.create(:data_file)  # Extracted samples
+    df_attributes = FactoryBot.create(:data_file) # Linked in sample's attributes only
+    df_ext_attr = FactoryBot.create(:data_file)   # Extracted samples and linked in sample's attributes
+    type = FactoryBot.create(:data_file_sample_type, project_ids: [project.id])
+    sample1 = Sample.new(sample_type: type, project_ids: [project.id], originating_data_file: df_extracted)
+    sample1.update(data: { 'data file': df_ext_attr.id })
+    sample1.save!
+    sample2 = Sample.new(sample_type: type, project_ids: [project.id], originating_data_file: df_ext_attr)
+    sample2.update(data: { 'data file': df_attributes.id })
+    sample2.save!
+
+    assert_equal [], df_no_samples.related_samples
+    assert_equal [sample1], df_extracted.related_samples
+    assert_equal [sample2].sort_by(&:id), df_attributes.related_samples
+    assert_equal [sample1, sample2].sort_by(&:id), df_ext_attr.related_samples.sort_by(&:id)
+  end
+
 end

test/unit/sample_test.rb

@@ -1352,4 +1352,34 @@ class SampleTest < ActiveSupport::TestCase
 
   end
 
+  test 'related datafiles include datafiles in attributes and originating datafile' do
+    project = FactoryBot.create(:project)
+    simple_type = FactoryBot.create(:simple_sample_type, project_ids: [project.id])
+    type_with_df_attr = FactoryBot.create(:data_file_sample_type, project_ids: [project.id])
+    df_attr = FactoryBot.build(:data_file_sample_attribute, title: 'data file 2', sample_type: type_with_df_attr)
+    type_with_df_attr.sample_attributes << df_attr
+    df1 = FactoryBot.create(:data_file)
+    df2 = FactoryBot.create(:data_file)
+    df3 = FactoryBot.create(:data_file)
+    # Sample with no data files linked as attribute nor originating data file
+    sample_no_df = Sample.new(sample_type: simple_type, project_ids: [project.id])
+    # Sample with originating data file only
+    sample_extracted = Sample.new(sample_type: simple_type, project_ids: [project.id], originating_data_file: df3)
+    # Sample with data files linked as attributes
+    sample_attributes = Sample.new(sample_type: type_with_df_attr, project_ids: [project.id])
+    sample_attributes.update(data: { 'data file': df1.id })
+    sample_attributes.update(data: { 'data file 2': df2.id })
+    sample_attributes.save!
+    # Sample with data files linked as attributes and originating data file
+    sample_ext_attr = Sample.new(sample_type: type_with_df_attr, project_ids: [project.id], originating_data_file: df3)
+    sample_ext_attr.update(data: { 'data file': df1.id })
+    sample_ext_attr.update(data: { 'data file 2': df2.id })
+    sample_ext_attr.save!
+
+    assert_equal [], sample_no_df.related_data_files
+    assert_equal [df3], sample_extracted.related_data_files
+    assert_equal [df1, df2].sort_by(&:id), sample_attributes.related_data_files.sort_by(&:id)
+    assert_equal [df1, df2, df3].sort_by(&:id), sample_ext_attr.related_data_files.sort_by(&:id)
+  end
+
 end