Updates to allow files in separate rows

- Use required headers for whole sheet for preflight (combines requirements for object types)
- Find parent work based on deduplication_field
- Raise an error for now if parent work is not found. Otherwise the stack trace is misleading.

app/uploaders/zizia/csv_manifest_validator.rb

@@ -87,7 +87,7 @@ def map_object_type(orig_value)
     end
 
     def missing_headers
-      required_headers.each do |header|
+      required_headers_for_sheet.each do |header|
         next if @transformed_headers.include?(header)
         @errors << "Missing required column: \"#{header.titleize}\".  Your spreadsheet must have this column."
       end
@@ -108,6 +108,9 @@ def required_headers(object_type = "w")
       end
     end
 
+    # TODO: Map these headers appropriately all the way through the ingest
+    # Right now the transformed headers only downcase and strip them, they don't translate them
+    # based on the associated mapper
     def work_headers
       ['title', 'creator', 'keyword', 'rights statement', 'visibility', 'files', 'deduplication_key']
     end

lib/zizia/hyrax/hyrax_record_importer.rb

@@ -262,9 +262,12 @@ def create_curation_concern(record, import_type)
         end
       end
 
+      # TODO: Right now we assume that the parent is a Work, but it would be more generally
+      # applicable to let this be any sort of CurationConcern
       def find_parent_work(record)
-        parent_identifier = record.parent.first
-        Work.where(identifier: parent_identifier).first
+        parent_work = Work.where("#{deduplication_field}": record.mapper.send(deduplication_field).to_s).first
+        return parent_work if parent_work.present?
+        raise "[zizia] Parent work for file not found, cannot attach file to work"
       end
 
       def create_file_set(record)

spec/dummy/spec/fixtures/csv_import/csv_files_with_problems/mix_of_object_types_missing_header.csv

@@ -0,0 +1,8 @@
+object type,title,creator,keyword,rights statement,visibility,files,deduplication_key
+c,Collection of vegetables,,,,public,,def/123
+Collection,Collection of fruits,,,,public,,def/234
+w,Work on tomatoes,"Tomato, Tommy",fruits|~|tomatoes,http://rightsstatements.org/vocab/InC/1.0/,public,birds.jpg,abc/123
+wOrk,Work on cabbages,"Cabbage, Carl",vegetables|~|cabbages,http://rightsstatements.org/vocab/InC/1.0/,public,dog.jpg,abs/234
+coLLEction,Collection of legumes,,,,public,,def/345
+,Work without object type,"Cabbage, Carl",vegetables|~|cabbages,http://rightsstatements.org/vocab/InC/1.0/,public,dog.jpg,abc/345
+f,,,,,,dog.jpg,,abc/123

spec/dummy/spec/fixtures/csv_import/csv_files_with_problems/mix_of_object_types_missing_parent_for_file.csv

@@ -0,0 +1,8 @@
+object type,title,creator,keyword,rights statement,visibility,files,deduplication_key,parent
+c,Collection of vegetables,,,,public,,def/123,
+Collection,Collection of fruits,,,,public,,def/234,
+w,Work on tomatoes,"Tomato, Tommy",fruits|~|tomatoes,http://rightsstatements.org/vocab/InC/1.0/,public,birds.jpg,abc/123,
+wOrk,Work on cabbages,"Cabbage, Carl",vegetables|~|cabbages,http://rightsstatements.org/vocab/InC/1.0/,public,dog.jpg,abs/234,
+coLLEction,Collection of legumes,,,,public,,def/345,
+,Work without object type,"Cabbage, Carl",vegetables|~|cabbages,http://rightsstatements.org/vocab/InC/1.0/,public,dog.jpg,abc/345,
+f,,,,,,dog.jpg,,

spec/dummy/spec/fixtures/csv_import/good/mix_of_object_types.csv

@@ -1,8 +1,8 @@
-object type,title,creator,keyword,rights statement,visibility,files,deduplication_key
-c,Collection of vegetables,,,,public,,def/123
-Collection,Collection of fruits,,,,public,,def/234
-w,Work on tomatoes,"Tomato, Tommy",fruits|~|tomatoes,http://rightsstatements.org/vocab/InC/1.0/,public,birds.jpg,abc/123
-wOrk,Work on cabbages,"Cabbage, Carl",vegetables|~|cabbages,http://rightsstatements.org/vocab/InC/1.0/,public,dog.jpg,abs/234
-coLLEction,Collection of legumes,,,,public,,def/345
-,Work without object type,"Cabbage, Carl",vegetables|~|cabbages,http://rightsstatements.org/vocab/InC/1.0/,public,dog.jpg,abc/345
-f,,,,,,dog.jpg,,abc/123
+object type,title,creator,keyword,rights statement,visibility,files,identifier,deduplication_key,parent
+c,Collection of vegetables,,,,public,,def/123,def/123,
+Collection,Collection of fruits,,,,public,,def/234,def/234,
+w,Work on tomatoes,"Tomato, Tommy",fruits|~|tomatoes,http://rightsstatements.org/vocab/InC/1.0/,public,birds.jpg,abc/123,abc/123,
+wOrk,Work on cabbages,"Cabbage, Carl",vegetables|~|cabbages,http://rightsstatements.org/vocab/InC/1.0/,public,dog.jpg,abs/234,abs/234,
+coLLEction,Collection of legumes,,,,public,,def/345,def/345,
+,Work without object type,"Cabbage, Carl",vegetables|~|cabbages,http://rightsstatements.org/vocab/InC/1.0/,public,dog.jpg,abc/345,abc/345,
+f,,,,,,dog.jpg,,,abc/123

spec/uploaders/zizia/csv_manfest_validator_spec.rb

@@ -54,6 +54,26 @@
     end
   end
 
+  context "with a mix of object types missing a field required for only one object type" do
+    let(:path_to_file) { Rails.root.join('spec', 'fixtures', 'csv_import', 'csv_files_with_problems', 'mix_of_object_types_missing_header.csv') }
+
+    it "gives an error for the missing header based on having a file row" do
+      validator.validate
+      expect(validator.errors).to eq(['Missing required column: "Parent".  Your spreadsheet must have this column.'])
+      expect(validator.warnings).to eq([])
+    end
+  end
+
+  context "with a mix of object types missing a value required for only one object type" do
+    let(:path_to_file) { Rails.root.join('spec', 'fixtures', 'csv_import', 'csv_files_with_problems', 'mix_of_object_types_missing_parent_for_file.csv') }
+
+    it "gives an error for the missing value based on having a file row" do
+      validator.validate
+      expect(validator.errors).to eq(['Missing required metadata in row 8: "Parent" field cannot be blank'])
+      expect(validator.warnings).to eq([])
+    end
+  end
+
   context "with a object type column" do
     let(:path_to_file) { Rails.root.join('spec', 'fixtures', 'csv_import', 'good', 'Postcards_Minneapolis_w_collection.csv') }
     let(:work_row) do
@@ -75,6 +95,7 @@
     it "returns required headers based on the object type" do
       required_work_headers = ['title', 'creator', 'keyword', 'rights statement', 'visibility', 'files', 'deduplication_key']
       required_collection_headers = ['title', 'visibility']
+      required_file_headers = ["files", "parent"]
       expect(validator.required_headers).to eq(required_work_headers)
       expect(validator.required_headers("w")).to eq(required_work_headers)
       expect(validator.required_headers("c")).to eq(required_collection_headers)
@@ -83,6 +104,7 @@
       expect(validator.required_headers("wOrk")).to eq(required_work_headers)
       expect(validator.required_headers('garbage')).to eq(required_work_headers)
       expect(validator.required_headers('')).to eq(required_work_headers)
+      expect(validator.required_headers('file')).to eq(required_file_headers)
     end
 
     it "returns different required column numbers based on the row" do