bolognese 1.6 → 1.6.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1a22da55003185429ddc212b5518e74cc414d94d5f87b8047f2ed07859bc8f0d
-  data.tar.gz: adcf4f5f25e7c820b2c23bf0f685f5af1e09c3ba3cd1ec4b850f180fe6919aa3
+  metadata.gz: 405a39bf0274cc91c5356f94b050f5e0bad7598d47601a3fac9a99aed3900eff
+  data.tar.gz: e32e05bbf8fdd50af362081e7dee9de0c6e64b21c12d4fadd676909808423955
 SHA512:
-  metadata.gz: 818dba03b12581b2345bf7da3728e99ed713bd16cf0367ccc61aeb072049564ad59f7e56b58d5e68a0c68d40eeafcb0402307721c5b42d8cfeb2c3f2a0fdc96a
-  data.tar.gz: 4dd76c4414a071860b76f108c5ef4145c188a1f909816c2d860c33dfc80b1dc55cffebd5060affabad5b79a6e2b88d61a0fe7107b3d133bfa0905cb866bc6199
+  metadata.gz: 1fc29ec5b40985218e1305d7e70eac714f8250d9f6c3b308b194785c6d3c8426facaf2b11c9332a5a1deb72fa6eb3084b60aaf58527fcb638b081ce08d55c322
+  data.tar.gz: 6a82235d1c16f6bf987c9a9ebd61337100ae7b8bc4c64d343d0de4f0037d399081f71a8bee68754937d482d0c5d7bf4f91f06a0afe55178818fe0a85e404c4f1

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (1.6)
+    bolognese (1.6.6)
       activesupport (>= 4.2.5)
       benchmark_methods (~> 0.7)
       bibtex-ruby (>= 5.1.0)
@@ -11,10 +11,10 @@ PATH
       concurrent-ruby (~> 1.1, >= 1.1.5)
       csl-styles (~> 1.0, >= 1.0.1.8)
       edtf (~> 3.0, >= 3.0.4)
-      faraday (= 0.17.0)
+      faraday (~> 1.0, >= 1.0.1)
       gender_detector (~> 0.1.2)
       iso8601 (~> 0.9.1)
-      json-ld (~> 2.2, >= 2.2.1)
+      json-ld (~> 3.1, >= 3.1.4)
       jsonlint (~> 0.3.0)
       loofah (~> 2.0, >= 2.0.3)
       maremma (>= 4.3, < 5)
@@ -25,12 +25,12 @@ PATH
       postrank-uri (~> 1.0, >= 1.0.18)
       rdf-rdfxml (~> 3.1)
       rdf-turtle (~> 3.1)
-      thor (~> 0.19)
+      thor (~> 1.0, >= 1.0.1)
 
 GEM
   remote: https://rubygems.org/
   specs:
-    activesupport (6.0.3)
+    activesupport (6.0.3.2)
       concurrent-ruby (~> 1.0, >= 1.0.2)
       i18n (>= 0.7, < 2)
       minitest (~> 5.1)
@@ -56,7 +56,7 @@ GEM
       namae (~> 1.0)
     csl-styles (1.0.1.10)
       csl (~> 1.0)
-    diff-lcs (1.3)
+    diff-lcs (1.4.1)
     docile (1.3.2)
     ebnf (1.2.0)
       rdf (~> 3.1)
@@ -64,12 +64,12 @@ GEM
     edtf (3.0.5)
       activesupport (>= 3.0, < 7.0)
     excon (0.71.1)
-    faraday (0.17.0)
+    faraday (1.0.1)
       multipart-post (>= 1.2, < 3)
     faraday-encoding (0.0.5)
       faraday
-    faraday_middleware (0.13.1)
-      faraday (>= 0.7.4, < 1.0)
+    faraday_middleware (1.0.0)
+      faraday (~> 1.0)
     gender_detector (0.1.2)
       unicode_utils (>= 1.3.0)
     haml (5.1.2)
@@ -79,28 +79,33 @@ GEM
       concurrent-ruby (~> 1.0)
     hashdiff (1.0.1)
     htmlentities (4.3.4)
-    i18n (1.8.2)
+    i18n (1.8.3)
       concurrent-ruby (~> 1.0)
     iso8601 (0.9.1)
-    json-ld (2.2.1)
-      multi_json (~> 1.12)
-      rdf (>= 2.2.8, < 4.0)
+    json-canonicalization (0.2.0)
+    json-ld (3.1.4)
+      htmlentities (~> 4.3)
+      json-canonicalization (~> 0.2)
+      link_header (~> 0.0, >= 0.0.8)
+      multi_json (~> 1.14)
+      rack (~> 2.0)
+      rdf (~> 3.1)
     jsonlint (0.3.0)
       oj (~> 3)
       optimist (~> 3)
     latex-decode (0.3.1)
     link_header (0.0.8)
-    loofah (2.5.0)
+    loofah (2.6.0)
       crass (~> 1.0.2)
       nokogiri (>= 1.5.9)
-    maremma (4.7)
+    maremma (4.8)
       activesupport (>= 4.2.5)
       addressable (>= 2.3.6)
       builder (~> 3.2, >= 3.2.2)
       excon (~> 0.71.0)
-      faraday (= 0.17.0)
+      faraday (~> 1.0, >= 1.0.1)
       faraday-encoding (~> 0.0.4)
-      faraday_middleware (~> 0.13.1)
+      faraday_middleware (~> 1.0)
       nokogiri (~> 1.10.4)
       oj (>= 2.8.3)
       oj_mimic_json (~> 1.0, >= 1.0.1)
@@ -119,29 +124,32 @@ GEM
       nokogiri (>= 1.8.0)
       public_suffix (>= 2.0.0, < 2.1)
     public_suffix (2.0.5)
-    rack (2.2.2)
+    rack (2.2.3)
     rack-test (0.8.3)
       rack (>= 1.0, < 3)
     rake (12.3.3)
-    rdf (3.1.1)
+    rdf (3.1.3)
       hamster (~> 3.0)
       link_header (~> 0.0, >= 0.0.8)
     rdf-aggregate-repo (3.1.0)
       rdf (~> 3.1)
-    rdf-rdfa (3.1.0)
+    rdf-rdfa (3.1.1)
       haml (~> 5.1)
       htmlentities (~> 4.3)
-      rdf (~> 3.1)
+      rdf (~> 3.1, >= 3.1.2)
       rdf-aggregate-repo (~> 3.1)
+      rdf-vocab (~> 3.1, >= 3.1.5)
       rdf-xsd (~> 3.1)
     rdf-rdfxml (3.1.0)
       htmlentities (~> 4.3)
       rdf (~> 3.1)
       rdf-rdfa (~> 3.1)
       rdf-xsd (~> 3.1)
-    rdf-turtle (3.1.0)
+    rdf-turtle (3.1.1)
       ebnf (~> 1.2)
-      rdf (~> 3.1)
+      rdf (~> 3.1, >= 3.1.2)
+    rdf-vocab (3.1.5)
+      rdf (~> 3.1, >= 3.1.2)
     rdf-xsd (3.1.0)
       rdf (~> 3.1)
     rspec (3.9.0)
@@ -168,7 +176,7 @@ GEM
     sxp (1.1.0)
       rdf (~> 3.1)
     temple (0.8.2)
-    thor (0.20.3)
+    thor (1.0.1)
     thread_safe (0.3.6)
     tilt (2.0.10)
     tzinfo (1.2.7)

data/bolognese.gemspec

@@ -17,13 +17,13 @@ Gem::Specification.new do |s|
 
   # Declary dependencies here, rather than in the Gemfile
   s.add_dependency 'maremma', '>= 4.3', '< 5'
-  s.add_dependency 'faraday', '0.17.0'
+  s.add_dependency 'faraday', '~> 1.0', '>= 1.0.1'
   s.add_dependency 'nokogiri', '~> 1.10.4'
   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'
   s.add_dependency 'bibtex-ruby', '>= 5.1.0'
-  s.add_dependency 'thor', '~> 0.19'
+  s.add_dependency 'thor', '~> 1.0', '>= 1.0.1'
   s.add_dependency 'colorize', '~> 0.8.1'
   s.add_dependency 'namae', '~> 1.0'
   s.add_dependency 'edtf', '~> 3.0', '>= 3.0.4'
@@ -31,7 +31,7 @@ Gem::Specification.new do |s|
   s.add_dependency 'csl-styles', '~> 1.0', '>= 1.0.1.8'
   s.add_dependency 'iso8601', '~> 0.9.1'
   s.add_dependency 'postrank-uri', '~> 1.0', '>= 1.0.18'
-  s.add_dependency 'json-ld', '~> 2.2', '>= 2.2.1'
+  s.add_dependency 'json-ld', '~> 3.1', '>= 3.1.4'
   s.add_dependency 'jsonlint', '~> 0.3.0'
   s.add_dependency 'oj', '~> 3.10'
   s.add_dependency "oj_mimic_json", "~> 1.0", ">= 1.0.1"

data/lib/bolognese/datacite_utils.rb

@@ -172,9 +172,9 @@ module Bolognese
     end
 
     def insert_version(xml)
-      return xml unless version.present?
+      return xml unless version_info.present?
 
-      xml.version(version)
+      xml.version(version_info)
     end
 
 
@@ -232,7 +232,13 @@ module Bolognese
             r["rightsUri"] = normalize_id(rights)
           end
 
-          attributes = { 'rightsURI' => r["rightsUri"], 'xml:lang' => r["lang"] }.compact
+          attributes = { 
+            "rightsURI" => r["rightsUri"],
+            "rightsIdentifier" => r["rightsIdentifier"],
+            "rightsIdentifierScheme" => r["rightsIdentifierScheme"],
+            "schemeURI" => r["schemeUri"],
+            "xml:lang" => r["lang"] 
+          }.compact
 
           xml.rights(r["rights"], attributes)
         end

data/lib/bolognese/metadata.rb

@@ -9,7 +9,7 @@ module Bolognese
     attr_accessor :string, :from, :sandbox, :meta, :regenerate, :issue, :show_errors
     attr_reader :doc, :page_start, :page_end
     attr_writer :id, :provider_id, :client_id, :doi, :identifiers, :creators, :contributors, :titles, :publisher, 
-                :rights_list, :dates, :publication_year, :volume, :url, :version,
+                :rights_list, :dates, :publication_year, :volume, :url, :version_info,
                 :subjects, :contributor, :descriptions, :language, :sizes,
                 :formats, :schema_version, :meta, :container, :agency,
                 :format, :funding_references, :state, :geo_locations,
@@ -106,7 +106,7 @@ module Bolognese
         :publication_year,
         :descriptions,
         :rights_list,
-        :version,
+        :version_info,
         :subjects,
         :language,
         :geo_locations,
@@ -117,7 +117,8 @@ module Bolognese
 
       @regenerate = options[:regenerate] || read_options.present?
       # generate name for method to call dynamically
-      @meta = @from.present? ? send("read_" + @from, { string: string, sandbox: options[:sandbox], doi: options[:doi], id: id, ra: ra }.merge(read_options)) : {}
+      opts = { string: string, sandbox: options[:sandbox], doi: options[:doi], id: id, ra: ra }.merge(read_options)
+      @meta = @from.present? ? send("read_" + @from, **opts) : {}
     end
 
     def id
@@ -189,13 +190,10 @@ module Bolognese
       @url ||= meta.fetch("url", nil)
     end
 
-    def version
-      @version ||= meta.fetch("version", nil)
+    def version_info
+      @version_info ||= meta.fetch("version_info", nil) || meta.fetch("version", nil)
     end
 
-    # for backwards compatibility
-    alias_attribute :version_info, :version
-
     def publication_year
       @publication_year ||= meta.fetch("publication_year", nil)
     end

data/lib/bolognese/metadata_utils.rb

@@ -140,8 +140,14 @@ module Bolognese
         author = to_citeproc(creators)
       end
 
+      if types["resourceTypeGeneral"] == "Software" && version_info.present?
+        type = "book"
+      else
+        type = types["citeproc"]
+      end
+
       {
-        "type" => types["citeproc"],
+        "type" => type,
         "id" => normalize_doi(doi),
         "categories" => Array.wrap(subjects).map { |k| parse_attributes(k, content: "subject", first: true) }.presence,
         "language" => language,
@@ -158,7 +164,8 @@ module Bolognese
         "publisher" => publisher,
         "title" => parse_attributes(titles, content: "title", first: true),
         "URL" => url,
-        "version" => version
+        "copyright" => Array.wrap(rights_list).map { |l| l["rights"] }.first,
+        "version" => version_info
       }.compact.symbolize_keys
     end
 
@@ -180,7 +187,7 @@ module Bolognese
         "identifiers" => identifiers,
         "sizes" => sizes,
         "formats" => formats,
-        "version" => version,
+        "version" => version_info,
         "rights_list" => rights_list,
         "descriptions" => descriptions,
         "geo_locations" => geo_locations,

data/lib/bolognese/readers/bibtex_reader.rb

@@ -76,6 +76,7 @@ module Bolognese
                     "dateType" => "Issued" }]
                 end
         publication_year =  meta.try(:date).present? ? meta.date.to_s[0..3] : nil
+        rights_list = meta.try(:copyright).present? ? [hsh_to_spdx("rightsURI" => meta[:copyright])] : []
 
         { "id" => normalize_doi(doi),
           "types" => types,
@@ -90,7 +91,7 @@ module Bolognese
           "dates" => dates,
           "publication_year" => publication_year,
           "descriptions" => meta.try(:abstract).present? ? [{ "description" => meta.try(:abstract) && sanitize(meta.abstract.to_s).presence, "descriptionType" => "Abstract" }] : [],
-          "rights_list" =>  meta.try(:copyright).present? ? [{ "rightsUri" => meta.try(:copyright).to_s.presence }.compact] : [],
+          "rights_list" => rights_list,
           "state" => state
         }.merge(read_options)
       end

data/lib/bolognese/readers/citeproc_reader.rb

@@ -60,7 +60,7 @@ module Bolognese
                 end
         publication_year = get_date_from_date_parts(meta.fetch("issued", nil)).to_s[0..3]
         rights_list = if meta.fetch("copyright", nil)
-                        [{ "rightsUri" => normalize_url(meta.fetch("copyright")) }.compact]
+                        [hsh_to_spdx("rightsURI" => meta.fetch("copyright"))]
                       end
         related_identifiers = if meta.fetch("container-title", nil).present? && meta.fetch("ISSN", nil).present?
                                 [{ "type" => "Periodical",
@@ -100,8 +100,10 @@ module Bolognese
         doi = Array.wrap(identifiers).find { |r| r["identifierType"] == "DOI" }.to_h.fetch("identifier", nil)
           
         state = id.present? || read_options.present? ? "findable" : "not_found"
-        subjects = Array.wrap(meta.fetch("categories", nil)).map do |s|
-          { "subject" => s }
+        subjects = Array.wrap(meta.fetch("categories", nil)).reduce([]) do |sum, subject|
+          sum += name_to_fos(subject)
+
+          sum
         end
 
         { "id" => id,
@@ -119,7 +121,7 @@ module Bolognese
           "publication_year" => publication_year,
           "descriptions" => meta.fetch("abstract", nil).present? ? [{ "description" => sanitize(meta.fetch("abstract")), "descriptionType" => "Abstract" }] : [],
           "rights_list" => rights_list,
-          "version" => meta.fetch("version", nil),
+          "version_info" => meta.fetch("version", nil),
           "subjects" => subjects,
           "state" => state
         }.merge(read_options)

data/lib/bolognese/readers/codemeta_reader.rb

@@ -57,13 +57,15 @@ module Bolognese
           "bibtex" => Bolognese::Utils::SO_TO_BIB_TRANSLATIONS[schema_org] || "misc",
           "ris" => Bolognese::Utils::SO_TO_RIS_TRANSLATIONS[schema_org] || "GEN"
         }.compact
-        subjects = Array.wrap(meta.fetch("tags", nil)).map do |s|
-          { "subject" => s }
+        subjects = Array.wrap(meta.fetch("tags", nil)).reduce([]) do |sum, subject|
+          sum += name_to_fos(subject)
+
+          sum
         end
 
         has_title = meta.fetch("title", nil)
-
         titles =  has_title.nil? ?  [{ "title" => meta.fetch("name", nil) }] : [{ "title" => has_title }]  
+        rights_list = meta.fetch("licenseId", nil).present? ? [hsh_to_spdx("rightsIdentifier" => meta.fetch("licenseId"))] : nil
 
         { "id" => id,
           "types" => types,
@@ -78,8 +80,8 @@ module Bolognese
           "dates" => dates,
           "publication_year" => publication_year,
           "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description")), "descriptionType" => "Abstract" }] : nil,
-          "rights_list" => [{ "rightsUri" => meta.fetch("license", nil) }.compact],
-          "version" => meta.fetch("version", nil),
+          "rights_list" => rights_list,
+          "version_info" => meta.fetch("version", nil),
           "subjects" => subjects,
           "state" => state
         }.merge(read_options)

data/lib/bolognese/readers/crossref_reader.rb

@@ -172,7 +172,7 @@ module Bolognese
           "publication_year" => publication_year,
           "descriptions" => crossref_description(bibliographic_metadata),
           "rights_list" => crossref_license(program_metadata),
-          "version" => nil,
+          "version_info" => nil,
           "subjects" => nil,
           "language" => nil,
           "sizes" => nil,
@@ -211,7 +211,7 @@ module Bolognese
         access_indicator = Array.wrap(program_metadata).find { |m| m["name"] == "AccessIndicators" }
         if access_indicator.present?
           Array.wrap(access_indicator["license_ref"]).map do |license|
-            { "rightsUri" => normalize_url(parse_attributes(license)) }
+            hsh_to_spdx("rightsURI" => parse_attributes(license))
           end.uniq
         else
           []

data/lib/bolognese/readers/datacite_reader.rb

@@ -93,7 +93,7 @@ module Bolognese
           "bibtex" => Bolognese::Utils::CR_TO_BIB_TRANSLATIONS[resource_type.to_s.underscore.camelcase] || Bolognese::Utils::SO_TO_BIB_TRANSLATIONS[schema_org] || "misc",
           "ris" => Bolognese::Utils::CR_TO_RIS_TRANSLATIONS[resource_type.to_s.underscore.camelcase] || Bolognese::Utils::DC_TO_RIS_TRANSLATIONS[resource_type_general.to_s.dasherize] || "GEN"
         }.compact
-        
+
         titles = Array.wrap(meta.dig("titles", "title")).map do |r|
           if r.blank?
             nil
@@ -117,20 +117,22 @@ module Bolognese
           if r.blank?
             nil
           elsif r.is_a?(String)
-            { "rights" => r }
+            name_to_spdx(r)
           elsif r.is_a?(Hash)
-            { "rights" => r["__content__"], "rightsUri" => normalize_url(r["rightsURI"]), "lang" => r["lang"] }.compact
+            hsh_to_spdx(r)
           end
         end.compact
-        subjects = Array.wrap(meta.dig("subjects", "subject")).map do |k|
-          if k.blank?
-            nil
-          elsif k.is_a?(String)
-            { "subject" => sanitize(k) }
-          elsif k.is_a?(Hash)
-            { "subject" => sanitize(k["__content__"]), "subjectScheme" => k["subjectScheme"], "schemeUri" => k["schemeURI"], "valueUri" => k["valueURI"], "lang" => k["lang"] }.compact
+        
+        subjects = Array.wrap(meta.dig("subjects", "subject")).reduce([]) do |sum, subject|
+          if subject.is_a?(String)
+            sum += name_to_fos(subject)
+          elsif subject.is_a?(Hash)
+            sum += hsh_to_fos(subject)
           end
-        end.compact
+
+          sum
+        end.uniq
+
         dates = Array.wrap(meta.dig("dates", "date")).map do |r|
           if r.is_a?(Hash) && date = sanitize(r["__content__"]).presence
             if Date.edtf(date).present? || Bolognese::Utils::UNKNOWN_INFORMATION.key?(date)
@@ -164,12 +166,15 @@ module Bolognese
         funding_references = Array.wrap(meta.dig("fundingReferences", "fundingReference")).compact.map do |fr|
           scheme_uri = parse_attributes(fr["funderIdentifier"], content: "schemeURI")
           funder_identifier = parse_attributes(fr["funderIdentifier"])
-          funder_identifier = !funder_identifier.to_s.start_with?("https://","http://") && scheme_uri.present? ? normalize_id(scheme_uri + funder_identifier) : normalize_id(funder_identifier)
-          
+          funder_identifier_type = parse_attributes(fr["funderIdentifier"], content: "funderIdentifierType")
+          if funder_identifier_type != "Other"
+            funder_identifier = !funder_identifier.to_s.start_with?("https://","http://") && scheme_uri.present? ? normalize_id(scheme_uri + funder_identifier) : normalize_id(funder_identifier)
+          end
+
           {
             "funderName" => fr["funderName"],
             "funderIdentifier" => funder_identifier,
-            "funderIdentifierType" => parse_attributes(fr["funderIdentifier"], content: "funderIdentifierType"),
+            "funderIdentifierType" => funder_identifier_type,
             "awardNumber" => parse_attributes(fr["awardNumber"]),
             "awardUri" => parse_attributes(fr["awardNumber"], content: "awardURI"),
             "awardTitle" => fr["awardTitle"] }.compact
@@ -181,7 +186,7 @@ module Bolognese
             rid = ri["__content__"]
           end
 
-          { 
+          {
             "relatedIdentifier" => rid,
             "relatedIdentifierType" => ri["relatedIdentifierType"],
             "relationType" => ri["relationType"],
@@ -230,7 +235,7 @@ module Bolognese
           "publication_year" => parse_attributes(meta.fetch("publicationYear", nil), first: true).to_s.strip.presence,
           "descriptions" => descriptions,
           "rights_list" => Array.wrap(rights_list),
-          "version" => meta.fetch("version", nil).to_s.presence,
+          "version_info" => meta.fetch("version", nil).to_s.presence,
           "subjects" => subjects,
           "language" => parse_attributes(meta.fetch("language", nil), first: true).to_s.strip.presence,
           "geo_locations" => geo_locations,