uc3-dmp-id 0.1.68 → 0.1.70

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 2f6ec06f994f12700ca9584e1a9f1bb42c5ff556e7ff6f1d9610c31dbd6cad2b
-  data.tar.gz: 32684f7693d73bc03c6eb7ea07906046895031d6aa3ec045c4c898fe04c03318
+  metadata.gz: ccb63f060db6bbbf29aacb329a163adcb1c3ee80ecd9eed385894b5d3f824380
+  data.tar.gz: 9caa229b0f0e270ac9da5160af0d1b36876cf95bbc5d4b9e2dc92a8a032c0406
 SHA512:
-  metadata.gz: 37817e517884f4fc2d79d4c63aa71bbdede1fe74b182212b84a6e17aa685ac4696415bdb2b7a3ad19ab4d55a83e1015f50dda730ca9f84cc125627b38c8606eb
-  data.tar.gz: 700da77d44d8302b2b4fdfd4d985ae4a0f48ffbed10a74b1924a6a7fd42f1eb7cf8bb722342c93a205d84eb70cfa65a2abe81e902e6be394df1e363e37f47977
+  metadata.gz: ba8bfca2d71f74097386240b43888e1f4fbe2d0443a881f5a264b0db475f224bfc7e96f9592f61b108d82a9570d15613a7d5b098d4495319c6fc601e40d194e5
+  data.tar.gz: e86673a1a825b4541ee868d5c7dac78d3cae186c328151c09e9ef6390bf687594ed36cd434d6f0ddd5a5b22cbd22c1627de0b5d6cbf465ca221f5c3b7b23d420

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

@@ -17,40 +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/'
-    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:)
-        org_pks = org.nil? ? [] : _by_org(org: org, client:, logger:)
-        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
@@ -152,65 +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]+$/
-        ror = "#{ROR_DOMAIN}/#{org.strip}" unless (org.to_s =~ regex).nil?
-
-        resp = client.get_item(key: { PK: 'AFFILIATION_INDEX', SK: ror }, logger:)
-        return [] unless resp.is_a?(Hash)
-
-        logger&.debug(message: "DMPs for AFFILIATION #{ror}", 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]+$/
-        ror = "#{ROR_DOMAIN}/#{funder.strip}" unless (funder.to_s =~ regex).nil?
-
-        resp = client.get_item(key: { PK: 'FUNDER_INDEX', SK: ror }, logger:)
-        return [] unless resp.is_a?(Hash)
-
-        logger&.debug(message: "DMPs for FUNDER #{ror}", 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
@@ -221,8 +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

@@ -44,13 +44,15 @@ module Uc3DmpId
         raise UpdaterError, Helper::MSG_DMP_FORBIDDEN if owner != updater
 
         # Handle any changes to the dmphub_modifications section
-        version = _process_harvester_mods(client:, p_key:, json: version, logger:)
+        version = _process_harvester_mods(client:, p_key:, json: mods, logger:)
 
         # Remove the version info because we don't want to save it on the record
         version.delete('dmphub_versions')
 
         # Splice the assertions
         version = _process_modifications(owner:, updater:, version:, mods:, logger:)
+        logger&.debug(message: 'Mods after process_modifications', details: version)
+
         # Set the :modified timestamps
         now = Time.now.utc
         version['modified'] = now.iso8601
@@ -192,6 +194,7 @@ module Uc3DmpId
 
       # Fetch any Harvester modifications to the JSON
       def _process_harvester_mods(client:, p_key:, json:, logger: nil)
+        logger&.debug(message: 'Incoming modifications', details: json)
         return json if json.fetch('dmphub_modifications', []).empty?
 
         # Fetch the `"SK": "HARVESTER_MODS"` record
@@ -201,6 +204,7 @@ module Uc3DmpId
         )
         return json unless resp.is_a?(Hash) && resp['related_works'].is_a?(Hash)
 
+        logger&.debug(message: 'Original HARVESTER_MODS record', details: resp)
         # The `dmphub_modifications` array will ONLY ever have things the harvester mods know about
         # so just find them and update the status accordingly
         mods = resp.dup
@@ -208,14 +212,18 @@ module Uc3DmpId
           next if entry.fetch('dmproadmap_related_identifiers', []).empty?
 
           entry['dmproadmap_related_identifiers'].each do |related|
-            next if mods['related_works'][related.identifier].nil?
+            related_id = mods['related_works'][related.identifier] if related.respond_to?(:identifier)
+            related_id = mods['related_works'][related['identifier']] if related_id.nil?
+            next if related_id.nil?
 
-            mods['related_works'][related.identifier]['status'] = related['status']
+            mods['related_works'][related_id]['status'] = related['status']
           end
         end
 
+        logger&.debug(message: 'Updating HARVESTER_MODS with:', details: mods)
         client.put_item(json: mods, logger:)
         json.delete('dmphub_modifications')
+        logger&.debug(message: 'After deleting dmphub_modifications:', details: json)
         json
       end
     end

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

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Uc3DmpId
-  VERSION = '0.1.68'
+  VERSION = '0.1.70'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: uc3-dmp-id
 version: !ruby/object:Gem::Version
-  version: 0.1.68
+  version: 0.1.70
 platform: ruby
 authors:
 - Brian Riley