uc3-dmp-id 0.1.85 → 0.1.87

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 48f18668e2920a15106853f5182786aa3ff7167a81d0ba7267fef83cd77f35fc
-  data.tar.gz: 7648b3a8d11c912262d026047bd3bf02438cdf348317288cef0be47078c2cfc0
+  metadata.gz: a5f7760c4099466f5a99377669e189dfc33cfe2e718d1bf0c96b06b251f5ff77
+  data.tar.gz: dd6b30e66d7848e8810bd4c735ce5d007e7c0dd80c059e2341813db873e74c35
 SHA512:
-  metadata.gz: 72b078e29a969a2bfa86a60a307f3cfcebdc196cc04f0361523b87ade76bf325abcddc9ba0aa7b7776474843c759b2926411f095f4d7426407a77a752f1b28bb
-  data.tar.gz: 213acbf3eb5d81c578ebdafdc9f8c4561007589d0a73afaa15083be3011a6710a7122ca5695b012614525c67ee116db8e220d382afaa9552b67b26922465fae6
+  metadata.gz: 9b2ebf8b2d1919f15bc9b538373fcafddb077c432754b0ba27c8706de6352db07d0db51ec41e5bb4bdc1469a55d45162d45127fc14e4df0c4699d63ed10963e5
+  data.tar.gz: 6470cf64e86138c08c343c299dea73a5d5add92755c6716684ad33e25378bfb36f6c3e440a142d5c66f439655d4b3600516c417efd31c9dc7ae4dac0047e416b

data/lib/uc3-dmp-id/creator.rb

@@ -53,7 +53,7 @@ module Uc3DmpId
         annotated['registered'] = annotated['created'] if annotated['registered'].nil?
 
         # Create the item
-        annotated.delete('dmphub_modifications')
+        annotated['dmphub_modifications'] = []
         resp = client.put_item(json: annotated, logger:)
         raise CreatorError, Helper::MSG_DMP_NO_DMP_ID if resp.nil?
 

data/lib/uc3-dmp-id/finder.rb

@@ -17,22 +17,43 @@ module Uc3DmpId
     MSG_MISSING_PROV_ID = 'No Provenance identifier was provided. \
                            Expected: `{ "dmp_id": { "identifier": "value", "type": "value" }`'
 
+    ORCID_DOMAIN = 'https://orcid.org/'
+    ROR_DOMAIN = 'https://ror.org/'
+    DOI_DOMAIN = 'https://doi.org/'
+    SORT_OPTIONS = %w[title modified]
+    SORT_DIRECTIONS = %w[asc desc]
+    MAX_PAGE_SIZE = 100
+    DEFAULT_PAGE_SIZE = 25
+    DEFAULT_SORT_OPTION = 'modified'
+    DEFAULT_SORT_DIR = 'desc'
+
     class << self
       # TODO: Replace this with ElasticSearch
       def search_dmps(args:, logger: nil)
-        client = Uc3DmpDynamo::Client.new
-        return _by_owner(owner_org: args['owner_orcid'], client:, logger:) unless args['owner_orcid'].nil?
+        # Fetch the DMPs for each of the possible filter options
+        client = Uc3DmpDynamo::Client.new(table: ENV['DYNAMO_INDEX_TABLE'])
+        owner = args['owner']
+        org = args['org']
+        funder = args['funder']
+
+        owner_pks = owner.nil? ? [] : _by_owner(owner: owner, client:, logger:)
+        # There may be multiple Org ids, so query them all
+        org_pks = org.nil? ? [] : org.split('|').map { |o| _by_org(org: o, client:, logger:) }
+        org_pks = org_pks.flatten.uniq
+        funder_pks = funder.nil? ? [] : _by_funder(funder: funder, client:, logger:)
+        # pks = [owner_pks, org_pks, funder_pks].reject(&:empty?)
+        logger&.debug(
+          message: 'PKs found',
+          details: { owner: owner_pks, org: org_pks, funder: funder_pks }
+        )
+        # return [] if pks.nil? || pks.empty?
 
-        unless args['owner_org_ror'].nil?
-          return _by_owner_org(owner_org: args['owner_org_ror'], client:,
-                               logger:)
-        end
-        unless args['modification_day'].nil?
-          return _by_mod_day(day: args['modification_day'], client:,
-                             logger:)
-        end
+        # Only use the DMPs that fit all of the filter criteria
+        # dmps = pks.reduce(:&).flatten.uniq
+        # return [] if dmps.nil? || dmps.empty?
+
+        [owner_pks, org_pks, funder_pks].flatten.uniq
 
-        []
       end
 
       # Find a DMP based on the contents of the incoming JSON
@@ -134,74 +155,68 @@ module Uc3DmpId
 
       private
 
-      # Fetch the DMP IDs for the specified owner's ORCID (the owner is the :dmphub_owner_id on the DMP ID record)
-      def _by_owner(owner_id:, client: nil, logger: nil)
-        regex = /^([0-9A-Z]{4}-){3}[0-9A-Z]{4}$/
-        raise FinderError, MSG_INVALID_OWNER_ID if owner_id.nil? || (owner_id.to_s =~ regex).nil?
+      # Fetch the DMP IDs for the specified person's ORCID (or email)
+      def _by_owner(owner:, client: nil, logger: nil)
+        orcid_regex = /^([0-9a-zA-Z]{4}-){3}[0-9a-zA-Z]{4}$/
+        email_regex = /^[^\s@]+@[^\s@]+\.[^\s@]+$/
+        orcid = owner.to_s.strip
+        return [] if (orcid =~ orcid_regex).nil? && (orcid =~ email_regex).nil?
 
-        args = {
-          index_name: 'dmphub_owner_id_gsi',
-          key_conditions: {
-            dmphub_owner_id: {
-              attribute_value_list: [
-                "http://orcid.org/#{owner_id}",
-                "https://orcid.org/#{owner_id}"
-              ],
-              comparison_operator: 'IN'
-            }
-          },
-          filter_expression: 'SK = :version',
-          expression_attribute_values: { ':version': Helper::DMP_LATEST_VERSION }
-        }
-        logger.info(message: "Querying _by_owner with #{args}") if logger.respond_to?(:info)
-        client = Uc3DmpDynamo::Client.new if client.nil?
-        _process_search_response(response: client.query(args:, logger:))
+        orcid = "#{ORCID_DOMAIN}#{orcid}" unless (orcid =~ orcid_regex).nil?
+        resp = client.get_item(key: { PK: 'PERSON_INDEX', SK: orcid }, logger:)
+        return [] unless resp.is_a?(Hash)
+
+        logger&.debug(message: "DMPs for PERSON #{orcid}", details: resp)
+        resp.fetch('dmps', [])
       end
 
-      # Fetch the DMP IDs for the specified organization/institution (the org is the :dmphub_owner_org
-      # on the DMP ID record)
-      def _by_owner_org(owner_org:, client: nil, logger: nil)
+      # Fetch the DMP IDs for the specified organization/institution
+      def _by_org(org:, client: nil, logger: nil)
         regex = /^[a-zA-Z0-9]+$/
-        raise FinderError, MSG_INVALID_OWNER_ID if owner_org.nil? || (owner_org.to_s.downcase =~ regex).nil?
+        id = "#{ROR_DOMAIN}#{org.strip}" unless (org.to_s =~ regex).nil?
+        return [] if id.nil?
+
+        resp = client.get_item(key: { PK: 'AFFILIATION_INDEX', SK: id }, logger:)
+        return [] unless resp.is_a?(Hash)
+
+        logger&.debug(message: "DMPs for AFFILIATION #{id}", details: resp)
+        resp.fetch('dmps', [])
+      end
 
+      # Fetch the DMP IDs for the specified funder
+      def _by_funder(funder:, client: nil, logger: nil)
+        regex = /^[a-zA-Z0-9]+$/
+        id = "#{ROR_DOMAIN}/#{funder.strip}" unless (funder.to_s =~ regex).nil?
+        id = "#{DOI_DOMAIN}#{funder.strip}" if id.nil? && !(funder.to_s =~ Helper::DOI_REGEX).nil?
+        return [] if id.nil?
+
+        resp = client.get_item(key: { PK: 'FUNDER_INDEX', SK: id }, logger:)
+        return [] unless resp.is_a?(Hash)
+
+        logger&.debug(message: "DMPs for FUNDER #{id}", details: resp)
+        resp.fetch('dmps', [])
+      end
+
+      # Fetch the DMP IDs that are marked as featured
+      def _by_featured(client: nil, logger: nil)
         args = {
-          index_name: 'dmphub_owner_org_gsi',
-          key_conditions: {
-            dmphub_owner_org: {
-              attribute_value_list: [
-                "https://ror.org/#{owner_org.to_s.downcase}",
-                "http://ror.org/#{owner_org.to_s.downcase}"
-              ],
-              comparison_operator: 'IN'
-            }
-          },
-          filter_expression: 'SK = :version',
-          expression_attribute_values: { ':version': Helper::DMP_LATEST_VERSION }
+          filter_expression: 'featured = :featured AND SK = :sk',
+          expression_attribute_values: { ':sk': 'METADATA', ':featured': 1 }
         }
-        logger.info(message: "Querying _by_owner_org with #{args}") if logger.respond_to?(:info)
+        logger&.debug(message: 'Fetch relevant DMPs _by_featured - scan args', details: args)
         client = Uc3DmpDynamo::Client.new if client.nil?
-        _process_search_response(response: client.query(args:, logger:))
+        _process_search_response(response: client.scan(args:))
       end
 
-      # Fetch the DMP IDs modified on the specified date (the date is the :dmphub_modification_day on the DMP ID record)
-      def _by_mod_day(day:, client: nil, logger: nil)
-        regex = /^[0-9]{4}(-[0-9]{2}){2}/
-        raise FinderError, MSG_INVALID_OWNER_ID if day.nil? || (day.to_s =~ regex).nil?
-
+      # Return all of the publicly visible DMPs
+      def _publicly_visible(client: nil, logger: nil)
         args = {
-          index_name: 'dmphub_modification_day_gsi',
-          key_conditions: {
-            dmphub_modification_day: {
-              attribute_value_list: [day.to_s],
-              comparison_operator: 'IN'
-            }
-          },
-          filter_expression: 'SK = :version',
-          expression_attribute_values: { ':version': Helper::DMP_LATEST_VERSION }
+          filter_expression: 'visibility = :visibility AND SK = :sk',
+          expression_attribute_values: { ':sk': 'METADATA', ':visibility': 'public' }
         }
-        logger.info(message: "Querying _by_mod_day with #{args}") if logger.respond_to?(:info)
+        logger&.debug(message: 'Fetch relevant DMPs _publicly_visible - scan args', details: args)
         client = Uc3DmpDynamo::Client.new if client.nil?
-        _process_search_response(response: client.query(args:, logger:))
+        _process_search_response(response: client.scan(args:))
       end
 
       # Transform the search results so that we do not include any of the DMPHub specific metadata
@@ -212,8 +227,8 @@ module Uc3DmpId
           next if item.nil?
 
           dmp = item['dmp'].nil? ? JSON.parse({ dmp: item }.to_json) : item
-          dmp = _remove_narrative_if_private(json: dmp)
-          Helper.cleanse_dmp_json(json: dmp)
+          # dmp = _remove_narrative_if_private(json: dmp)
+          # Helper.cleanse_dmp_json(json: dmp)
         end
         results.compact.uniq
       end

data/lib/uc3-dmp-id/updater.rb

@@ -112,8 +112,7 @@ module Uc3DmpId
 
       def _purge_mods(payload:)
         # Ensure we've deleted the dmphub_mods!!!!
-        payload.delete('dmphub_modifications')
-        payload.delete(:dmphub_modifications)
+        payload['dmphub_modifications'] = []
         payload
       end
 

data/lib/uc3-dmp-id/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Uc3DmpId
-  VERSION = '0.1.85'
+  VERSION = '0.1.87'
 end

data/lib/uc3-dmp-id/versioner.rb

@@ -58,7 +58,7 @@ module Uc3DmpId
 
         # Create the prior version record ()
         client = Uc3DmpDynamo::Client.new if client.nil?
-        prior.delete('dmphub_modifications')
+        prior['dmphub_modifications'] = []
         resp = client.put_item(json: prior, logger:)
         return nil if resp.nil?
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: uc3-dmp-id
 version: !ruby/object:Gem::Version
-  version: 0.1.85
+  version: 0.1.87
 platform: ruby
 authors:
 - Brian Riley
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-07-16 00:00:00.000000000 Z
+date: 2024-07-17 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: json