adiwg-mdtranslator 2.21.0.pre.beta.0 → 2.21.0.pre.beta.2

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_dcat_us.rb

@@ -0,0 +1,104 @@
+require 'jbuilder'
+require_relative 'dcat_us_keyword'
+require_relative 'dcat_us_publisher'
+require_relative 'dcat_us_contact_point'
+require_relative 'dcat_us_identifier'
+require_relative 'dcat_us_distribution'
+require_relative 'dcat_us_spatial'
+require_relative 'dcat_us_temporal'
+require_relative 'dcat_us_modified'
+require_relative 'dcat_us_access_level'
+require_relative 'dcat_us_rights'
+require_relative 'dcat_us_license'
+require_relative 'dcat_us_issued'
+require_relative 'dcat_us_described_by'
+require_relative 'dcat_us_is_part_of'
+require_relative 'dcat_us_theme'
+require_relative 'dcat_us_references'
+require_relative 'dcat_us_landing_page'
+require_relative 'dcat_us_system_of_records'
+require_relative 'dcat_us_description'
+require_relative 'dcat_us_described_by_type'
+require_relative 'dcat_us_accrualPeriodicity'
+require_relative 'dcat_us_language'
+require_relative 'dcat_us_primaryITInvestmentUII'
+require_relative 'dcat_us_program_code'
+require_relative 'dcat_us_bureau_code'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+
+            def self.build(intObj, responseObj)
+               metadataInfo = intObj[:metadata][:metadataInfo]
+               resourceInfo = intObj[:metadata][:resourceInfo]
+               citation = resourceInfo[:citation]
+
+               title = citation[:title]
+               description = Description.build(intObj)
+               keyword = Keyword.build(intObj)
+               modified = Modified.build(intObj)
+               publisher = Publisher.build(intObj)
+               contactPoint = ContactPoint.build(intObj)
+               accessLevel = AccessLevel.build(intObj)
+               identifier = Identifier.build(intObj)
+               distribution = Distribution.build(intObj)
+               rights = Rights.build(intObj, accessLevel)
+               spatial = Spatial.build(intObj)
+               temporal = Temporal.build(intObj)
+               license = License.build(intObj)
+               issued = Issued.build(intObj)
+               language = Language.build(intObj)
+               describedBy = DescribedBy.build(intObj)
+               isPartOf = IsPartOf.build(intObj)
+               theme = Theme.build(intObj)
+               references = References.build(intObj)
+               landingPage = LandingPage.build(intObj)
+               systemOfRecords = SystemOfRecords.build(intObj)
+               describedByType = DescribedByType.build(intObj)
+               accrualPeriodicity = AccrualPeriodicity.build(intObj)
+               primaryITInvestmentUII = PrimaryITInvestmentUII.build(intObj)
+               programCode = ProgramCode.build(intObj)
+               bureauCode = BureauCode.build(intObj)
+
+               @Namespace = ADIWG::Mdtranslator::Writers::Dcat_us
+
+               Jbuilder.new do |json|
+                  json.set!('@type', 'dcat:Dataset')
+                  json.set!('title', title)
+                  json.set!('description', description)
+                  json.set!('keyword', keyword)
+                  json.set!('modified', modified)
+                  json.set!('publisher', publisher) 
+                  json.set!('contactPoint', contactPoint)
+                  json.set!('identifier', identifier)
+                  json.set!('accessLevel', accessLevel)
+                  json.set!('bureauCode', bureauCode)
+                  json.set!('programCode', programCode)
+                  json.set!('distribution', distribution)
+
+                  json.set!('license', license)
+                  json.set!('rights', rights)
+                  json.set!('spatial', spatial)
+                  json.set!('temporal', temporal)
+
+                  json.set!('issued', issued)
+                  json.set!('accrualPeriodicity', accrualPeriodicity)
+                  json.set!('language', language)
+                  # json.set!('dataQuality', metadataInfo[:metadataMaintenance][:maintenanceNote])
+                  json.set!('theme', theme)
+                  json.set!('references', references)
+                  json.set!('landingPage', landingPage)
+                  json.set!('isPartOf', isPartOf)
+                  json.set!('systemOfRecords', systemOfRecords)
+                  json.set!('primaryITInvestmentUII', primaryITInvestmentUII)
+                  json.set!('describedBy', describedBy)
+                  json.set!('describedByType', describedByType)
+                  # json.set!('conformsTo', metadataInfo[:metadataStandards][0][:standardName])
+               end
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_described_by.rb

@@ -0,0 +1,29 @@
+require 'jbuilder'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module DescribedBy
+
+               def self.build(intObj)
+                  dataDictionaries = intObj[:dataDictionaries]
+                  dataDictionaries.each do |dataDictionary|
+                     unless dataDictionary[:includedWithDataset]
+                        onlineResources = dataDictionary[:citation][:onlineResources]
+                        onlineResources.each do |resource|
+                           if resource[:olResURI]
+                              return resource[:olResURI]
+                           end
+                        end
+                     end
+                  end
+                
+                  nil
+                end                
+                
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_described_by_type.rb

@@ -0,0 +1,33 @@
+require 'jbuilder'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module DescribedByType
+
+               def self.build(intObj)
+
+                  #  metadataInfo[:metadataOnlineOptions][0][:olResProtocol]
+                  dataDictionaries = intObj[:dataDictionaries]
+                  describedByType = ''
+                  dataDictionaries.each do |dataDictionary|
+                     unless dataDictionary[:includedWithDataset]
+                        onlineResources = dataDictionary[:citation][:onlineResources]
+                        onlineResources.each do |resource|
+                           if resource[:olResURI] && !resource[:olResURI].end_with?('.html')
+                              describedByType = resource[:olResProtocol]
+                              break
+                           end
+                        end
+                     end
+                  end
+                
+                  describedByType
+               end                
+
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_description.rb

@@ -0,0 +1,19 @@
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module Description
+
+               def self.build(intObj)
+                  resourceInfo = intObj.dig(:metadata, :resourceInfo)
+                  description = resourceInfo&.dig(:abstract)
+
+                  return nil unless description
+               end
+                                    
+
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_distribution.rb

@@ -0,0 +1,59 @@
+require 'jbuilder'
+require_relative 'dcat_us_access_url'
+require_relative 'dcat_us_download_url'
+require_relative 'dcat_us_media_type'
+
+
+module ADIWG
+  module Mdtranslator
+    module Writers
+      module Dcat_us
+        module Distribution
+
+          def self.build(intObj)
+
+            resourceDistributions = intObj.dig(:metadata, :distributorInfo)
+            distributions = []
+
+            resourceDistributions&.each do |resource|
+              description = resource[:description] || ''
+              break_flag = false  # Flag to control nested loop breaks
+
+              resource[:distributor]&.each do |distributor|
+                break if break_flag
+
+                distributor[:transferOptions]&.each do |transfer|
+                  break if break_flag
+
+                  mediaType = MediaType.build(transfer)
+
+                  transfer[:onlineOptions]&.each do |option|
+                    next unless option[:olResURI]
+                    accessURL = AccessURL.build(option)
+                    downloadURL = DownloadURL.build(option)
+                    title = option[:olResName] || ''
+
+                    distribution = Jbuilder.new do |json|
+                      json.set!('@type', 'dcat:Distribution')
+                      json.set!('description', description)
+                      json.set!('accessURL', accessURL) if accessURL
+                      json.set!('downloadURL', downloadURL) if downloadURL
+                      json.set!('mediaType', mediaType)
+                      json.set!('title', title)
+                    end
+
+                    distributions << distribution.attributes!
+                    break_flag = true
+                    break
+                  end
+                end
+              end
+            end
+            distributions
+          end
+
+        end
+      end
+    end
+  end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_download_url.rb

@@ -0,0 +1,17 @@
+require 'jbuilder'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module DownloadURL
+
+               def self.build(option)
+                  option[:olResURI] unless option[:olResURI].end_with?('.html')
+                end                
+
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_identifier.rb

@@ -0,0 +1,29 @@
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module Identifier
+
+               def self.build(intObj)
+                  citation = intObj.dig(:metadata, :resourceInfo, :citation)
+                  identifiers = citation&.dig(:identifiers)
+                  onlineResources = citation&.dig(:onlineResources)
+                  uri = onlineResources.dig(0, :olResURI)
+                
+                  namespace_is_doi = identifiers&.any? { |identifier| identifier[:namespace]&.casecmp?("DOI") }
+                
+                  if namespace_is_doi
+                    return uri
+                  elsif uri && uri.downcase.include?("doi")
+                    return uri
+                  end
+                
+                  nil
+               end
+                                    
+
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_is_part_of.rb

@@ -0,0 +1,27 @@
+require 'jbuilder'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module IsPartOf
+
+               def self.build(intObj)
+                  associatedResources = intObj.dig(:metadata, :associatedResources)
+                
+                  associatedResources.each do |resource|
+                    next unless resource[:initiativeType] == "collection" && resource[:associationType] == "collectiveTitle"
+                
+                    onlineResources = resource.dig(:resourceCitation, :onlineResources) || []
+                    uri = onlineResources.find { |onlineResource| onlineResource[:olResURI] }&.dig(:olResURI)
+                    return uri if uri
+                  end
+                
+                  nil
+                end                
+
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_issued.rb

@@ -0,0 +1,19 @@
+require 'jbuilder'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module Issued
+
+               def self.build(intObj)
+                  dates = intObj[:metadata][:resourceInfo][:citation][:dates].map { |obj| obj[:date] }
+                  earliest_date = dates.min
+                  return earliest_date
+                end
+                
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_keyword.rb

@@ -0,0 +1,19 @@
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module Keyword
+ 
+               def self.build(intObj)
+                  resourceInfo = intObj.dig(:metadata, :resourceInfo)
+                  keywords = resourceInfo&.dig(:keywords)
+                
+                  return keywords&.flat_map { |keyword| keyword[:keywords]&.map { |kw| kw[:keyword] } } || []
+                end                
+                
+            end
+         end
+      end
+   end
+end
+ 

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_landing_page.rb

@@ -0,0 +1,27 @@
+require 'jbuilder'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module LandingPage
+
+               def self.build(intObj)
+                  onlineResources = intObj.dig(:metadata, :resourceInfo, :citation, :onlineResources)
+                
+                  if onlineResources
+                    onlineResources.each do |resource|
+                      if resource.dig(:olResFunction) == 'landingPage'
+                        return resource.dig(:olResURI)
+                      end
+                    end
+                  end
+                  
+                  return nil
+                end                
+                
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_language.rb

@@ -0,0 +1,23 @@
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module Language
+ 
+               def self.build(intObj)
+                  metadataInfo = intObj.dig(:metadata, :metadataInfo)
+                  defaultLanguage = metadataInfo&.dig(:defaultMetadataLocale, :languageCode)
+                  additionalLanguage = metadataInfo&.dig(:otherMetadataLocales)
+                  languages = []
+                  languages << defaultLanguage if defaultLanguage
+                  additionalLanguage.each { |lang| languages << lang[:languageCode] } if additionalLanguage
+
+                  return languages
+                end                
+                
+            end
+         end
+      end
+   end
+end
+ 

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_license.rb

@@ -0,0 +1,18 @@
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module License
+
+               def self.build(intObj)
+                  resourceInfo = intObj.dig(:metadata, :resourceInfo)
+                  title = resourceInfo.dig(:constraints, 0, :reference, 0, :title)
+                  license = title || 'https://creativecommons.org/publicdomain/zero/1.0/'
+                  license
+               end
+
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_media_type.rb

@@ -0,0 +1,17 @@
+require 'jbuilder'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module MediaType
+
+               def self.build(transfer)
+                  transfer.dig(:distributionFormats)&.find { |format| format.dig(:formatSpecification, :title) }&.dig(:formatSpecification, :title) || ''
+                end                
+
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_modified.rb

@@ -0,0 +1,29 @@
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module Modified
+
+               def self.build(intObj)
+                  resourceInfo = intObj[:metadata][:resourceInfo]
+                  citation = resourceInfo[:citation]
+                  dates = citation[:dates]
+
+                  mostRecentDate = nil
+
+                  dates.each do |date|
+                     if date[:dateType] == "lastUpdated" || date[:dateType] == "lastRevised" || date[:dateType] == "revision"
+                        if mostRecentDate.nil? || date[:date] > mostRecentDate
+                           mostRecentDate = date[:date]
+                        end
+                     end
+                  end
+
+                  return mostRecentDate
+               end               
+
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_primaryITInvestmentUII.rb

@@ -0,0 +1,19 @@
+require 'jbuilder'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module PrimaryITInvestmentUII
+
+               def self.build(intObj)
+                  primaryITInvestmentUII = intObj[:metadata][:metadataInfo][:metadataIdentifier][:identifier]
+                  
+                  primaryITInvestmentUII ? primaryITInvestmentUII : nil
+                end
+                
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_program_code.rb

@@ -0,0 +1,33 @@
+require 'jbuilder'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module ProgramCode
+
+               def self.build(intObj)
+                
+                    responsibleParties = intObj[:metadata][:resourceInfo][:citation][:responsibleParties]
+                    contacts = []
+                    responsibleParties.each do |party|
+                        contactId = party[:parties][0][:contactId]
+                        contacts << Dcat_us.get_contact_by_id(contactId)
+                    end
+
+                    programContacts = contacts&.select { |contact| contact[:externalIdentifier].any? { |id| id[:namespace] == 'programCode'} }
+
+                    programsCodes = []
+                    programContacts.each do |contact|
+                        programCode = contact[:externalIdentifier].find { |id| id[:namespace] == 'programCode' }
+                        programsCodes << programCode[:identifier]
+                    end
+
+                    return programsCodes
+                end
+                
+            end
+         end
+      end
+   end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_publisher.rb

@@ -0,0 +1,74 @@
+module ADIWG
+  module Mdtranslator
+    module Writers
+      module Dcat_us
+        module Publisher
+          def self.build(intObj)
+            contacts = intObj.dig(:contacts)
+            metadata = intObj.dig(:metadata)
+            responsible_parties = metadata&.dig(:resourceInfo, :citation, :responsibleParties)
+            resource_distributions = metadata&.dig(:distributorInfo)
+            
+            publisher = find_publisher(responsible_parties) || find_distributions_publisher(intObj)
+            
+            return if publisher.nil?
+            
+            name, org_name = extract_names(publisher, contacts)
+            
+            return if name.nil?
+            
+            build_json(name, org_name)
+          end
+          
+          private
+          
+          def self.find_publisher(responsible_parties)
+            responsible_parties&.detect do |party|
+              party[:roleName] == 'publisher' && party.dig(:parties)&.first&.dig(:contactType) == 'organization'
+            end
+          end
+
+          def self.find_distributions_publisher(intObj)
+            distributor_info = intObj.dig(:metadata, :distributorInfo) || []
+            distributor_info.each do |distribution|
+              distributors = distribution[:distributor] || []
+              distributors.each do |dist|
+                contact_type = dist.dig(:contact, :parties, 0, :contactType)
+                return dist[:contact] if contact_type == 'organization'
+              end
+            end
+            nil
+          end            
+          
+          def self.extract_names(publisher, contacts)
+            contact_id = publisher.dig(:parties)&.first&.dig(:contactId) || publisher[:contactId]
+            related_contact = contacts&.detect { |contact| contact[:contactId] == contact_id }
+            org_name = nil
+            if related_contact
+              member_of_org_ids = related_contact[:memberOfOrgs]
+              member_of_orgs = contacts.select { |contact| member_of_org_ids&.include?(contact[:contactId]) && contact[:isOrganization] }
+              org_name = member_of_orgs.first&.dig(:name)
+            end
+            name = publisher.dig(:parties)&.first&.dig(:contactName) || publisher[:contactName]
+            [name, org_name]
+          end
+          
+          def self.build_json(name, org_name)
+            Jbuilder.new do |json|
+              json.set!('@type', 'org:Organization')
+              json.set!('name', name)
+              
+              if org_name
+                json.subOrganizationOf do |json_sub|
+                  json_sub.set!('@type', 'org:Organization')
+                  json_sub.set!('name', org_name)
+                end
+              end
+            end
+          end
+
+        end
+      end
+    end
+  end
+end

data/lib/adiwg/mdtranslator/writers/dcat_us/sections/dcat_us_references.rb

@@ -0,0 +1,50 @@
+require 'jbuilder'
+
+module ADIWG
+   module Mdtranslator
+      module Writers
+         module Dcat_us
+            module References
+
+               def self.build(intObj)
+                  metadata = intObj.dig(:metadata)
+                  associated_resources = metadata&.dig(:associatedResources)
+                  additional_documents = metadata&.dig(:additionalDocuments)
+                  uris = []
+                
+                  if associated_resources
+                    associated_resources.each do |associated|
+                      citation = associated.dig(:resourceCitation)
+                      online_resources = citation&.dig(:onlineResources)
+                      next unless online_resources
+                
+                      online_resources.each do |online|
+                        uri = online.dig(:olResURI)
+                        uris << uri if uri
+                      end
+                    end
+                  end
+                
+                  if additional_documents
+                    additional_documents.each do |additional|
+                      citations = additional&.dig(:citation) || []
+                      citations.each do |citation|
+                        online_resources = citation&.dig(:onlineResources)
+                        next unless online_resources
+                
+                        online_resources.each do |online|
+                          uri = online.dig(:olResURI)
+                          uris << uri if uri
+                        end
+                      end
+                    end
+                  end
+                
+                  uris.join(',')
+               end
+
+            end
+         end
+      end
+   end
+end