Merge pull request #375 from sul-dlss/manifest_cols

remove manifest_cols

app/lib/pre_assembly/bundle.rb

@@ -15,7 +15,6 @@ class Bundle
              :content_exclusion,
              :content_md_creation,
              :content_structure,
-             :manifest_cols,
              :manifest_rows,
              :path_in_bundle,
              :progress_log_file,
@@ -89,8 +88,8 @@ def digital_objects
         DigitalObject.new(self, params).tap do |dobj|
           r = manifest_rows[i]
           # Get label and source_id from column names declared in YAML config.
-          dobj.label        = manifest_cols[:label] ? r[manifest_cols[:label]] : ''
-          dobj.source_id    = r[manifest_cols[:source_id]] if manifest_cols[:source_id]
+          dobj.label        = r['label'] || ''
+          dobj.source_id    = r['sourceid']
           # Also store a hash of all values from the manifest row, using column names as keys.
           dobj.manifest_row = r
         end
@@ -102,11 +101,8 @@ def digital_objects
     # manifest columns. The column name to use is configured by the
     # user invoking the pre-assembly script.
     def discover_containers_via_manifest
-      raise ':manifest_cols must be specified' unless manifest_cols
-      # TODO: note that manifest_cols is a constant in bundle_context
-      obj_sym = manifest_cols[:object_container].to_sym
-      manifest_rows.each_with_index { |r, i| raise "Missing #{obj_sym} in row #{i}: #{r}" unless r[obj_sym] }
-      manifest_rows.map { |r| path_in_bundle r[obj_sym] }
+      manifest_rows.each_with_index { |r, i| raise "Missing 'object' in row #{i}: #{r}" unless r[:object] }
+      manifest_rows.map { |r| path_in_bundle r[:object] }
     end
 
     # A method to discover object containers or stageable items.
@@ -168,7 +164,7 @@ def load_checksums(dobj)
 
     # confirm that the all of the source IDs supplied within a manifest are locally unique
     def manifest_sourceids_unique?
-      all_source_ids = manifest_rows.collect { |r| r[manifest_cols[:source_id]] }
+      all_source_ids = manifest_rows.collect { |r| r['sourceid'] }
       all_source_ids.size == all_source_ids.uniq.size
     end
 

app/models/bundle_context.rb

@@ -87,16 +87,6 @@ def manifest_rows
     @manifest_rows ||= CsvImporter.parse_to_hash(path_in_bundle(manifest))
   end
 
-  # TODO: make this simpler / remove it (#329)
-  def manifest_cols
-    {
-      label: 'label', # only used by SMPL manifests
-      source_id: 'sourceid', # only used by SMPL manifests
-      object_container: 'object', # object referring to filename or foldername
-      druid: 'druid'
-    }
-  end
-
   private
 
   def default_enums

spec/lib/pre_assembly/bundle_spec.rb

@@ -63,7 +63,6 @@
   describe '#object_discovery: discovery via manifest and crawl' do
     it 'discover_containers_via_manifest() should return expected information' do
       vals = %w[123.tif 456.tif 789.tif]
-      flat_dir_images.manifest_cols[:object_container] = :col_foo
       allow(flat_dir_images).to receive(:manifest_rows).and_return(vals.map { |v| { object: v } })
       expect(flat_dir_images.discover_containers_via_manifest).to eq(vals.map { |v| flat_dir_images.path_in_bundle v })
     end

spec/models/bundle_context_spec.rb

@@ -135,17 +135,6 @@
     end
   end
 
-  describe 'manifest_cols' do
-    it 'sets the column names' do
-      expect(bc.manifest_cols).to eq(
-        label: 'label',
-        source_id: 'sourceid',
-        object_container: 'object', # object referring to filename or foldername
-        druid: 'druid'
-      )
-    end
-  end
-
   describe '#verify_output_dir (private method)' do
     before { FileUtils.mkdir_p(Settings.job_output_parent_dir) }
 

spec/test_data/exemplar_templates/TEMPLATE.yaml

@@ -132,13 +132,9 @@ checksums_file:   'checksums.txt'      # A provider checksum file (in default md
 desc_md_template: 'mods_template.xml'  # An optional descriptive metadata XML template to use in conjunction with the manifest.  Path can be absolute or relative to the bundle path.
 
 ####
-# You must indicate the column names of certain fields in the CSV file.
-# These should always be lowercase.  Note that if your project is using a manifest for object discovery and has objects already registered, you will
-# need a column called "druid" in addition to the columns below.  The "source_id" and "label" columns must exist in this case but can be blank (since objects are already registered).
-manifest_cols:
-  object_container:   'filename'   # Always required, indicates the column name containing the filename (single file per object) or folder name (multiple files per object) relative to the bundle directory.
-  source_id:          'sourceid'   # optional
-  label:              'label'      # optional
+# The CSV file is expected with columns (always lowercase):
+# - 'druid', required
+# - 'object', required
 
 ####
 # Attributes related to content metadata generation.