uc3-dmp-id 0.1.81 → 0.1.82

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3de2c7edd72200592a63d6bad11a19ed289c9f3dfa35272d82ff98ec666b89ab
-  data.tar.gz: 7b887212a1babb8f7928ba8bc3062c4c6a282d29af346c36d5e0f61083e67356
+  metadata.gz: 9c9845166f4bdaebc57491464cf46037645a6fccd2cc3274aa988ee6c7a735ad
+  data.tar.gz: 39d2e885ac1b4a8fd95796464d117c6a86b060e4842b99814dc30972160dc9ce
 SHA512:
-  metadata.gz: 8685a865bb59f270cced7d0ffab66cd12a87a03a2d078accb2676a4d28ba73f13cf3d728b829e7e207aecddaa4cc10a53257414cb53c8e13f8cd0559b34e1210
-  data.tar.gz: 13886474742a157b539d70b7e77961ab2a5272818ab0bd6df519ea11b7a6d18720b1f4268ec43ee532f3e4493bd92b9a5b31485d0530b38e6450972a821ef429
+  metadata.gz: d850ae3e53fb18bae12856a45f2fda95d66664f70052ed6f9c33e047de47728a5936012529a8727c59ccd671d687c6834d6880a97f915c9e4538d282ec3ea320
+  data.tar.gz: 2fd983b7d683a5da85f21a1f7001d5f8b1f7a47e798f5fffcb9db98e3486890d6982746d7a65b8d30377aa7eab5c8fcde377ba9b4d31fdc2cfd633d4bcb9fcb1

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

@@ -17,43 +17,22 @@ 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)
-        # 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?
-
-        # Only use the DMPs that fit all of the filter criteria
-        # dmps = pks.reduce(:&).flatten.uniq
-        # return [] if dmps.nil? || dmps.empty?
+        client = Uc3DmpDynamo::Client.new
+        return _by_owner(owner_org: args['owner_orcid'], client:, logger:) unless args['owner_orcid'].nil?
 
-        [owner_pks, org_pks, funder_pks].flatten.uniq
+        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
 
+        []
       end
 
       # Find a DMP based on the contents of the incoming JSON
@@ -155,68 +134,74 @@ module Uc3DmpId
 
       private
 
-      # 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?
+      # 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?
 
-        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
-      def _by_org(org:, client: nil, logger: nil)
-        regex = /^[a-zA-Z0-9]+$/
-        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', [])
+        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:))
       end
 
-      # Fetch the DMP IDs for the specified funder
-      def _by_funder(funder:, client: nil, logger: nil)
+      # 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)
         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
+        raise FinderError, MSG_INVALID_OWNER_ID if owner_org.nil? || (owner_org.to_s.downcase =~ regex).nil?
 
-      # Fetch the DMP IDs that are marked as featured
-      def _by_featured(client: nil, logger: nil)
         args = {
-          filter_expression: 'featured = :featured AND SK = :sk',
-          expression_attribute_values: { ':sk': 'METADATA', ':featured': 1 }
+          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 }
         }
-        logger&.debug(message: 'Fetch relevant DMPs _by_featured - scan args', details: args)
+        logger.info(message: "Querying _by_owner_org with #{args}") if logger.respond_to?(:info)
         client = Uc3DmpDynamo::Client.new if client.nil?
-        _process_search_response(response: client.scan(args:))
+        _process_search_response(response: client.query(args:, logger:))
       end
 
-      # Return all of the publicly visible DMPs
-      def _publicly_visible(client: nil, logger: nil)
+      # 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?
+
         args = {
-          filter_expression: 'visibility = :visibility AND SK = :sk',
-          expression_attribute_values: { ':sk': 'METADATA', ':visibility': 'public' }
+          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 }
         }
-        logger&.debug(message: 'Fetch relevant DMPs _publicly_visible - scan args', details: args)
+        logger.info(message: "Querying _by_mod_day with #{args}") if logger.respond_to?(:info)
         client = Uc3DmpDynamo::Client.new if client.nil?
-        _process_search_response(response: client.scan(args:))
+        _process_search_response(response: client.query(args:, logger:))
       end
 
       # Transform the search results so that we do not include any of the DMPHub specific metadata
@@ -227,8 +212,8 @@ 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

@@ -49,27 +49,21 @@ module Uc3DmpId
         logger&.debug(message: 'Version after process_harvester_mods', details: version)
         raise UpdaterError, Helper::MSG_SERVER_ERROR if version.nil?
 
-        # Remove the version info any any lingering modification blocks
-        version.delete('dmphub_versions')
-        version.delete('dmphub_modifications')
-
-        # Set the :modified timestamps
-        now = Time.now.utc
-        version['modified'] = now.iso8601
-        version['dmphub_modification_day'] = now.strftime('%Y-%m-%d')
+        # Process the incoming payload
+        payload = _process_modifications(owner:, updater:, version:, payload:, logger:)
 
         # Save the changes
-        resp = client.put_item(json: version, logger:)
+        resp = client.put_item(json: payload, logger:)
         raise UpdaterError, Helper::MSG_DMP_UNABLE_TO_VERSION if resp.nil?
 
         # Send the updates to EZID
-        _post_process(provenance:, json: version, logger:)
+        _post_process(provenance:, json: payload, logger:)
 
         # Return the new version record
         logger.info(message: "Updated DMP ID: #{p_key}") if logger.respond_to?(:debug)
 
         # Append the :dmphub_versions Array
-        out = JSON.parse({ dmp: version }.to_json)
+        out = JSON.parse({ dmp: payload }.to_json)
         out = Versioner.append_versions(p_key:, dmp: out, client:, logger:)
         Helper.cleanse_dmp_json(json: out)
       end
@@ -235,6 +229,40 @@ module Uc3DmpId
         logger&.debug(message: 'Returning updated VERSION:', details: version)
         version
       end
+
+      # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
+      def _process_modifications(owner:, updater:, version:, payload:, logger: nil)
+        return version unless payload.is_a?(Hash) && !updater.nil?
+        return payload unless version.is_a?(Hash) && !owner.nil?
+
+        logger.debug(message: 'Modifications before processing.', details: payload) if logger.respond_to?(:debug)
+
+        excluded_keys = %w[modified dmphub_modifications dmphub_versions]
+
+        # Always remove the dmphub_versions and dmphub_modifications
+        excluded_keys.each { |key| payload.delete(key) }
+
+        # Always include the PK and SK!
+        payload['PK'] = version['PK']
+        payload['SK'] = Helper::DMP_LATEST_VERSION
+
+        # Set the :modified timestamps
+        now = Time.now.utc
+        version['modified'] = now.iso8601
+        version['dmphub_modification_day'] = now.strftime('%Y-%m-%d')
+
+        # Retain all the other attributes on the original version unless they are
+        # in the list of exclusions OR the incoming payload already has a value for it
+        version.keys.each do |key|
+          next if excluded_keys.include?(key) || !payload[key].nil?
+
+          payload[key] = version[key]
+        end
+        logger.debug(message: 'Modifications after processing.', details: payload) if logger.respond_to?(:debug)
+        payload
+      end
+      # rubocop:enable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
+
     end
   end
 end

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

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Uc3DmpId
-  VERSION = '0.1.81'
+  VERSION = '0.1.82'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: uc3-dmp-id
 version: !ruby/object:Gem::Version
-  version: 0.1.81
+  version: 0.1.82
 platform: ruby
 authors:
 - Brian Riley