bolognese 1.11.3 → 2.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 84e06f3ca24650902cd1d96f777b2103a242fc44346d9995667961f090c1e300
-  data.tar.gz: 047d14ab54b8a9aa0569bdaeb90806a8a835809f399693c2b29c0a97e9eb0658
+  metadata.gz: ce52f2eb08396d0e56bdccf06702901fe7c04f1925cacafc71f315a1edda91c3
+  data.tar.gz: 29fb5a1d987a95220bbc7a3912aa24735f54924f47c8a74d61d1f51e25b75314
 SHA512:
-  metadata.gz: 2547f9e0c41915163269e3802c32c2f73abf6977987748eb70eec7d324585487585a3952a8874bb40e35e4d65d748e6968c0d1f7d92f76992ee3068b2cd718dd
-  data.tar.gz: 3beb48ca299e9b0a1ab95e7e74fabd99e127c374a513b9b775644695e12ac489f3d35851bc1553bc01a593c2ce3c84bc37e8f5e2bd452f6ac561d431edd4f669
+  metadata.gz: 4d528d2597900dfb90410504833a83e79b1bda1974d9f77102071cecd3f71c3e1f899d8e59abe24cf1110469807cc0c6382c50a5a62571ceceb9278e7cfa37b7
+  data.tar.gz: 341d76ed37284899ea1e74e331026ba6d042d92037a260d7591b9a48b38b7c8a0b06a6908e97917b2c345c5be8b3ac6475cc63a5478891589d624f248137b315

data/.github/workflows/ci.yml

@@ -7,7 +7,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        ruby: ["2.6", "2.7", "3.0", "3.1"]
+        ruby: ["3.0", "3.1", "3.2", "3.3"]
     runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v3

data/.github/workflows/release.yml

@@ -11,10 +11,10 @@ jobs:
     runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v3
-    - name: Set up Ruby 2.6
+    - name: Set up Ruby 3.1.4
       uses: ruby/setup-ruby@v1
       with:
-        ruby-version: "2.6"
+        ruby-version: "3.1.4"
 
     - name: Build
       run: |

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (1.11.3)
+    bolognese (2.2.0)
       activesupport (>= 4.2.5)
       benchmark_methods (~> 0.7)
       bibtex-ruby (>= 5.1.0)
@@ -16,9 +16,9 @@ PATH
       json-ld-preloaded (~> 3.1, >= 3.1.3)
       jsonlint (~> 0.3.0)
       loofah (~> 2.0, >= 2.0.3)
-      maremma (>= 4.9.4, < 5)
+      maremma (~> 5.0)
       namae (~> 1.0)
-      nokogiri (>= 1.13.2, < 1.14)
+      nokogiri (~> 1.16, >= 1.16.2)
       oj (~> 3.10)
       oj_mimic_json (~> 1.0, >= 1.0.1)
       postrank-uri (~> 1.0, >= 1.0.18)
@@ -41,6 +41,7 @@ GEM
     bibtex-ruby (6.0.0)
       latex-decode (~> 0.0)
     builder (3.2.4)
+    byebug (11.1.3)
     citeproc (1.0.10)
       namae (~> 1.0)
     citeproc-ruby (1.1.14)
@@ -94,10 +95,10 @@ GEM
       concurrent-ruby (~> 1.0)
     iso8601 (0.9.1)
     json (2.6.2)
-    json-canonicalization (0.3.2)
-    json-ld (3.2.5)
+    json-canonicalization (0.3.1)
+    json-ld (3.2.4)
       htmlentities (~> 4.3)
-      json-canonicalization (~> 0.3, >= 0.3.2)
+      json-canonicalization (~> 0.3)
       link_header (~> 0.0, >= 0.0.8)
       multi_json (~> 1.15)
       rack (>= 2.2, < 4)
@@ -113,7 +114,7 @@ GEM
     loofah (2.21.3)
       crass (~> 1.0.2)
       nokogiri (>= 1.12.0)
-    maremma (4.9.9)
+    maremma (5.0.0)
       activesupport (>= 4.2.5)
       addressable (>= 2.3.6)
       builder (~> 3.2, >= 3.2.2)
@@ -124,17 +125,17 @@ GEM
       faraday-follow_redirects (~> 0.3.0)
       faraday-gzip (~> 0.1.0)
       faraday-multipart (~> 1.0.4)
-      nokogiri (>= 1.13.1, < 1.14.0)
+      nokogiri (~> 1.16, >= 1.16.2)
       oj (>= 2.8.3)
       oj_mimic_json (~> 1.0, >= 1.0.1)
     matrix (0.4.2)
-    mini_portile2 (2.8.0)
+    mini_portile2 (2.8.5)
     minitest (5.18.0)
     multi_json (1.15.0)
     multipart-post (2.3.0)
     namae (1.1.1)
-    nokogiri (1.13.9)
-      mini_portile2 (~> 2.8.0)
+    nokogiri (1.16.2)
+      mini_portile2 (~> 2.8.2)
       racc (~> 1.4)
     oj (3.14.2)
     oj_mimic_json (1.0.1)
@@ -222,7 +223,9 @@ PLATFORMS
 DEPENDENCIES
   bolognese!
   bundler (>= 1.0)
+  byebug
   hashdiff (>= 1.0.0.beta1, < 2.0.0)
+  json-canonicalization (= 0.3.1)
   rack-test (~> 0)
   rake (~> 12.0)
   rspec (~> 3.4)
@@ -232,4 +235,4 @@ DEPENDENCIES
   webmock (~> 3.0, >= 3.0.1)
 
 BUNDLED WITH
-   2.4.10
+   2.5.5

data/bolognese.gemspec

@@ -13,12 +13,12 @@ Gem::Specification.new do |s|
   s.version       = Bolognese::VERSION
   s.extra_rdoc_files = ["README.md"]
   s.license       = 'MIT'
-  s.required_ruby_version = ['>=2.3']
+  s.required_ruby_version = ['>=3.0']
 
   # Declare dependencies here, rather than in the Gemfile
-  s.add_dependency 'maremma', '>= 4.9.4', '< 5'
+  s.add_dependency 'maremma', '~> 5.0'
   #s.add_dependency 'faraday', '~> 0.17.3'
-  s.add_dependency 'nokogiri', '>= 1.13.2', '< 1.14'
+  s.add_dependency 'nokogiri', '~> 1.16', '>= 1.16.2'
   s.add_dependency 'loofah', '~> 2.0', '>= 2.0.3'
   s.add_dependency 'builder', '~> 3.2', '>= 3.2.2'
   s.add_dependency 'activesupport', '>= 4.2.5'
@@ -49,6 +49,8 @@ Gem::Specification.new do |s|
   s.add_development_dependency 'webmock', '~> 3.0', '>= 3.0.1'
   s.add_development_dependency 'simplecov', '0.17.1'
   s.add_development_dependency 'hashdiff', ['>= 1.0.0.beta1', '< 2.0.0']
+  s.add_development_dependency 'byebug'
+  s.add_development_dependency 'json-canonicalization', '0.3.1'
 
   s.require_paths = ["lib"]
   s.files       = `git ls-files`.split($/)

data/lib/bolognese/author_utils.rb

@@ -30,19 +30,20 @@ module Bolognese
       name_type = parse_attributes(author.fetch("creatorName", nil), content: "nameType", first: true) || parse_attributes(author.fetch("contributorName", nil), content: "nameType", first: true)
 
       name_identifiers = Array.wrap(author.fetch("nameIdentifier", nil)).map do |ni|
+        name_identifier = ni["__content__"].strip if ni["__content__"].present?
         if ni["nameIdentifierScheme"] == "ORCID"
           {
-            "nameIdentifier" => normalize_orcid(ni["__content__"]),
+            "nameIdentifier" => normalize_orcid(name_identifier),
             "schemeUri" => "https://orcid.org",
             "nameIdentifierScheme" => "ORCID" }.compact
         elsif ni["nameIdentifierScheme"] == "ROR"
           {
-            "nameIdentifier" => normalize_ror(ni["__content__"]),
+            "nameIdentifier" => normalize_ror(name_identifier),
             "schemeUri" => "https://ror.org",
             "nameIdentifierScheme" => "ROR" }.compact
         else
           {
-            "nameIdentifier" => ni["__content__"],
+            "nameIdentifier" => name_identifier,
             "schemeUri" => ni.fetch("schemeURI", nil),
             "nameIdentifierScheme" => ni["nameIdentifierScheme"] }.compact
         end

data/lib/bolognese/datacite_utils.rb

@@ -106,9 +106,19 @@ module Bolognese
         end
       end
     end
-
+    
     def insert_publisher(xml)
-      xml.publisher(publisher || container && container["title"])
+      if publisher.is_a?(Hash)
+        attributes = {
+          'publisherIdentifier' => publisher["publisherIdentifier"],
+          'publisherIdentifierScheme' => publisher["publisherIdentifierScheme"],
+          'schemeURI' => publisher["schemeUri"],
+          "xml:lang" => publisher["lang"]
+        }.compact
+        xml.publisher(publisher["name"] || container && container["title"], attributes)
+      else
+        xml.publisher(publisher || container && container["title"])
+      end
     end
 
     def insert_publication_year(xml)
@@ -171,7 +181,7 @@ module Bolognese
             s["subject"] = subject
           end
 
-          attributes = { "subjectScheme" => s["subjectScheme"], "schemeURI" => s["schemeUri"], "valueURI" => s["valueUri"], "xml:lang" => s["lang"] }.compact
+          attributes = { "subjectScheme" => s["subjectScheme"], "schemeURI" => s["schemeUri"], "valueURI" => s["valueUri"], "classificationCode" => s["classificationCode"], "xml:lang" => s["lang"] }.compact
 
           xml.subject(s["subject"], attributes)
         end
@@ -375,13 +385,11 @@ module Bolognese
               end
             end
             if geo_location["geoLocationPolygon"]
-              geo_location["geoLocationPolygon"].each do |geo_location_polygon|
-                xml.geoLocationPolygon do
-                  geo_location_polygon.each do |polygon_point|
-                    xml.polygonPoint do
-                      xml.pointLatitude(polygon_point.dig("polygonPoint", "pointLatitude"))
-                      xml.pointLongitude(polygon_point.dig("polygonPoint", "pointLongitude"))
-                    end
+              xml.geoLocationPolygon do
+                Array.wrap(geo_location["geoLocationPolygon"]).each do |polygon_point|
+                  xml.polygonPoint do
+                    xml.pointLatitude(polygon_point.dig("polygonPoint", "pointLatitude"))
+                    xml.pointLongitude(polygon_point.dig("polygonPoint", "pointLongitude"))
                   end
                 end
               end

data/lib/bolognese/metadata.rb

@@ -156,11 +156,6 @@ module Bolognese
       @descriptions ||= meta.fetch("descriptions", nil)
     end
 
-    def abstract_description
-      # Fetch the first description with descriptionType "Abstract"
-      @abstract_description ||= descriptions&.find { |d| d["descriptionType"] == "Abstract" }
-    end
-
     def rights_list
       @rights_list ||= meta.fetch("rights_list", nil)
     end
@@ -222,7 +217,7 @@ module Bolognese
     end
 
     def publisher
-      @publisher ||= meta.fetch("publisher", nil)
+      @publisher ||= normalize_publisher(meta["publisher"]) if meta.fetch("publisher", nil).present?
     end
 
     def identifiers

data/lib/bolognese/metadata_utils.rb

@@ -96,7 +96,7 @@ module Bolognese
       if container.present?
         container["title"]
       elsif types["citeproc"] == "article-journal"
-        publisher
+        publisher["name"] if publisher.present?
       else
         nil
       end
@@ -161,7 +161,7 @@ module Bolognese
         "volume" => container.to_h["volume"],
         "issue" => container.to_h["issue"],
         "page" => page,
-        "publisher" => publisher,
+        "publisher" => publisher["name"],
         "title" => parse_attributes(titles, content: "title", first: true),
         "URL" => url,
         "copyright" => Array.wrap(rights_list).map { |l| l["rights"] }.first,

data/lib/bolognese/readers/bibtex_reader.rb

@@ -86,11 +86,11 @@ module Bolognese
           "titles" => meta.try(:title).present? ? [{ "title" => meta.try(:title).to_s }] : [],
           "creators" => creators,
           "container" => container,
-          "publisher" => meta.try(:publisher).to_s.presence,
+          "publisher" => meta.try(:publisher).present? ? { "name" => meta.publisher.to_s } : nil,
           "related_identifiers" => related_identifiers,
           "dates" => dates,
           "publication_year" => publication_year,
-          "descriptions" => meta.try(:abstract).present? ? [{ "description" => meta.try(:abstract) && sanitize(meta.abstract.to_s).presence, "descriptionType" => "Abstract" }] : [],
+          "descriptions" => meta.try(:abstract).present? ? [{ "description" => meta.try(:abstract) && sanitize(meta.abstract.to_s, new_line: true).presence, "descriptionType" => "Abstract" }] : [],
           "rights_list" => rights_list,
           "state" => state
         }.merge(read_options)

data/lib/bolognese/readers/citeproc_reader.rb

@@ -107,7 +107,7 @@ module Bolognese
           "related_identifiers" => related_identifiers,
           "dates" => dates,
           "publication_year" => publication_year,
-          "descriptions" => meta.fetch("abstract", nil).present? ? [{ "description" => sanitize(meta.fetch("abstract")), "descriptionType" => "Abstract" }] : [],
+          "descriptions" => meta.fetch("abstract", nil).present? ? [{ "description" => sanitize(meta.fetch("abstract"), new_line: true), "descriptionType" => "Abstract" }] : [],
           "rights_list" => rights_list,
           "version_info" => meta.fetch("version", nil),
           "subjects" => subjects,

data/lib/bolognese/readers/codemeta_reader.rb

@@ -43,7 +43,7 @@ module Bolognese
         dates << { "date" => meta.fetch("dateCreated"), "dateType" => "Created" } if meta.fetch("dateCreated", nil).present?
         dates << { "date" => meta.fetch("dateModified"), "dateType" => "Updated" } if meta.fetch("dateModified", nil).present?
         publication_year = meta.fetch("datePublished")[0..3] if meta.fetch("datePublished", nil).present?
-        publisher = meta.fetch("publisher", nil)
+        publisher = { "name" => meta.fetch("publisher", nil) } if meta.fetch("publisher", nil).present?
         state = meta.present? || read_options.present? ? "findable" : "not_found"
         schema_org = meta.fetch("@type", nil)
         types = {
@@ -76,7 +76,7 @@ module Bolognese
           #{}"is_part_of" => is_part_of,
           "dates" => dates,
           "publication_year" => publication_year,
-          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description")), "descriptionType" => "Abstract" }] : nil,
+          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description"), new_line: true), "descriptionType" => "Abstract" }] : nil,
           "rights_list" => rights_list,
           "version_info" => meta.fetch("version", nil),
           "subjects" => subjects,

data/lib/bolognese/readers/crosscite_reader.rb

@@ -7,7 +7,10 @@ module Bolognese
         errors = jsonlint(string)
         return { "errors" => errors } if errors.present?
 
-        string.present? ? Maremma.from_json(string) : {}
+        crosscite = string.present? ? Maremma.from_json(string) : {}
+        crosscite["publisher"] = normalize_publisher(crosscite["publisher"]) if crosscite.fetch("publisher", nil).present?
+
+        crosscite
       end
     end
   end

data/lib/bolognese/readers/crossref_reader.rb

@@ -40,8 +40,7 @@ module Bolognese
         journal_metadata = nil
         journal_issue = {}
         journal_metadata = nil
-        publisher = query.dig("crm_item", 0)
-        publisher = nil unless publisher.is_a?(String)
+        publisher = query.dig("crm_item", 0).is_a?(String) ? { "name" => query.dig("crm_item", 0) } : nil
 
         case model
         when "book"
@@ -232,11 +231,11 @@ module Bolognese
 
       def crossref_description(bibliographic_metadata)
         abstract = Array.wrap(bibliographic_metadata.dig("abstract")).map do |r|
-          { "descriptionType" => "Abstract", "description" => sanitize(parse_attributes(r, content: 'p')) }.compact
+          { "descriptionType" => "Abstract", "description" => sanitize(parse_attributes(r, content: 'p'), new_line: true) }.compact
         end
 
         description = Array.wrap(bibliographic_metadata.dig("description")).map do |r|
-          { "descriptionType" => "Other", "description" => sanitize(parse_attributes(r)) }.compact
+          { "descriptionType" => "Other", "description" => sanitize(parse_attributes(r), new_line: true) }.compact
         end
 
         (abstract + description)

data/lib/bolognese/readers/datacite_json_reader.rb

@@ -7,7 +7,10 @@ module Bolognese
         errors = jsonlint(string)
         return { "errors" => errors } if errors.present?
 
-        string.present? ? Maremma.from_json(string).transform_keys! { |key| key.underscore } : {}
+        datacite_json = string.present? ? Maremma.from_json(string).transform_keys! { |key| key.underscore } : {}
+        datacite_json["publisher"] = normalize_publisher(datacite_json["publisher"]) if datacite_json.fetch("publisher", nil).present?
+
+        datacite_json
       end
     end
   end

data/lib/bolognese/readers/datacite_reader.rb

@@ -94,13 +94,29 @@ module Bolognese
 
         titles = get_titles(meta)
 
+        publisher = Array.wrap(meta.dig("publisher")).map do |r|
+          if r.blank?
+            nil
+          elsif r.is_a?(String)
+            { "name" => r.strip }
+          elsif r.is_a?(Hash)
+            {
+              "name" => r["__content__"].present? ? r["__content__"].strip : nil,
+              "publisherIdentifier" => r["publisherIdentifierScheme"] == "ROR" ? normalize_ror(r["publisherIdentifier"]) : r["publisherIdentifier"],
+              "publisherIdentifierScheme" => r["publisherIdentifierScheme"],
+              "schemeUri" => r["schemeURI"],
+              "lang" => r["lang"],
+            }.compact
+          end
+        end.compact.first
+
         descriptions = Array.wrap(meta.dig("descriptions", "description")).map do |r|
           if r.blank?
             nil
           elsif r.is_a?(String)
-            { "description" => sanitize(r), "descriptionType" => "Abstract" }
+            { "description" => sanitize(r, new_line: true), "descriptionType" => "Abstract" }
           elsif r.is_a?(Hash)
-            { "description" => sanitize(r["__content__"]), "descriptionType" => r["descriptionType"], "lang" => r["lang"] }.compact
+            { "description" => sanitize(r["__content__"], new_line: true), "descriptionType" => r["descriptionType"], "lang" => r["lang"] }.compact
           end
         end.compact
         rights_list = Array.wrap(meta.dig("rightsList", "rights")).map do |r|
@@ -287,7 +303,7 @@ module Bolognese
           "creators" => get_authors(Array.wrap(meta.dig("creators", "creator"))),
           "contributors" => get_authors(Array.wrap(meta.dig("contributors", "contributor"))),
           "container" => set_container(meta),
-          "publisher" => parse_attributes(meta.fetch("publisher", nil), first: true).to_s.strip.presence,
+          "publisher" => publisher,
           "agency" => "datacite",
           "funding_references" => funding_references,
           "dates" => dates,

data/lib/bolognese/readers/npm_reader.rb

@@ -103,7 +103,7 @@ module Bolognese
           #"related_identifiers" => related_identifiers,
           #"dates" => dates,
           #"publication_year" => publication_year,
-          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description")), "descriptionType" => "Abstract" }] : [],
+          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description"), new_line: true), "descriptionType" => "Abstract" }] : [],
           "rights_list" => rights_list,
           "version_info" => meta.fetch("version", nil),
           "subjects" => subjects

data/lib/bolognese/readers/ris_reader.rb

@@ -89,12 +89,12 @@ module Bolognese
           "url" => meta.fetch("UR", nil),
           "titles" => meta.fetch("T1", nil).present? ? [{ "title" => meta.fetch("T1", nil) }] : nil,
           "creators" => get_authors(author),
-          "publisher" => meta.fetch("PB", "(:unav)"),
+          "publisher" => { "name" => meta.fetch("PB", "(:unav)") },
           "container" => container,
           "related_identifiers" => related_identifiers,
           "dates" => dates,
           "publication_year" => publication_year,
-          "descriptions" => meta.fetch("AB", nil).present? ? [{ "description" => sanitize(meta.fetch("AB")), "descriptionType" => "Abstract" }] : nil,
+          "descriptions" => meta.fetch("AB", nil).present? ? [{ "description" => sanitize(meta.fetch("AB"), new_line: true), "descriptionType" => "Abstract" }] : nil,
           "subjects" => subjects,
           "language" => meta.fetch("LA", nil),
           "state" => state

data/lib/bolognese/readers/schema_org_reader.rb

@@ -74,7 +74,11 @@ module Bolognese
           creators = get_authors(from_schema_org_creators(Array.wrap(authors)))
         end
         contributors = get_authors(from_schema_org_contributors(Array.wrap(meta.fetch("editor", nil))))
-        publisher = parse_attributes(meta.fetch("publisher", nil), content: "name", first: true)
+
+        publisher = {
+          "name" => parse_attributes(meta.fetch("publisher", nil), content: "name", first: true),
+          "publisherIdentifier" => parse_attributes(meta.fetch("publisher", nil), content: "@id", first: true),
+        }.compact if meta.fetch("publisher", nil).present?
 
         ct = (schema_org == "Dataset") ? "includedInDataCatalog" : "Periodical"
         container = if meta.fetch(ct, nil).present?
@@ -180,7 +184,7 @@ module Bolognese
           "related_identifiers" => related_identifiers,
           "publication_year" => publication_year,
           "dates" => dates,
-          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description")), "descriptionType" => "Abstract" }] : nil,
+          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description"), new_line: true), "descriptionType" => "Abstract" }] : nil,
           "rights_list" => rights_list,
           "version_info" => meta.fetch("version", nil).to_s.presence,
           "subjects" => subjects,

data/lib/bolognese/utils.rb

@@ -600,12 +600,12 @@ module Bolognese
     end
 
     def validate_orcid(orcid)
-      orcid = Array(/\A(?:(?:http|https):\/\/(?:(?:www|sandbox)?\.)?orcid\.org\/)?(\d{4}[[:space:]-]\d{4}[[:space:]-]\d{4}[[:space:]-]\d{3}[0-9X]+)\z/.match(orcid)).last
+      orcid = Array(/\A(?:(?:http|https):\/\/(?:(?:www|sandbox)?\.)?orcid\.org\/)?(\d{4}[[:space:]-]\d{4}[[:space:]-]\d{4}[[:space:]-]\d{3}[0-9X]+)\/{0,1}\z/.match(orcid)).last
       orcid.gsub(/[[:space:]]/, "-") if orcid.present?
     end
 
     def validate_ror(ror)
-      Array(/^(?:(?:(?:http|https):\/\/)?ror\.org\/)?(0\w{6}\d{2})$/.match(ror)).last
+      Array(/^(?:(?:(?:http|https):\/\/)?ror\.org\/)?(0\w{6}\d{2})\/{0,1}$/.match(ror)).last
     end
 
     def validate_orcid_scheme(orcid_scheme)
@@ -773,6 +773,14 @@ module Bolognese
       nil
     end
 
+    def normalize_publisher(publisher)
+      if publisher.respond_to?(:to_hash)
+        publisher
+      elsif publisher.respond_to?(:to_str)
+        { "name" => publisher }
+      end
+    end
+
     def to_datacite_json(element, options={})
       a = Array.wrap(element).map do |e|
         e.inject({}) {|h, (k,v)| h[k.dasherize] = v; h }
@@ -1057,12 +1065,16 @@ module Bolognese
       custom_scrubber = Bolognese::WhitelistScrubber.new(options)
 
       if text.is_a?(String)
-        # remove excessive internal whitespace with squish
-        Loofah.scrub_fragment(text, custom_scrubber).to_s.squish
+        if options[:new_line]
+          # Remove multiple spaces, tabs, and other whitespace characters while preserving single spaces and new lines
+          Loofah.scrub_fragment(text, custom_scrubber).to_s.gsub(/[ \t]+/, ' ').strip
+        else
+          Loofah.scrub_fragment(text, custom_scrubber).to_s.squish
+        end
       elsif text.is_a?(Hash)
-        sanitize(text.fetch(content, nil))
+        sanitize(text.fetch(content, nil), new_line: options[:new_line])
       elsif text.is_a?(Array)
-        a = text.map { |e| e.is_a?(Hash) ? sanitize(e.fetch(content, nil)) : sanitize(e) }.uniq
+        a = text.map { |e| e.is_a?(Hash) ? sanitize(e.fetch(content, nil), new_line: options[:new_line]) : sanitize(e, new_line: options[:new_line]) }.uniq
         a = options[:first] ? a.first : a.unwrap
       else
         nil
@@ -1410,5 +1422,10 @@ module Bolognese
         }
       end
     end
+
+    def abstract_description
+      # Fetch the first description with descriptionType "Abstract"
+      descriptions&.find { |d| d["descriptionType"] == "Abstract" }
+    end
   end
 end

data/lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "1.11.3"
+  VERSION = "2.2.0"
 end

data/lib/bolognese/writers/bibtex_writer.rb

@@ -21,7 +21,7 @@ module Bolognese
           volume: container.to_h["volume"],
           issue: container.to_h["issue"],
           pages: pages,
-          publisher: publisher,
+          publisher: publisher["name"],
           year: publication_year,
           copyright: Array.wrap(rights_list).map { |l| l["rights"] }.first,
         }.compact

data/lib/bolognese/writers/codemeta_writer.rb

@@ -19,7 +19,7 @@ module Bolognese
           "tags" => subjects.present? ? Array.wrap(subjects).map { |k| parse_attributes(k, content: "subject", first: true) } : nil,
           "datePublished" => get_date(dates, "Issued") || publication_year,
           "dateModified" => get_date(dates, "Updated"),
-          "publisher" => publisher,
+          "publisher" => publisher["name"],
           "license" => Array.wrap(rights_list).map { |l| l["rightsUri"] }.compact.unwrap,
         }.compact
         JSON.pretty_generate hsh.presence

data/lib/bolognese/writers/csv_writer.rb

@@ -15,7 +15,7 @@ module Bolognese
           resource_type: types["resourceType"],
           title: parse_attributes(titles, content: "title", first: true),
           author: authors_as_string(creators),
-          publisher: publisher,
+          publisher: publisher["name"],
           publication_year: publication_year
         }.values
 

data/lib/bolognese/writers/datacite_json_writer.rb

@@ -4,7 +4,9 @@ module Bolognese
   module Writers
     module DataciteJsonWriter
       def datacite_json
-        JSON.pretty_generate crosscite_hsh.transform_keys! { |key| key.camelcase(uppercase_first_letter = :lower) } if crosscite_hsh.present?
+        if crosscite_hsh.present?
+          JSON.pretty_generate crosscite_hsh.transform_keys! { |key| key.camelcase(uppercase_first_letter = :lower) }
+        end
       end
     end
   end

data/lib/bolognese/writers/jats_writer.rb

@@ -77,16 +77,19 @@ module Bolognese
 
       def insert_source(xml)
         if is_chapter?
-          xml.source(publisher)
+          xml.source(publisher["name"])
         elsif is_article? || is_data?
-          xml.source(container && container["title"] || publisher)
+          xml.source(container && container["title"] || publisher["name"])
         else
           xml.source(parse_attributes(titles, content: "title", first: true))
         end
       end
 
       def insert_publisher_name(xml)
-        xml.send("publisher-name", publisher)
+        attributes = {
+          "xml:lang" => publisher["lang"]
+        }.compact
+        xml.send("publisher-name", attributes, publisher["name"])
       end
 
       def insert_publication_date(xml)

data/lib/bolognese/writers/ris_writer.rb

@@ -14,7 +14,7 @@ module Bolognese
           "AB" => parse_attributes(abstract_description, content: "description", first: true),
           "KW" => Array.wrap(subjects).map { |k| parse_attributes(k, content: "subject", first: true) }.presence,
           "PY" => publication_year,
-          "PB" => publisher,
+          "PB" => publisher["name"],
           "LA" => language,
           "VL" => container.to_h["volume"],
           "IS" => container.to_h["issue"],

data/lib/bolognese/writers/schema_org_writer.rb

@@ -37,7 +37,7 @@ module Bolognese
           "schemaVersion" => schema_version,
           "periodical" => types.present? ? ((types["schemaOrg"] != "Dataset") && container.present? ? to_schema_org(container) : nil) : nil,
           "includedInDataCatalog" => types.present? ? ((types["schemaOrg"] == "Dataset") && container.present? ? to_schema_org_container(container, type: "Dataset") : nil) : nil,
-          "publisher" => publisher.present? ? { "@type" => "Organization", "name" => publisher } : nil,
+          "publisher" => publisher.present? ? { "@type" => "Organization", "@id" => publisher["publisherIdentifier"], "name" => publisher["name"] }.compact : nil,
           "funder" => to_schema_org_funder(funding_references),
           "provider" => agency.present? ? { "@type" => "Organization", "name" => agency } : nil
         }.compact.presence

data/resources/kernel-4/include/datacite-relationType-v4.xsd

@@ -46,6 +46,8 @@
       <xs:enumeration value="IsRequiredBy" />
       <xs:enumeration value="Obsoletes" />
       <xs:enumeration value="IsObsoletedBy" />
+      <xs:enumeration value="Collects" />
+      <xs:enumeration value="IsCollectedBy" />
     </xs:restriction>
   </xs:simpleType>
 </xs:schema>

data/resources/kernel-4/include/datacite-resourceType-v4.xsd

@@ -22,6 +22,7 @@
       <xs:enumeration value="Dissertation" />
       <xs:enumeration value="Event" />
       <xs:enumeration value="Image" />
+      <xs:enumeration value="Instrument" />
       <xs:enumeration value="InteractiveResource" />
       <xs:enumeration value="Journal" />
       <xs:enumeration value="JournalArticle" />
@@ -35,6 +36,7 @@
       <xs:enumeration value="Software" />
       <xs:enumeration value="Sound" />
       <xs:enumeration value="Standard" />
+      <xs:enumeration value="StudyRegistration" />
       <xs:enumeration value="Text" />
       <xs:enumeration value="Workflow" />
       <xs:enumeration value="Other" />