uc3-dmp-id 0.0.139 → 0.1.0

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

@@ -9,30 +9,37 @@ module Uc3DmpId
   class Finder
     MSG_INVALID_ARGS = 'Expected JSON to be structured as `{ "dmp": { "PK": "value"} } OR \
                         { "dmp": { "dmp_id": { "identifier": "value", "type": "value" } }`'
-    MSG_INVALID_OWNER_ID = 'Invalid :owner_orcid. Expected value to start with `https://orcid.org/`.'
-    MSG_INVALID_OWNER_ORG = 'Invalid :owner_org_ror. Expected value to start with `https://ror.org/`.'
+    MSG_INVALID_OWNER_ID = 'Invalid :owner_orcid. Expected a valid ORCID id (excluding the domain)`.'
+    MSG_INVALID_OWNER_ORG = 'Invalid :owner_org_ror. Expected a valid ROR id (excluding the domain)`.'
     MSG_INVALID_MOD_DATE = 'Invalid :modification_day. Expected value to be in the `YYYY-MM-DD` format.'
     MSG_MISSING_PK = 'No PK was provided'
     MSG_MISSING_PROV_ID = 'No Provenance identifier was provided. \
                            Expected: `{ "dmp_id": { "identifier": "value", "type": "value" }`'
 
-
     class << self
       # TODO: Replace this with ElasticSearch
       def search_dmps(args:, logger: nil)
-        return _by_owner(owner_org: args['owner_orcid'], logger: logger) unless args['owner_orcid'].nil?
-        return _by_owner_org(owner_org: args['owner_org_ror'], logger: logger) unless args['owner_org_ror'].nil?
-        return _by_mod_day(day: args['modification_day'], logger: logger) unless args['modification_day'].nil?
+        client = Uc3DmpDynamo::Client.new
+        return _by_owner(owner_org: args['owner_orcid'], client: client, logger: logger) unless args['owner_orcid'].nil?
+
+        unless args['owner_org_ror'].nil?
+          return _by_owner_org(owner_org: args['owner_org_ror'], client: client,
+                               logger: logger)
+        end
+        unless args['modification_day'].nil?
+          return _by_mod_day(day: args['modification_day'], client: client,
+                             logger: logger)
+        end
 
         []
       end
-      # rubocop:enable Metrics/MethodLength
 
       # Find a DMP based on the contents of the incoming JSON
       # -------------------------------------------------------------------------
-      def by_json(json:, cleanse: true, logger: nil)
+      # rubocop:disable Metrics/AbcSize
+      def by_json(json:, client: nil, cleanse: true, logger: nil)
         json = Helper.parse_json(json: json)&.fetch('dmp', {})
-        raise FinderError, MSG_INVALID_ARGS if json.nil? || (json['PK'].nil? && json['dmp_id'].nil?)
+        raise FinderError, MSG_INVALID_ARGS if !json.is_a?(Hash) || (json['PK'].nil? && json['dmp_id'].nil?)
 
         p_key = json['PK']
         # Translate the incoming :dmp_id into a PK
@@ -46,9 +53,11 @@ module Uc3DmpId
         # find_by_PK
         p_key.nil? ? nil : by_pk(p_key: p_key, s_key: json['SK'], client: client, cleanse: cleanse, logger: logger)
       end
+      # rubocop:enable Metrics/AbcSize
 
       # Find the DMP by its PK and SK
       # -------------------------------------------------------------------------
+      # rubocop:disable Metrics/AbcSize
       def by_pk(p_key:, s_key: Helper::DMP_LATEST_VERSION, client: nil, cleanse: true, logger: nil)
         raise FinderError, MSG_MISSING_PK if p_key.nil?
 
@@ -57,7 +66,7 @@ module Uc3DmpId
         resp = client.get_item(
           key: {
             PK: Helper.append_pk_prefix(p_key: p_key),
-            SK: s_key
+            SK: Helper.append_sk_prefix(s_key: s_key)
           },
           logger: logger
         )
@@ -66,9 +75,10 @@ module Uc3DmpId
         dmp = resp['dmp'].nil? ? JSON.parse({ dmp: resp }.to_json) : resp
         return nil if dmp['dmp']['PK'].nil?
 
-        dmp = Versioner.append_versions(p_key: dmp['dmp']['PK'], dmp: dmp, client: client, logger: logger)
+        dmp = Versioner.append_versions(p_key: dmp['dmp']['PK'], dmp: dmp, client: client, logger: logger) if cleanse
         cleanse ? Helper.cleanse_dmp_json(json: dmp) : dmp
       end
+      # rubocop:enable Metrics/AbcSize
 
       # Fetch just the PK to see if a record exists
       # -------------------------------------------------------------------------
@@ -79,7 +89,7 @@ module Uc3DmpId
         client.pk_exists?(
           key: {
             PK: Helper.append_pk_prefix(p_key: p_key),
-            SK: s_key
+            SK: Helper.append_sk_prefix(s_key: s_key)
           },
           logger: logger
         )
@@ -89,7 +99,10 @@ module Uc3DmpId
       # -------------------------------------------------------------------------
       # rubocop:disable Metrics/AbcSize
       def by_provenance_identifier(json:, client: nil, cleanse: true, logger: nil)
-        raise FinderError, MSG_MISSING_PROV_ID if json.nil? || json.fetch('dmp_id', {})['identifier'].nil?
+        raise FinderError, MSG_MISSING_PROV_ID unless json.is_a?(Hash)
+
+        json = json['dmp'] unless json['dmp'].nil?
+        raise FinderError, MSG_MISSING_PROV_ID if json.fetch('dmp_id', {})['identifier'].nil?
 
         args = {
           index_name: 'dmphub_provenance_identifier_gsi',
@@ -117,17 +130,17 @@ 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:, logger: nil)
-        regex = %r{^([0-9A-Z]{4}-){3}[0-9A-Z]{4}$}
-        raise FinderError, MSG_INVALID_OWNER_ID if owner_id.nil? || (owner_id.to_s.downcase =~  regex).nil?
+      def _by_owner(owner_id:, client: nil, logger: nil)
+        regex = /^([0-9A-Z]{4}-){3}[0-9A-Z]{4}$/
+        raise FinderError, MSG_INVALID_OWNER_ID if owner_id.nil? || (owner_id.to_s =~ regex).nil?
 
         args = {
           index_name: 'dmphub_owner_id_gsi',
           key_conditions: {
             dmphub_owner_id: {
               attribute_value_list: [
-                "http://orcid.org/#{owner_id.to_s.downcase}",
-                "https://orcid.org/#{owner_id.to_s.downcase}"
+                "http://orcid.org/#{owner_id}",
+                "https://orcid.org/#{owner_id}"
               ],
               comparison_operator: 'IN'
             }
@@ -135,35 +148,39 @@ module Uc3DmpId
           filter_expression: 'SK = :version',
           expression_attribute_values: { ':version': Helper::DMP_LATEST_VERSION }
         }
-        logger.info(message: "Querying _by_owner with #{args}") unless logger.nil?
+        logger.info(message: "Querying _by_owner with #{args}") if logger.respond_to?(:info)
         client = client.nil? ? Uc3DmpDynamo::Client.new : client
         _process_search_response(response: client.query(args: args, logger: logger))
       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:, logger: nil)
-        regex = %r{^[a-zA-Z0-9]+$}
-        raise FinderError, MSG_INVALID_OWNER_ID if owner_org.nil? ||(owner_org.to_s.downcase =~ regex).nil?
+      # Fetch the DMP IDs for the specified organization/institution (the org is the :dmphub_owner_org
+      # on the DMP ID record)
+      def _by_owner_org(owner_org:, client: nil, logger: nil)
+        regex = /^[a-zA-Z0-9]+$/
+        raise FinderError, MSG_INVALID_OWNER_ID if owner_org.nil? || (owner_org.to_s.downcase =~ regex).nil?
 
         args = {
           index_name: 'dmphub_owner_org_gsi',
           key_conditions: {
             dmphub_owner_org: {
-              attribute_value_list: ["https://ror.org/#{owner_org.to_s.downcase}"],
-              comparison_operator: 'EQ'
+              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.info(message: "Querying _by_owner_org with #{args}") unless logger.nil?
+        logger.info(message: "Querying _by_owner_org with #{args}") if logger.respond_to?(:info)
         client = client.nil? ? Uc3DmpDynamo::Client.new : client
         _process_search_response(response: client.query(args: args, logger: logger))
       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:, logger: nil)
-        regex = %r{^[0-9]{4}(-[0-9]{2}){2}}
+      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 = {
@@ -177,17 +194,18 @@ module Uc3DmpId
           filter_expression: 'SK = :version',
           expression_attribute_values: { ':version': Helper::DMP_LATEST_VERSION }
         }
-        logger.info(message: "Querying _by_mod_day with #{args}") unless logger.nil?
+        logger.info(message: "Querying _by_mod_day with #{args}") if logger.respond_to?(:info)
         client = client.nil? ? Uc3DmpDynamo::Client.new : client
         _process_search_response(response: client.query(args: args, logger: logger))
       end
 
-
       # Transform the search results so that we do not include any of the DMPHub specific metadata
       def _process_search_response(response:)
         return [] unless response.is_a?(Array) && response.any?
 
-        results = response.each do |item|
+        results = response.map do |item|
+          next if item.nil?
+
           dmp = item['dmp'].nil? ? JSON.parse({ dmp: item }.to_json) : item
           Helper.cleanse_dmp_json(json: dmp)
         end

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

@@ -1,10 +1,8 @@
 # frozen_string_literal: true
 
-
 # TODO: Be sure to update the API functions so that they call cleanse_dmp_json before
 #       calling Uc3DmpApiCore::Responder.respond !!!!!!!!!!
 
-
 module Uc3DmpId
   # Helper functions for working with DMP IDs
   class Helper
@@ -21,6 +19,24 @@ module Uc3DmpId
     DMP_LATEST_VERSION = "#{SK_DMP_PREFIX}latest"
     DMP_TOMBSTONE_VERSION = "#{SK_DMP_PREFIX}tombstone"
 
+    DEFAULT_API_URL = 'https://api.dmphub.uc3dev.cdlib.net/dmps/'
+    DEFAULT_LANDING_PAGE_URL = 'https://dmphub.uc3dev.cdlib.net/dmps/'
+
+    MSG_DMP_EXISTS = 'DMP already exists. Try :update instead.'
+    MSG_DMP_FORBIDDEN = 'You do not have permission.'
+    MSG_DMP_INVALID_DMP_ID = 'Invalid DMP ID format.'
+    MSG_DMP_NO_DMP_ID = 'A DMP ID could not be registered at this time.'
+    MSG_DMP_NO_HISTORICALS = 'You cannot modify a historical version of the DMP.'
+    MSG_DMP_NO_TOMBSTONE = 'Unable to tombstone the DMP ID at this time.'
+    MSG_DMP_NO_UPDATE = 'Unable to update the DMP ID at this time.'
+    MSG_DMP_NOT_FOUND = 'DMP does not exist.'
+    MSG_DMP_UNABLE_TO_VERSION = 'Unable to version this DMP.'
+    MSG_DMP_UNKNOWN = 'DMP does not exist. Try :create instead.'
+    MSG_NO_CHANGE = 'The updated record has no changes.'
+    MSG_NO_OWNER_ORG = 'Could not determine ownership of the DMP ID.'
+    MSG_NO_PROVENANCE_OWNER = 'Unable to determine the provenance of the DMP ID.'
+    MSG_SERVER_ERROR = 'Something went wrong.'
+
     class << self
       # Append the PK prefix for the object
       # -------------------------------------------------------------------------------------
@@ -48,19 +64,13 @@ module Uc3DmpId
 
       # Return the base URL for a DMP ID
       def dmp_id_base_url
-        url = ENV.fetch('DMP_ID_BASE_URL', 'https://dmphub.uc3dev.cdlib.net/dmps/')
+        url = ENV.fetch('DMP_ID_BASE_URL', DEFAULT_LANDING_PAGE_URL)
         url&.end_with?('/') ? url : "#{url}/"
       end
 
       # The landing page URL (NOT the DOI URL)
       def landing_page_url
-        url = ENV.fetch('DMP_ID_LANDING_URL', 'https://dmphub.uc3dev.cdlib.net/dmps/')
-        url&.end_with?('/') ? url : "#{url}/"
-      end
-
-      # Return the base URL for the API
-      def api_base_url
-        url = ENV.fetch('DMP_ID_BASE_URL', 'https://api.dmphub.uc3dev.cdlib.net/dmps/')
+        url = ENV.fetch('DMP_ID_LANDING_URL', DEFAULT_LANDING_PAGE_URL)
         url&.end_with?('/') ? url : "#{url}/"
       end
 
@@ -69,7 +79,7 @@ module Uc3DmpId
         dmp_id = value.match(DOI_REGEX).to_s
         return nil if dmp_id.nil? || dmp_id == ''
         # If it's already a URL, return it as is
-        return value if value.start_with?('http')
+        return with_protocol ? value : value.gsub(%r{https?://}, '') if value.start_with?('http')
 
         dmp_id = dmp_id.gsub('doi:', '')
         dmp_id = dmp_id.start_with?('/') ? dmp_id[1..dmp_id.length] : dmp_id
@@ -90,7 +100,7 @@ module Uc3DmpId
 
       # Append the :PK prefix to the :dmp_id
       def dmp_id_to_pk(json:)
-        return nil if json.nil? || json['identifier'].nil?
+        return nil if !json.is_a?(Hash) || json['identifier'].nil?
 
         # If it's a DOI format it correctly
         dmp_id = format_dmp_id(value: json['identifier'].to_s)
@@ -117,6 +127,7 @@ module Uc3DmpId
       end
 
       # Compare the DMP IDs to see if they are the same
+      # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
       def eql?(dmp_a:, dmp_b:)
         return dmp_a == dmp_b unless dmp_a.is_a?(Hash) && !dmp_a['dmp'].nil? && dmp_b.is_a?(Hash) && !dmp_b['dmp'].nil?
 
@@ -132,9 +143,9 @@ module Uc3DmpId
         b = deep_copy_dmp(obj: dmp_b)
 
         # ignore some of the attributes before comparing
-        %w[SK dmphub_modification_day modified created dmphub_assertions].each do |key|
-        a['dmp'].delete(key) unless a['dmp'][key].nil?
-        b['dmp'].delete(key) unless b['dmp'][key].nil?
+        %w[SK dmphub_modification_day modified created dmphub_versions].each do |key|
+          a['dmp'].delete(key) unless a['dmp'][key].nil?
+          b['dmp'].delete(key) unless b['dmp'][key].nil?
         end
         a == b
       end
@@ -147,7 +158,7 @@ module Uc3DmpId
         id = dmp.fetch('contact', {}).fetch('contact_id', {})['identifier']
         return id unless id.nil?
 
-        dmp.fetch('contributor', []).map { |contributor| contributor.fetch('contact_id', {})['identifier'] }.first
+        dmp.fetch('contributor', []).map { |contributor| contributor.fetch('contributor_id', {})['identifier'] }.first
       end
 
       # Extract the Contact's affiliaiton ROR ID
@@ -155,7 +166,8 @@ module Uc3DmpId
         return nil unless json.is_a?(Hash)
 
         dmp = json['dmp'].nil? ? json : json['dmp']
-        owner_org = dmp.fetch('contact', {}).fetch('dmproadmap_affiliation', {}).fetch('affiliation_id', {})['identifier']
+        owner_org = dmp.fetch('contact', {}).fetch('dmproadmap_affiliation', {}).fetch('affiliation_id',
+                                                                                       {})['identifier']
         return owner_org unless owner_org.nil?
 
         orgs = dmp.fetch('contributor', []).map do |contributor|
@@ -163,8 +175,10 @@ module Uc3DmpId
         end
         orgs.compact.max_by { |i| orgs.count(i) }
       end
+      # rubocop:enable Metrics/AbcSize
 
       # Add DMPHub specific fields to the DMP ID JSON
+      # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
       def annotate_dmp_json(provenance:, p_key:, json:)
         json = parse_json(json: json)
         bool_vals = [1, '1', true, 'true', 'yes']
@@ -179,7 +193,7 @@ module Uc3DmpId
         annotated['SK'] = DMP_LATEST_VERSION
 
         # Ensure that the :dmp_id matches the :PK
-        annotated['dmp_id'] = pk_to_dmp_id(p_key: remove_pk_prefix(p_key: annotated['PK']))
+        annotated['dmp_id'] = JSON.parse(pk_to_dmp_id(p_key: remove_pk_prefix(p_key: annotated['PK'])).to_json)
 
         owner_id = extract_owner_id(json: json)
         owner_org = extract_owner_org(json: json)
@@ -189,7 +203,7 @@ module Uc3DmpId
         annotated['dmproadmap_featured'] = bool_vals.include?(featured.to_s.downcase) ? '1' : '0'
 
         # Update the modification timestamps
-        annotated['dmphub_modification_day'] = Time.now.strftime('%Y-%m-%d')
+        annotated['dmphub_modification_day'] = Time.now.utc.strftime('%Y-%m-%d')
         annotated['dmphub_owner_id'] = owner_id unless owner_id.nil?
         annotated['dmphub_owner_org'] = owner_org unless owner_org.nil?
         return annotated unless json['dmphub_provenance_id'].nil?
@@ -210,6 +224,8 @@ module Uc3DmpId
         end
         annotated
       end
+      # rubocop:enable Metrics/AbcSize, Metrics/MethodLength
+      # rubocop:enable Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
 
       # Recursive method that strips out any DMPHub related metadata from a DMP record before sending
       # it to the caller
@@ -220,7 +236,7 @@ module Uc3DmpId
         return json.map { |obj| cleanse_dmp_json(json: obj) }.compact if json.is_a?(Array)
 
         cleansed = {}
-        allowable = %w[dmphub_versions]
+        allowable = %w[dmphub_modifications dmphub_versions]
         json.each_key do |key|
           next if (key.to_s.start_with?('dmphub') && !allowable.include?(key)) || %w[PK SK].include?(key.to_s)
 
@@ -289,8 +305,8 @@ module Uc3DmpId
 
         # Remove the homepage or callbackUri because we will add this when needed. we just want the id
         val = value.downcase
-                  .gsub(provenance.fetch('callbackUri', '').downcase, '')
-                  .gsub(provenance.fetch('homepage', '').downcase, '')
+                   .gsub(provenance.fetch('callbackUri', '').downcase, '')
+                   .gsub(provenance.fetch('homepage', '').downcase, '')
         val = val.gsub(%r{https?://}, '')
         val = val[1..val.length] if val.start_with?('/')
         id = provenance['PK']&.gsub('PROVENANCE#', '')