[#805] Export Metadata Subfield Funding References

Add script to convert CSV of funding data into ruby hash stored within
text file. Use function to read file and extract funding data.
closes #805

app/models/concerns/galtersufia/generic_file/funding_data.txt

@@ -0,0 +1 @@
+{"ns0646000"=>[{:funder=>{:name=>"National Library of Medicine (NLM)", :identifier=>"0060t0j89", :scheme=>"ror"}, :award=>{:title=>"", :number=>"F37 LM009568 ", :identifier=>"", :scheme=>""}}], "n870zq819"=>[{:funder=>{:name=>"National Library of Medicine (NLM)", :identifier=>"0060t0j89", :scheme=>"ror"}, :award=>{:title=>"", :number=>"F37 LM009568 ", :identifier=>"", :scheme=>""}}], "9z902z89q"=>[{:funder=>{:name=>"National Library of Medicine (NLM)", :identifier=>"0060t0j89", :scheme=>"ror"}, :award=>{:title=>"", :number=>"K01LM011973-01", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"UL1RR025741", :identifier=>"", :scheme=>""}}], "0z708w45r"=>[{:funder=>{:name=>"Swiss National Science Foundation", :identifier=>"00yjd3n13", :scheme=>"ror"}, :award=>{:title=>"", :number=>"3200B0-108261", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"Mach-Gaensslen Foundation", :identifier=>"0009ezc57", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "hx11xf327"=>[{:funder=>{:name=>"Otho S. A. Sprague Memorial Institute", :identifier=>"00c2b9157", :scheme=>"ror"}, :award=>{:title=>"", :number=>"HE 07057", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"Otho S. A. Sprague Memorial Institute", :identifier=>"00c2b9157", :scheme=>"ror"}, :award=>{:title=>"", :number=>"HE 10573", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"Ernest S. Bazley Trust", :identifier=>"", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "8p58pc98c"=>[{:funder=>{:name=>"National Library of Medicine (NLM)", :identifier=>"0060t0j89", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"Oak Ridge Institute for Science and Education", :identifier=>"", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "c0acc95c-7f32-48de-83a4-c830bec4bb29"=>[{:funder=>{:name=>"National Center for Chronic Disease Prevention and Health Promotion", :identifier=>"021rths28", :scheme=>"ror"}, :award=>{:title=>"", :number=>"DP006255", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Institute on Drug Abuse", :identifier=>"00fq5cm18", :scheme=>"ror"}, :award=>{:title=>"", :number=>"DA027828", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"Department of Health and Human Services", :identifier=>"00b9jhh55", :scheme=>"ror"}, :award=>{:title=>"", :number=>"SM60563-40", :identifier=>"", :scheme=>""}}], "4de23834-0309-48f9-9b9b-05dbaa6940ec"=>[{:funder=>{:name=>"National Institute on Drug Abuse (NIDA)", :identifier=>"00fq5cm18", :scheme=>"ror"}, :award=>{:title=>"", :number=>"2P30DA027828-07", :identifier=>"", :scheme=>""}}], "3e5cf487-2383-4d1f-b697-ed40a8b79670"=>[{:funder=>{:name=>"National Library of Medicine (NLM)", :identifier=>"0060t0j89", :scheme=>"ror"}, :award=>{:title=>"", :number=>"K01LM011973-01", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"UL1TR001422", :identifier=>"", :scheme=>""}}], "2315fdf3-f8ad-461f-8151-0daa68ebfd2c"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"UL1TR001422", :identifier=>"", :scheme=>""}}], "1bff6ec0-af92-48dd-a8e2-d17e4ac4f1ec"=>[{:funder=>{:name=>"Penn Center for AIDS Research (CFAR)", :identifier=>"047939x15", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Institute of Allergy and Infectious Diseases (NIAID)", :identifier=>"043z4tv69", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "37a5967d-8e25-47d1-92d3-9861755ccdc8"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U24TR002306", :identifier=>"", :scheme=>""}}], "6f032b75-efdb-4ff9-8154-7443b2cac1e0"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"UL1TR001422", :identifier=>"", :scheme=>""}}], "d282192b-5d42-4f97-853d-72530ae15ba8"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"UL1TR001422", :identifier=>"", :scheme=>""}}], "1aaa14ed-50b0-4c3c-a249-d088b8424a1d"=>[{:funder=>{:name=>"Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)", :identifier=>"04byxyr05", :scheme=>"ror"}, :award=>{:title=>"", :number=>"P50HD076188", :identifier=>"", :scheme=>""}}], "7f794595-230e-48a0-aee1-d92f567d09a7"=>[{:funder=>{:name=>"Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)", :identifier=>"04byxyr05", :scheme=>"ror"}, :award=>{:title=>"", :number=>"P50HD076188", :identifier=>"", :scheme=>""}}], "28201f5d-8f20-4d1c-add3-409ca6f061a5"=>[{:funder=>{:name=>"Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)", :identifier=>"04byxyr05", :scheme=>"ror"}, :award=>{:title=>"", :number=>"P50HD076188", :identifier=>"", :scheme=>""}}], "50abe4ff-81d9-40ab-bafd-24d68b1d9c40"=>[{:funder=>{:name=>"Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)", :identifier=>"04byxyr05", :scheme=>"ror"}, :award=>{:title=>"", :number=>"P50HD076188", :identifier=>"", :scheme=>""}}], "1b68450e-37e2-4dbe-8630-3ffc812c3040"=>[{:funder=>{:name=>"Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)", :identifier=>"04byxyr05", :scheme=>"ror"}, :award=>{:title=>"", :number=>"P50HD076188", :identifier=>"", :scheme=>""}}], "11306426-f23a-4db9-bb68-b49f43362d41"=>[{:funder=>{:name=>"Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)", :identifier=>"04byxyr05", :scheme=>"ror"}, :award=>{:title=>"", :number=>"P50HD076188", :identifier=>"", :scheme=>""}}], "10014f57-93aa-4fe1-bc0b-7444813e9b3a"=>[{:funder=>{:name=>"Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)", :identifier=>"04byxyr05", :scheme=>"ror"}, :award=>{:title=>"", :number=>"P50HD076188", :identifier=>"", :scheme=>""}}], "09c20041-3882-4d6d-8b74-5469e33612e6"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U24TR002306", :identifier=>"", :scheme=>""}}], "c6f785cf-f992-44ea-9905-ab4316181d91"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"UL1TR001422", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Cancer Institute (NCI)", :identifier=>"040gcmg81", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U54CA202995", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Cancer Institute (NCI)", :identifier=>"040gcmg81", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U54CA202997", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Cancer Institute (NCI)", :identifier=>"040gcmg81", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U54CA203000", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Library of Medicine (NLM)", :identifier=>"0060t0j89", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U24TR002306", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U24TR002306", :identifier=>"", :scheme=>""}}], "fb4aa42b-ffe8-4548-b6cf-2fa6e32a80ac"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U24TR002306", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"UL1TR001422", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Cancer Institute (NCI)", :identifier=>"040gcmg81", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U54CA202995", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Cancer Institute (NCI)", :identifier=>"040gcmg81", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U54CA202997 ", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Cancer Institute (NCI)", :identifier=>"040gcmg81", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U54CA203000", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS)", :identifier=>"006zn3t30", :scheme=>"ror"}, :award=>{:title=>"", :number=>"P30AR072579 ", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Library of Medicine (NLM)", :identifier=>"0060t0j89", :scheme=>"ror"}, :award=>{:title=>"", :number=>"G08LM012688", :identifier=>"", :scheme=>""}}], "f9070cb2-95e5-4d5d-83f6-d68b50d4e343"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U24TR002306", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"UL1TR001422", :identifier=>"", :scheme=>""}}], "370b309f-c33b-4936-996b-3717ed0f6074"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U24TR002306", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>" UL1TR001422", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Cancer Institute (NCI)", :identifier=>"040gcmg81", :scheme=>"ror"}, :award=>{:title=>"", :number=>" U54CA202995", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Cancer Institute (NCI)", :identifier=>"040gcmg81", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U54CA202997", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Cancer Institute (NCI)", :identifier=>"040gcmg81", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U54CA203000", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS)", :identifier=>"006zn3t30", :scheme=>"ror"}, :award=>{:title=>"", :number=>"P30AR072579 ", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Library of Medicine (NLM)", :identifier=>"0060t0j89", :scheme=>"ror"}, :award=>{:title=>"", :number=>"G08LM012688", :identifier=>"", :scheme=>""}}], "691d9bf4-5be8-479f-bb74-eff2d98bfa9e"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U24TR002306", :identifier=>"", :scheme=>""}}], "2d858062-bd0a-4b26-bf24-3f1637e7aa94"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"UL1TR001", :identifier=>"", :scheme=>""}}], "a1419467-3a15-4e85-b2f4-1ac0c057c60f"=>[{:funder=>{:name=>"Fogarty International Center", :identifier=>"02xey9a22", :scheme=>"ror"}, :award=>{:title=>"", :number=>"D43TW009575", :identifier=>"", :scheme=>""}}], nil=>[{:funder=>{:name=>"National Cancer Institute (NCI)", :identifier=>"040gcmg81", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U54CA221205", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"American Association for the Study of Liver Diseases", :identifier=>"005tjs461", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)", :identifier=>"00adh9b73", :scheme=>"ror"}, :award=>{:title=>"", :number=>"R01DK110024", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Institute on Alcohol Abuse and Alcoholism (NIAAA)", :identifier=>"02jzrsm59", :scheme=>"ror"}, :award=>{:title=>"", :number=>"R01AA027179", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"", :identifier=>"", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "72802492-bdd9-4bdb-80a9-85e63cdc09e4"=>[{:funder=>{:name=>"National Center for Advancing Translational Sciences (NCATS)", :identifier=>"04pw6fb54", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U24TR002306", :identifier=>"", :scheme=>""}}], "78254bed-92a7-4708-bcff-16383bab9ce3"=>[{:funder=>{:name=>"", :identifier=>"", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "c423cf8b-8cbb-4407-9b3e-dd30bd86769a"=>[{:funder=>{:name=>"", :identifier=>"", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "d43f1bbc-9099-45fa-83e2-79d1301a0aeb"=>[{:funder=>{:name=>"", :identifier=>"", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "ae10b9d6-f1d1-43bc-a5bc-2b6e024d263b"=>[{:funder=>{:name=>"", :identifier=>"", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "b876a078-d5da-4f9f-b8e4-cc1cd75489a5"=>[{:funder=>{:name=>"", :identifier=>"", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "d593a1e5-bc29-4db0-94d2-3ec5471ba11f"=>[{:funder=>{:name=>"National Heart Lung and Blood Institute (NHLBI)", :identifier=>"012pb6c26", :scheme=>"ror"}, :award=>{:title=>"", :number=>"R01HL115828", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Heart Lung and Blood Institute (NHLBI)", :identifier=>"012pb6c26", :scheme=>"ror"}, :award=>{:title=>"", :number=>"R01HL133504", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Heart Lung and Blood Institute (NHLBI)", :identifier=>"012pb6c26", :scheme=>"ror"}, :award=>{:title=>"", :number=>"K25HL119608", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"National Heart Lung and Blood Institute (NHLBI)", :identifier=>"012pb6c26", :scheme=>"ror"}, :award=>{:title=>"", :number=>"F30HL145995", :identifier=>"", :scheme=>""}}], "45a68a08-9aa6-4e13-8f7e-fd20f9d87010"=>[{:funder=>{:name=>"National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)", :identifier=>"00adh9b73", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}], "b93105d8-0142-4ddb-baa0-64fea8b981b2"=>[{:funder=>{:name=>"National Institute on Aging", :identifier=>"049v75w11", :scheme=>"ror"}, :award=>{:title=>"", :number=>" U2CAG057441", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"The Environmental Influences on Child Health Outcomes (ECHO)", :identifier=>"", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"Office of the Director, National Institutes of Health (OD)", :identifier=>"00fj8a872", :scheme=>"ror"}, :award=>{:title=>"", :number=>"U24OD023319", :identifier=>"", :scheme=>""}}, {:funder=>{:name=>"", :identifier=>"", :scheme=>"ror"}, :award=>{:title=>"", :number=>"", :identifier=>"", :scheme=>""}}]}

app/models/invenio_rdm_record_converter.rb

@@ -15,6 +15,7 @@ class InvenioRdmRecordConverter < Sufia::Export::Converter
   def initialize(generic_file=nil)
     return unless generic_file
     @@header_lookup ||= HeaderLookup.new
+    @@funding_data ||= eval(File.read('app/models/concerns/galtersufia/generic_file/funding_data.txt'))
 
     @record = record_for_export(generic_file)
     @file = filename_and_content_path(generic_file)
@@ -114,7 +115,8 @@ def invenio_metadata(gf)
       "subjects": SUBJECT_SCHEMES.map{ |subject_type| subjects_for_scheme(gf.send(subject_type), subject_type) }.flatten,
       "dates": gf.date_created.map{ |date| {"date": date, "type": "other", "description": "When the item was originally created."} },
       "formats": gf.mime_type,
-      "locations": gf.based_near.present? ? gf.based_near.shift.split("', ").map{ |location| {place: location.gsub("'", "")} } : {}
+      "locations": gf.based_near.present? ? gf.based_near.shift.split("', ").map{ |location| {place: location.gsub("'", "")} } : {},
+      "funding": funding(gf.id)
     }
   end
 
@@ -200,4 +202,8 @@ def additional_descriptions(descriptions)
     return nil if additional_descriptions_size < 0
     descriptions.last(additional_descriptions_size).map{ |add_desc| {"description": add_desc, "type": "other", "lang": "eng"} }
   end
+
+  def funding(file_id)
+    @@funding_data[file_id] || {}
+  end
 end

funding_csv_to_hash.rb

@@ -0,0 +1,45 @@
+require 'csv'
+
+funding_csv = CSV.read("prod_funding_sources.csv")
+funding_csv.shift
+formatted_funding_data = funding_csv.each_with_object({}) do |row, hash|
+  # name, ror-identifier, award-number, file-id
+  name = row[0]
+  identifier = row[1]
+  award_number = row[2]
+  file_id = row[3]
+
+  if hash.keys.include?(file_id)
+    hash[file_id] +=
+      [{
+        "funder": {
+          "name": "#{name}",
+          "identifier": "#{identifier}",
+          "scheme": "ror"
+        },
+        "award": {
+          "title": "", # always blank
+          "number": "#{award_number}",
+          "identifier": "", # always blank
+          "scheme": "" # always blank
+        }
+      }]
+  else
+    hash[file_id] =
+      [{
+        "funder": {
+          "name": "#{name}",
+          "identifier": "#{identifier}",
+          "scheme": "ror"
+        },
+        "award": {
+          "title": "", # always blank
+          "number": "#{award_number}",
+          "identifier": "", # always blank
+          "scheme": "" # always blank
+        }
+      }]
+  end
+end
+
+File.write("app/models/concerns/galtersufia/generic_file/funding_data.txt", formatted_funding_data)