commonmeta-ruby 3.2.8 → 3.2.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c4efef118736aa6738cde85e83a7074cfc0746339c6ee4a1c9d7fb464253a8c7
-  data.tar.gz: 02665e140aae9c74d6bab0e0d09472992e1342d74784d33153104dba316fe3f9
+  metadata.gz: 7def7640d11d8d88a8a9fb8b18394cd88d5de57e44832c9b05d7c23213857970
+  data.tar.gz: 9adc93a35e9c9060ddd3bda1bb859e88f45a00d31454af13c54b54d4c96209b8
 SHA512:
-  metadata.gz: 1beb77ff61c956fb294b7c072151970326d62a2cc9d9cfe6a48ff6d652a24829f8bf832f9442c970ba09e6c589fc0d73ba0cef6e5f1852d30e2205da47a06732
-  data.tar.gz: 20608ff7d3a77b9a40c74ab4c5fc0e16377714d56bb60859c00a3c3e1378ea2484943c9a1ad25b062d454896278df81ca75a4426bac626b81cbf4ad23aaf1491
+  metadata.gz: 61ab2849532d1967e78111a68eec7c4ece6c1cbb968ae705fde933e11dab50079a4f441a4ab4c3acfe2e6d5e9d76dd1496d7b2092244cde2e5a0d588c8747b0d
+  data.tar.gz: cd470bf78ce018a78d89133ec5e17ab5dc6f3f35f982cc0e26736b24661bb316b6b9491e93ff897d98773ab027264b4532d367da2c03e546331ee20600c12a24

data/Gemfile.lock

@@ -1,10 +1,10 @@
 PATH
   remote: .
   specs:
-    commonmeta-ruby (3.2.8)
+    commonmeta-ruby (3.2.10)
       activesupport (>= 4.2.5, < 8.0)
       addressable (~> 2.8.1, < 2.8.2)
-      base32-url (>= 0.5.0, < 1)
+      base32-url (>= 0.7.0, < 1)
       bibtex-ruby (~> 6.0)
       builder (~> 3.2, >= 3.2.4)
       citeproc-ruby (~> 2.0)
@@ -34,7 +34,8 @@ GEM
     addressable (2.8.1)
       public_suffix (>= 2.0.2, < 6.0)
     ast (2.4.2)
-    base32-url (0.6.0)
+    base32-url (0.7.0)
+      uuidtools (~> 2.1, >= 2.1.5)
     bibtex-ruby (6.0.0)
       latex-decode (~> 0.0)
     builder (3.2.4)
@@ -241,6 +242,7 @@ GEM
     unf_ext (0.0.8.2)
     unicode-display_width (2.4.2)
     unicode-types (1.8.0)
+    uuidtools (2.2.0)
     vcr (6.1.0)
     webmock (3.18.1)
       addressable (>= 2.8.0)

data/commonmeta.gemspec

@@ -20,7 +20,7 @@ Gem::Specification.new do |s|
   # Declare dependencies here, rather than in the Gemfile
   s.add_dependency 'activesupport', '>= 4.2.5', '< 8.0'
   s.add_dependency 'addressable', '~> 2.8.1', '< 2.8.2'
-  s.add_dependency 'base32-url', '>= 0.5.0', '< 1'
+  s.add_dependency 'base32-url', '>= 0.7.0', '< 1'
   s.add_dependency 'bibtex-ruby', '~> 6.0'
   s.add_dependency 'builder', '~> 3.2', '>= 3.2.4'
   s.add_dependency 'citeproc-ruby', '~> 2.0'

data/lib/commonmeta/crossref_utils.rb

@@ -3,12 +3,12 @@
 module Commonmeta
   module CrossrefUtils
     def write_crossref_xml
-      @crossref_xml ||= Nokogiri::XML::Builder.new(encoding: 'UTF-8') do |xml|
+      @crossref_xml ||= Nokogiri::XML::Builder.new(encoding: "UTF-8") do |xml|
         xml.doi_batch(crossref_root_attributes) do
           xml.head do
             # we use a uuid as batch_id
             xml.doi_batch_id(SecureRandom.uuid)
-            xml.timestamp(Time.now.utc.strftime('%Y%m%d%H%M%S'))
+            xml.timestamp(Time.now.utc.strftime("%Y%m%d%H%M%S"))
             xml.depositor do
               xml.depositor_name(depositor)
               xml.email_address(email)
@@ -23,12 +23,12 @@ module Commonmeta
     end
 
     def crossref_errors(xml: nil)
-      filepath = File.expand_path('../../resources/crossref/crossref5.3.1.xsd', __dir__)
-      File.open(filepath, 'r') do |f|
+      filepath = File.expand_path("../../resources/crossref/crossref5.3.1.xsd", __dir__)
+      File.open(filepath, "r") do |f|
         schema = Nokogiri::XML::Schema(f)
       end
 
-      schema.validate(Nokogiri::XML(xml, nil, 'UTF-8')).map(&:to_s).unwrap
+      schema.validate(Nokogiri::XML(xml, nil, "UTF-8")).map(&:to_s).unwrap
     rescue Nokogiri::XML::SyntaxError => e
       e.message
     end
@@ -37,9 +37,9 @@ module Commonmeta
       return xml if doi_from_url(id).blank?
 
       case type
-      when 'JournalArticle'
+      when "JournalArticle"
         insert_journal(xml)
-      when 'Article'
+      when "Article"
         insert_posted_content(xml)
       end
     end
@@ -47,21 +47,21 @@ module Commonmeta
     def insert_journal(xml)
       xml.journal do
         if language.present?
-          xml.journal_metadata('language' => language[0..1]) do
-            xml.full_title(container['title'])
+          xml.journal_metadata("language" => language[0..1]) do
+            xml.full_title(container["title"])
           end
         else
           xml.journal_metadata do
-            xml.full_title(container['title'])
+            xml.full_title(container["title"])
           end
         end
-        xml.journal_article('publication_type' => 'full_text') do
+        xml.journal_article("publication_type" => "full_text") do
           insert_crossref_titles(xml)
           insert_crossref_creators(xml)
           insert_crossref_publication_date(xml)
           insert_crossref_abstract(xml)
           insert_crossref_issn(xml)
-          insert_publisher_item(xml)
+          insert_item_number(xml)
           insert_crossref_access_indicators(xml)
           insert_doi_data(xml)
           insert_citation_list(xml)
@@ -70,7 +70,7 @@ module Commonmeta
     end
 
     def insert_posted_content(xml)
-      posted_content = { 'type' => 'other', 'language' => language ? language[0..1] : nil }.compact
+      posted_content = { "type" => "other", "language" => language ? language[0..1] : nil }.compact
 
       xml.posted_content(posted_content) do
         insert_group_title(xml)
@@ -78,7 +78,7 @@ module Commonmeta
         insert_crossref_titles(xml)
         insert_posted_date(xml)
         insert_institution(xml)
-        insert_publisher_item(xml)
+        insert_item_number(xml)
         insert_crossref_abstract(xml)
         insert_crossref_access_indicators(xml)
         insert_doi_data(xml)
@@ -89,14 +89,14 @@ module Commonmeta
     def insert_group_title(xml)
       return xml if subjects.blank?
 
-      xml.group_title(subjects.first['subject'].titleize)
+      xml.group_title(subjects.first["subject"].titleize)
     end
 
     def insert_crossref_creators(xml)
       xml.contributors do
         Array.wrap(creators).each_with_index do |person, index|
-          xml.person_name('contributor_role' => 'author',
-                          'sequence' => index.zero? ? 'first' : 'additional') do
+          xml.person_name("contributor_role" => "author",
+                          "sequence" => index.zero? ? "first" : "additional") do
             insert_crossref_person(xml, person)
           end
         end
@@ -104,15 +104,15 @@ module Commonmeta
     end
 
     def insert_crossref_person(xml, person)
-      xml.given_name(person['givenName']) if person['givenName'].present?
-      xml.surname(person['familyName']) if person['familyName'].present?
-      if person.dig('id') && URI.parse(person.dig('id')).host == 'orcid.org'
-        xml.ORCID(person.dig('id'))
+      xml.given_name(person["givenName"]) if person["givenName"].present?
+      xml.surname(person["familyName"]) if person["familyName"].present?
+      if person.dig("id") && URI.parse(person.dig("id")).host == "orcid.org"
+        xml.ORCID(person.dig("id"))
       end
-      Array.wrap(person['affiliation']).each do |affiliation|
-        attributes = { 'affiliationIdentifier' => affiliation['affiliationIdentifier'],
-                       'affiliationIdentifierScheme' => affiliation['affiliationIdentifierScheme'], 'schemeURI' => affiliation['schemeUri'] }.compact
-        xml.affiliation(affiliation['name'], attributes)
+      Array.wrap(person["affiliation"]).each do |affiliation|
+        attributes = { "affiliationIdentifier" => affiliation["affiliationIdentifier"],
+                       "affiliationIdentifierScheme" => affiliation["affiliationIdentifierScheme"], "schemeURI" => affiliation["schemeUri"] }.compact
+        xml.affiliation(affiliation["name"], attributes)
       end
     end
 
@@ -120,7 +120,7 @@ module Commonmeta
       xml.titles do
         Array.wrap(titles).each do |title|
           if title.is_a?(Hash)
-            xml.title(title['title'])
+            xml.title(title["title"])
           else
             xml.title(title)
           end
@@ -131,15 +131,15 @@ module Commonmeta
     def insert_citation_list(xml)
       xml.citation_list do
         Array.wrap(references).each do |ref|
-          xml.citation('key' => ref['key']) do
-            xml.journal_article(ref['journal_title']) if ref['journal_title'].present?
-            xml.author(ref['author']) if ref['author'].present?
-            xml.volume(ref['volume']) if ref['volume'].present?
-            xml.first_page(ref['first_page']) if ref['first_page'].present?
-            xml.cYear(ref['cYear']) if ref['cYear'].present?
-            xml.article_title(ref['article_title']) if ref['article_title'].present?
-            xml.doi(doi_from_url(ref['doi'])) if ref['doi'].present?
-            xml.unstructured_citation(ref['url']) if ref['url'].present?
+          xml.citation("key" => ref["key"]) do
+            xml.journal_article(ref["journal_title"]) if ref["journal_title"].present?
+            xml.author(ref["author"]) if ref["author"].present?
+            xml.volume(ref["volume"]) if ref["volume"].present?
+            xml.first_page(ref["first_page"]) if ref["first_page"].present?
+            xml.cYear(ref["cYear"]) if ref["cYear"].present?
+            xml.article_title(ref["article_title"]) if ref["article_title"].present?
+            xml.doi(doi_from_url(ref["doi"])) if ref["doi"].present?
+            xml.unstructured_citation(ref["url"]) if ref["url"].present?
           end
         end
       end
@@ -163,12 +163,12 @@ module Commonmeta
     def insert_crossref_access_indicators(xml)
       return xml if license.blank?
 
-      rights_uri = license['url']
+      rights_uri = license["url"]
 
-      xml.program('xmlns' => 'http://www.crossref.org/AccessIndicators.xsd',
-                  'name' => 'AccessIndicators') do
-        xml.license_ref(rights_uri, 'applies_to' => 'vor')
-        xml.license_ref(rights_uri, 'applies_to' => 'tdm')
+      xml.program("xmlns" => "http://www.crossref.org/AccessIndicators.xsd",
+                  "name" => "AccessIndicators") do
+        xml.license_ref(rights_uri, "applies_to" => "vor")
+        xml.license_ref(rights_uri, "applies_to" => "tdm")
       end
     end
 
@@ -204,7 +204,7 @@ module Commonmeta
       xml.subjects do
         subjects.each do |subject|
           if subject.is_a?(Hash)
-            xml.subject(subject['subject'])
+            xml.subject(subject["subject"])
           else
             xml.subject(subject)
           end
@@ -225,11 +225,11 @@ module Commonmeta
     end
 
     def insert_crossref_publication_date(xml)
-      return xml if date['registered'].blank?
+      return xml if date["registered"].blank?
 
-      date_ = get_datetime_from_iso8601(date['registered'])
+      date_ = get_datetime_from_iso8601(date["registered"])
 
-      xml.publication_date('media_type' => 'online') do
+      xml.publication_date("media_type" => "online") do
         xml.month(date_.month) if date_.month.present?
         xml.day(date_.day) if date_.day.present?
         xml.year(date_.year) if date_.year.present?
@@ -237,9 +237,9 @@ module Commonmeta
     end
 
     def insert_posted_date(xml)
-      return xml if date['published'].blank?
+      return xml if date["published"].blank?
 
-      date_ = get_datetime_from_iso8601(date['published'])
+      date_ = get_datetime_from_iso8601(date["published"])
 
       xml.posted_date do
         xml.month(date_.month) if date_.month.present?
@@ -252,24 +252,22 @@ module Commonmeta
       return xml if publisher.blank?
 
       xml.institution do
-        xml.institution_name(publisher['name'])
+        xml.institution_name(publisher["name"])
       end
     end
 
-    def insert_publisher_item(xml)
+    def insert_item_number(xml)
       return xml if alternate_identifiers.blank?
 
-      xml.publisher_item do
-        alternate_identifiers.each do |alternate_identifier|
-          attributes = {
-            'item_number_type' => alternate_identifier['alternateIdentifierType'] ? alternate_identifier['alternateIdentifierType'].downcase : nil
-          }.compact
-          
-          # remove dashes from UUIDs, as item_number can only be 32 characters long
-          alternate_identifier['alternateIdentifier'].gsub!('-', '') if alternate_identifier['alternateIdentifierType'] == 'UUID'
+      alternate_identifiers.each do |alternate_identifier|
+        attributes = {
+          "item_number_type" => alternate_identifier["alternateIdentifierType"] ? alternate_identifier["alternateIdentifierType"].downcase : nil,
+        }.compact
 
-          xml.item_number(alternate_identifier['alternateIdentifier'], attributes)
-        end
+        # convert UUIDs into base32 encoded strings, as item_number can only be 32 characters long (UUIDv4 is 36 characters long)
+        alternate_identifier["alternateIdentifier"] = Base32::URL.encode_uuid(alternate_identifier["alternateIdentifier"], split: 7, checksum: true) if alternate_identifier["alternateIdentifierType"] == "UUID"
+
+        xml.item_number(alternate_identifier["alternateIdentifier"], attributes)
       end
     end
 
@@ -279,9 +277,9 @@ module Commonmeta
       xml.doi_data do
         xml.doi(doi_from_url(id).downcase)
         xml.resource(url)
-        xml.collection('property' => 'text-mining') do
+        xml.collection("property" => "text-mining") do
           xml.item do
-            xml.resource(url, 'mime_type' => 'text/html')
+            xml.resource(url, "mime_type" => "text/html")
           end
         end
       end
@@ -296,26 +294,26 @@ module Commonmeta
           r = rights
         else
           r = {}
-          r['rights'] = rights
-          r['rightsUri'] = normalize_id(rights)
+          r["rights"] = rights
+          r["rightsUri"] = normalize_id(rights)
         end
 
         attributes = {
-          'rightsURI' => r['rightsUri'],
-          'rightsIdentifier' => r['rightsIdentifier'],
-          'rightsIdentifierScheme' => r['rightsIdentifierScheme'],
-          'schemeURI' => r['schemeUri'],
-          'xml:lang' => r['lang']
+          "rightsURI" => r["rightsUri"],
+          "rightsIdentifier" => r["rightsIdentifier"],
+          "rightsIdentifierScheme" => r["rightsIdentifierScheme"],
+          "schemeURI" => r["schemeUri"],
+          "xml:lang" => r["lang"],
         }.compact
 
-        xml.rights(r['rights'], attributes)
+        xml.rights(r["rights"], attributes)
       end
     end
 
     def insert_crossref_issn(xml)
-      issn = if container.to_h.fetch('identifierType', nil) == 'ISSN'
-               container.to_h.fetch('identifier', nil)
-             end
+      issn = if container.to_h.fetch("identifierType", nil) == "ISSN"
+          container.to_h.fetch("identifier", nil)
+        end
 
       return xml if issn.blank?
 
@@ -329,22 +327,22 @@ module Commonmeta
         d = descriptions.first
       else
         d = {}
-        d['description'] = descriptions.first
+        d["description"] = descriptions.first
       end
 
-      xml.abstract('xmlns' => 'http://www.ncbi.nlm.nih.gov/JATS1') do
-        xml.p(d['description'])
+      xml.abstract("xmlns" => "http://www.ncbi.nlm.nih.gov/JATS1") do
+        xml.p(d["description"])
       end
     end
 
     def crossref_root_attributes
-      { 'xmlns:xsi': 'http://www.w3.org/2001/XMLSchema-instance',
-        'xsi:schemaLocation': 'http://www.crossref.org/schema/5.3.1 https://www.crossref.org/schemas/crossref5.3.1.xsd',
-        xmlns: 'http://www.crossref.org/schema/5.3.1',
-        'xmlns:jats': 'http://www.ncbi.nlm.nih.gov/JATS1',
-        'xmlns:fr': 'http://www.crossref.org/fundref.xsd',
-        'xmlns:mml': 'http://www.w3.org/1998/Math/MathML',
-        version: '5.3.1' }
+      { 'xmlns:xsi': "http://www.w3.org/2001/XMLSchema-instance",
+        'xsi:schemaLocation': "http://www.crossref.org/schema/5.3.1 https://www.crossref.org/schemas/crossref5.3.1.xsd",
+        xmlns: "http://www.crossref.org/schema/5.3.1",
+        'xmlns:jats': "http://www.ncbi.nlm.nih.gov/JATS1",
+        'xmlns:fr': "http://www.crossref.org/fundref.xsd",
+        'xmlns:mml': "http://www.w3.org/1998/Math/MathML",
+        version: "5.3.1" }
     end
   end
 end

data/lib/commonmeta/utils.rb

@@ -1367,19 +1367,27 @@ module Commonmeta
       end
     end
 
-    def encode_doi(prefix)
+    def encode_doi(prefix, options = {})
       # DOI suffix is a generated from a random number, encoded in base32
       # suffix has 8 digits plus two checksum digits. With base32 there are
       # 32 possible digits, so 8 digits gives 32^8 possible combinations
-      random_int = SecureRandom.random_number(32 ** 7..(32 ** 8) - 1)
-      suffix = Base32::URL.encode(random_int, checksum: true)
-      str = "#{suffix[0, 5]}-#{suffix[5, 10]}"
+      if options[:uuid]
+        str = Base32::URL.encode_uuid(options[:uuid], split: 7, checksum: true)
+      else
+        random_int = SecureRandom.random_number(32 ** 7..(32 ** 8) - 1)
+        suffix = Base32::URL.encode(random_int, checksum: true)
+        str = "#{suffix[0, 5]}-#{suffix[5, 10]}"
+      end
       "https://doi.org/#{prefix}/#{str}"
     end
 
-    def decode_doi(doi)
+    def decode_doi(doi, options = {})
       suffix = doi.split("/", 5).last
-      Base32::URL.decode(suffix)
+      if options[:uuid]
+        Base32::URL.decode_uuid(suffix)
+      else
+        Base32::URL.decode(suffix)
+      end
     end
 
     def encode_container_id

data/lib/commonmeta/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Commonmeta
-  VERSION = '3.2.8'
+  VERSION = '3.2.10'
 end

data/spec/utils_spec.rb

@@ -650,6 +650,15 @@ describe Commonmeta::Metadata, vcr: true do
       expect(response.length).to eq(36)
     end
 
+    it "encode doi from uuid" do
+      prefix = "10.53731"
+      uuid = "c3095752-2af0-40a4-a229-3ceb7424bce2"
+      response = subject.encode_doi(prefix, uuid: uuid)
+      puts response
+      expect(response).to match(%r{#{prefix}/[a-z0-9]+})
+      expect(response.length).to eq(56)
+    end
+
     it "decode doi" do
       doi = "https://doi.org/10.53731/revzwnv-rpd913d-8drwz"
       response = subject.decode_doi(doi)
@@ -661,6 +670,12 @@ describe Commonmeta::Metadata, vcr: true do
       response = subject.decode_doi(doi)
       expect(response).to eq(30_198_793_950_250_854_133_601_922_433)
     end
+
+    it "decode doi to uuid" do
+      doi = "https://doi.org/10.53731/6315bn4-aqg82ja-4a9wxdt-29f7279"
+      response = subject.decode_doi(doi, uuid: true)
+      expect(response).to eq('255d48ab-c102-9288-a4f3-add092f388e9')
+    end
   end
 
   context "random id" do

data/spec/writers/crossref_xml_writer_spec.rb

@@ -191,6 +191,7 @@ describe Commonmeta::Metadata, vcr: true do
                                          "person_name")).first).to eq("ORCID" => "https://orcid.org/0000-0001-9165-2757", "contributor_role" => "author", "given_name" => "Esha", "sequence" => "first", "surname" => "Datta")
       expect(crossref_xml.dig("titles",
                               "title")).to eq("Attempts at automating journal subject classification")
+      expect(crossref_xml.dig('item_number')).to eq("__content__"=>"2x2kzts-edc9rgb-vg6s4j6-z9t0d56", "item_number_type"=>"uuid")
     end
 
     it "json_feed_item with references" do
@@ -218,7 +219,7 @@ describe Commonmeta::Metadata, vcr: true do
                               "title")).to eq("The Research Software Alliance (ReSA)")
       expect(crossref_xml.dig("citation_list", "citation").length).to eq(11)
       expect(crossref_xml.dig("citation_list", "citation").last).to eq("doi"=>"10.5281/zenodo.3699950", "key"=>"ref11")
-      expect(crossref_xml.dig('publisher_item', 'item_number')).to eq("__content__"=>"954f81380ecd409087c5cef1297f1470", "item_number_type"=>"uuid")
+      expect(crossref_xml.dig('item_number')).to eq("__content__"=>"4n9y0kg-3pd8288-fheey4m-qy53g93", "item_number_type"=>"uuid")
     end
 
     it "json_feed_item from rogue scholar with doi" do
@@ -242,6 +243,7 @@ describe Commonmeta::Metadata, vcr: true do
                                          "person_name")).first).to eq("ORCID"=>"https://orcid.org/0000-0003-3334-2771", "contributor_role"=>"author", "given_name"=>"Heinz", "sequence"=>"first", "surname"=>"Pampel")
       expect(crossref_xml.dig("titles",
                               "title")).to eq("EU-Mitgliedstaaten betonen die Rolle von wissenschaftsgeleiteten Open-Access-Modellen jenseits von APCs")
+      expect(crossref_xml.dig('item_number')).to eq("__content__"=>"way2ng-4s48j9v-hbw4rjf-aqh9f33", "item_number_type"=>"uuid")
     end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: commonmeta-ruby
 version: !ruby/object:Gem::Version
-  version: 3.2.8
+  version: 3.2.10
 platform: ruby
 authors:
 - Martin Fenner
@@ -56,7 +56,7 @@ dependencies:
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 0.5.0
+        version: 0.7.0
     - - "<"
       - !ruby/object:Gem::Version
         version: '1'
@@ -66,7 +66,7 @@ dependencies:
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 0.5.0
+        version: 0.7.0
     - - "<"
       - !ruby/object:Gem::Version
         version: '1'