dmp-dynamo_adapter 0.1.3 → 0.1.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c4c2a510b026b803c920fd6554a4196ca88ba8eaac1664adbe84205aac6f043a
-  data.tar.gz: c2ac05ece5c801f7119e2544b88c860666cd1fd7010debe24191490f22c2cb5a
+  metadata.gz: 8fe879d1aefc8db1c3b1a8160171c2f3cd8f6ee1986d494abf69083af19a6133
+  data.tar.gz: 58fd4512a6c72bfaee714cfb19be7c9dd3f9a81b2df1742f5a7a35ff6b5fa045
 SHA512:
-  metadata.gz: 2a5417148acb494de40019d4b323c5aae7f13c4c65699abdea86b47c6f98768a70f4cca8ac58bca6012ec92c23e6f4ef3c61c77f2c821bb0b6fcd64db4001282
-  data.tar.gz: dba9ee2fc94ef98acebce7dc4f1954c0936afe7fb07bc5365b27b173da1b29a63e7c94ee06982e60a18542adc436c7a9a510cec76ea6c924f339bd94f879f12c
+  metadata.gz: 52c93f6323e77810246cceeba743a71866872931716f82b1dfe482ff613d1361ef596deed64ff8f22eec32ea38c7679d0c64916c7e1941d076eed1357537ac37
+  data.tar.gz: 01f5d12e5227007fc434520e135b82edbbb4601a517c6a9898cf510fe2dcca9163900145fb50b19d69a40cdc8e3bf81fd345c0e0361d24df2a650ab15a9bf2cb

data/lib/dmp/dmp_id_handler.rb

@@ -0,0 +1,63 @@
+# frozen_string_literal: true
+
+require 'securerandom'
+require 'dmp/metadata_handler'
+
+module Dmp
+  # Methods that handle PK generation
+  class DmpIdHandler
+    DOI_REGEX = %r{[0-9]{2}\.[0-9]{5}/[a-zA-Z0-9/_.]+}.freeze
+
+    class << self
+      def dmp_id_base_url
+        ENV['DMP_ID_BASE_URL'].end_with?('/') ? ENV['DMP_ID_BASE_URL'] : "#{ENV['DMP_ID_BASE_URL']}/"
+      end
+
+      # Preassign a DMP ID that will leater be sent to the DOI minting authority (EZID)
+      def preregister_dmp_id
+        dmp_id = ''
+
+        counter = 0
+        while dmp_id == '' && counter <= 10
+          prefix = "#{ENV['DMP_ID_SHOULDER']}.#{SecureRandom.hex(4).upcase}"
+          dmp_id = prefix if find_by_pk(p_key: Dmp::MetadataHandler.append_pk_prefix(dmp: dmp_id)).empty?
+          counter += 1
+        end
+        # Something went wrong and it was unable to identify a unique id
+        return nil if counter >= 10
+
+        "#{Dmp::MetadataHandler::PK_DMP_PREFIX}#{dmp_id_base_url}#{dmp_id}"
+      end
+
+      # Format the DMP ID in the way we want it
+      def format_dmp_id(value:)
+        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')
+
+        dmp_id = dmp_id.gsub('doi:', '')
+        dmp_id = dmp_id.start_with?('/') ? dmp_id[1..dmp_id.length] : dmp_id
+        "#{dmp_id_base_url}#{dmp_id}"
+      end
+
+      # Append the :PK prefix to the :dmp_id
+      def dmp_id_to_pk(json:)
+        return nil if json.nil? || json['identifier'].nil?
+
+        # If it's a DOI format it correctly
+        dmp_id = format_dmp_id(value: json['identifier'].to_s)
+        return nil if dmp_id.nil? || dmp_id == ''
+
+        Dmp::MetadataHandler.append_pk_prefix(dmp: dmp_id)
+      end
+
+      # Derive the DMP ID by removing the :PK prefix
+      def pk_to_dmp_id(p_key:)
+        return nil if p_key.nil?
+
+        { type: 'doi', identifier: Dmp::MetadataHandler.remove_pk_prefix(dmp: p_key) }
+      end
+    end
+  end
+end

data/lib/dmp/dynamo_adapter.rb

@@ -3,8 +3,8 @@
 require 'json'
 require 'aws-sdk-dynamodb'
 
-require 'dmp_id_handler'
-require 'metadata_handler'
+require 'dmp/dmp_id_handler'
+require 'dmp/metadata_handler'
 
 module Dmp
   # DMP adapter for an AWS DynamoDB Table

data/lib/dmp/metadata_handler.rb

@@ -0,0 +1,251 @@
+ # frozen_string_literal: true
+
+require 'dmp/dmp_id_handler'
+
+module Dmp
+  # Handles alterations to DMP metadata elements
+  class MetadataHandler
+    PK_DMP_PREFIX = 'DMP#'.freeze
+    PK_PROVENANCE_PREFIX = 'PROVENANCE#'.freeze
+
+    SK_PREFIX = 'VERSION#'.freeze
+
+    LATEST_VERSION = "#{SK_PREFIX}latest".freeze
+    TOMBSTONE_VERSION = "#{SK_PREFIX}tombstone".freeze
+
+    class << self
+      # determine if the objects are equal. This ignores :SK, :dmphub_modification_day
+      # and :dmphub_updated_at attributes
+      def eql(dmp_a:, dmp_b:)
+        dmp_a = {} if dmp_a.nil?
+        dmp_b = {} if dmp_b.nil?
+        # They are not equal if the :PK do not match (and aren't blank)
+        return false if !dmp_a['PK'].nil? && !dmp_b['PK'].nil? && dmp_a['PK'] != dmp_b['PK']
+
+        a = deep_copy(obj: dmp_a)
+        b = deep_copy(obj: dmp_b)
+
+        # ignore some of the attributes before comparing
+        %w[SK dmphub_modification_day dmphub_updated_at dmphub_created_at].each do |key|
+          a.delete(key) unless a[key].nil?
+          b.delete(key) unless b[key].nil?
+        end
+        a == b
+      end
+
+      # Append the PK prefix for the object
+      def append_pk_prefix(dmp: nil, provenance: nil)
+        # If all the :PK types were passed return nil because we only want one
+        return nil if !dmp.nil? && !provenance.nil?
+
+        return "#{PK_DMP_PREFIX}#{remove_pk_prefix(dmp: dmp)}" unless dmp.nil?
+        return "#{PK_PROVENANCE_PREFIX}#{remove_pk_prefix(provenance: provenance)}" unless provenance.nil?
+
+        nil
+      end
+
+      # Strip off the PK prefix
+      def remove_pk_prefix(dmp: nil, provenance: nil)
+        # If all the :PK types were passed return nil because we only want one
+        return nil if !dmp.nil? && !provenance.nil?
+
+        return dmp.gsub(PK_DMP_PREFIX, '') unless dmp.nil?
+        return provenance.gsub(PK_PROVENANCE_PREFIX, '') unless provenance.nil?
+
+        nil
+      end
+
+      # Add all attributes necessary for the DMPHub
+      # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
+      def annotate_json(provenance:, p_key:, json:)
+        return nil if provenance.nil? || p_key.nil? || json.nil?
+
+        # Fail if the :PK does not match the :dmp_id if the json has a :PK
+        id = Dmp::DmpIdHandler.dmp_id_to_pk(json: json.fetch('dmp_id', {}))
+        id = nil if id != p_key && !json['PK'].nil?
+
+        annotated = deep_copy(obj: json)
+        annotated['PK'] = json['PK'] || p_key
+        annotated['SK'] = LATEST_VERSION
+
+        # Ensure that the :dmp_id matches the :PK
+        annotated['dmp_id'] = Dmp::DmpIdHandler.pk_to_dmp_id(p_key: annotated['PK'])
+
+        # Update the modification timestamps
+        annotated['dmphub_modification_day'] = Time.now.strftime('%Y-%M-%d')
+        annotated['dmphub_updated_at'] = Time.now.iso8601
+        # Only add the Creation date if it is blank
+        annotated['dmphub_created_at'] = Time.now.iso8601 if json['dmphub_created_at'].nil?
+        return annotated unless json['dmphub_provenance_id'].nil?
+
+        annotated['dmphub_provenance_id'] = provenance
+        return annotated if !annotated['dmphub_provenance_identifier'].nil? ||
+                            json.fetch('dmp_id', {})['identifier'].nil?
+
+        # Record the original Provenance system's identifier
+        annotated['dmphub_provenance_identifier'] = json['dmp_id']
+        annotated
+      end
+      # rubocop:enable Metrics/MethodLength, Metrics/AbcSize
+
+      # Process an update on the DMP metadata
+      def process_update(updater:, original_version:, new_version:)
+        return nil if updater.nil? || new_version.nil?
+        # If there is no :original_version then assume it's a new DMP
+        return new_version if original_version.nil?
+        # does not allow tombstoned DMPs to be updated
+        return original_version if original_version['SK'] == TOMBSTONE_VERSION
+        return original_version if eql(dmp_a: original_version, dmp_b: new_version)
+
+        owner = original_version['dmphub_provenance_id']
+        args = { owner: owner, updater: updater }
+
+        # If the system of provenance is making the change then just use the
+        # new version as the base and then splice in any mods made by others
+        args = args.merge({ base: new_version, mods: original_version})
+        return splice_for_owner(args) if owner == updater
+
+        # Otherwise use the original version as the base and then update the
+        # metadata owned by the updater system
+        args = args.merge({ base: original_version, mods: new_version})
+        splice_for_others(args)
+      end
+
+      private
+
+      # Splice changes from other systems back onto the system of provenance's changes
+      def splice_for_owner(owner:, updater:, base:, mods:)
+        return base if owner.nil? || updater.nil? || mods.nil?
+        return mods if base.nil?
+
+        provenance_regex = %r{"dmphub_provenance_id":"#{PK_PROVENANCE_PREFIX}[a-zA-Z\-_]+"}
+        others = base.to_json.match(provenance_regex)
+        # Just return it as is if there are no mods by other systems
+        return mods if others.nil?
+
+        spliced = deep_copy(obj: base)
+        cloned_mods = deep_copy(obj: mods)
+
+        # ensure that the :project and :funding are defined
+        spliced['project'] = [{}] if spliced['project'].nil? || spliced['project'].empty?
+        spliced['project'].first['funding'] = [] if spliced['project'].first['funding'].nil?
+        # get all the new funding and retain other system's funding metadata
+        mod_fundings = cloned_mods.fetch('project', [{}]).first.fetch('funding', [])
+        other_fundings = spliced['project'].first['funding'].reject { |fund| fund['dmphub_provenance_id'].nil? }
+        # process funding (just attach all funding not owned by the system of provenance)
+        spliced['project'].first['funding'] = mod_fundings
+        spliced['project'].first['funding'] << other_fundings if other_fundings.any?
+        return spliced if cloned_mods['dmproadmap_related_identifiers'].nil?
+
+        # process related_identifiers (just attach all related identifiers not owned by the system of provenance)
+        spliced['dmproadmap_related_identifiers'] = [] if spliced['dmproadmap_related_identifiers'].nil?
+        mod_relateds = cloned_mods.fetch('dmproadmap_related_identifiers', [])
+        other_relateds = spliced['dmproadmap_related_identifiers'].reject { |id| id['dmphub_provenance_id'].nil? }
+        spliced['dmproadmap_related_identifiers'] = mod_relateds
+        spliced['dmproadmap_related_identifiers'] << other_relateds if other_relateds.any?
+        spliced
+      end
+
+      # Splice changes from other systems back onto the system of provenance's changes
+      def splice_for_others(owner:, updater:, base:, mods:)
+        return base if owner.nil? || updater.nil? || base.nil? || mods.nil?
+
+        spliced = deep_copy(obj: base)
+         base_funds = spliced.fetch('project', [{}]).first.fetch('funding', [])
+        base_relateds = spliced.fetch('dmproadmap_related_identifiers', [])
+
+        mod_funds = mods.fetch('project', [{}]).first.fetch('funding', [])
+        mod_relateds = mods.fetch('dmproadmap_related_identifiers', [])
+
+        # process funding
+        spliced['project'].first['funding'] = update_funding(
+          updater: updater, base: base_funds, mods: mod_funds
+        )
+        return spliced if mod_relateds.empty?
+
+        # process related_identifiers
+        spliced['dmproadmap_related_identifiers'] = update_related_identifiers(
+          updater: updater, base: base_relateds, mods: mod_relateds
+        )
+        spliced
+      end
+
+      # Splice funding changes
+      def update_funding(updater:, base:, mods:)
+        return base if updater.nil? || mods.nil? || mods.empty?
+
+        spliced = deep_copy(obj: base)
+        mods.each do |funding|
+          # Ignore it if it has no status or grant id
+          next if funding['funding_status'].nil? && funding['grant_id'].nil?
+
+          # See if there is an existing funding record for the funder that's waiting on an update
+          spliced = [] if spliced.nil?
+          items = spliced.select do |orig|
+            !orig['funder_id'].nil? &&
+              orig['funder_id'] == funding['funder_id'] &&
+              %w[applied planned].include?(orig['funding_status'])
+          end
+          # Always grab the most current
+          item = items.sort { |a, b| b.fetch('dmphub_created_at', '') <=> a.fetch('dmphub_created_at', '') }.first
+
+          # Out with the old and in with the new
+          spliced.delete(item) unless item.nil?
+          # retain the original name
+          funding['name'] = item['name'] unless item.nil?
+          item = deep_copy(obj: funding)
+
+          item['funding_status'] == funding['funding_status'] unless funding['funding_status'].nil?
+          spliced << item if funding['grant_id'].nil?
+          next if funding['grant_id'].nil?
+
+          item['grant_id'] = funding['grant_id']
+          item['funding_status'] = funding['grant_id'].nil? ? 'rejected' : 'granted'
+
+          # Add the provenance to the entry
+          item['grant_id']['dmphub_provenance_id'] = updater
+          item['grant_id']['dmphub_created_at'] = Time.now.iso8601
+          spliced << item
+        end
+        spliced
+      end
+
+      # Splice related identifier changes
+      def update_related_identifiers(updater:, base:, mods:)
+        return base if updater.nil? || mods.nil? || mods.empty?
+
+        # Remove the updater's existing related identifiers and replace with the new set
+        spliced = base.nil? ? [] : deep_copy(obj: base)
+        spliced = spliced.reject { |related| related['dmphub_provenance_id'] == updater }
+        # Add the provenance to the entry
+        updates = mods.nil? ? [] : deep_copy(obj: mods)
+        updates = updates.map do |related|
+          related['dmphub_provenance_id'] = updater
+          related
+        end
+        spliced + updates
+      end
+
+      # Ruby's clone/dup methods do not clone/dup the children, so we need to do it here
+      def deep_copy(obj:)
+        case obj.class.name
+        when 'Array'
+          obj.map { |item| deep_copy(obj: item) }
+        when 'Hash'
+          hash = obj.dup
+          hash.each_pair do |key, value|
+            if ::String === key || ::Symbol === key
+              hash[key] = deep_copy(obj: value)
+            else
+              hash.delete(key)
+              hash[deep_copy(obj: key)] = deep_copy(obj: value)
+            end
+          end
+          hash
+        else
+          obj.dup
+        end
+      end
+    end
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: dmp-dynamo_adapter
 version: !ruby/object:Gem::Version
-  version: 0.1.3
+  version: 0.1.6
 platform: ruby
 authors:
 - briri
@@ -58,7 +58,9 @@ executables: []
 extensions: []
 extra_rdoc_files: []
 files:
+- lib/dmp/dmp_id_handler.rb
 - lib/dmp/dynamo_adapter.rb
+- lib/dmp/metadata_handler.rb
 homepage: https://github.com/CDLUC3/dmphub-v2/tree/main/gems/dmp-dynamo_adapter
 licenses:
 - MIT