bolognese 1.9.18 → 2.3.9

data/lib/bolognese/doi_utils.rb

@@ -2,6 +2,10 @@
 
 module Bolognese
   module DoiUtils
+    class << self
+      include DoiUtils
+    end
+
     def validate_doi(doi)
       doi = Array(/\A(?:(http|https):\/(\/)?(dx\.)?(doi.org|handle.stage.datacite.org|handle.test.datacite.org)\/)?(doi:)?(10\.\d{4,5}\/.+)\z/.match(doi)).last
       # remove non-printing whitespace and downcase

data/lib/bolognese/metadata.rb

@@ -217,7 +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
@@ -153,6 +153,7 @@ module Bolognese
         "language" => language,
         "author" => author,
         "contributor" => to_citeproc(contributors),
+        "translator" => contributors ? to_citeproc(contributors.select { |c| c["contributorType"] == "Translator" }) : nil,
         "issued" => get_date(dates, "Issued") ? get_date_parts(get_date(dates, "Issued")) : get_date_parts(publication_year.to_s),
         "submitted" => Array.wrap(dates).find { |d| d["dateType"] == "Submitted" }.to_h.fetch("__content__", nil),
         "abstract" => parse_attributes(descriptions, content: "description", first: true),
@@ -161,7 +162,7 @@ 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

@@ -52,6 +52,12 @@ module Bolognese
           [{ "nameType" => "Organizational", "name" => ":(unav)" }]
         end
         contributors = get_authors(from_citeproc(Array.wrap(meta.fetch("editor", nil))))
+        translators = get_authors(from_citeproc(Array.wrap(meta.fetch("translator", nil))))
+        translators.each do |translator|
+          translator["contributorType"] = "Translator"
+        end
+        contributors += translators
+
         dates = if date = get_date_from_date_parts(meta.fetch("issued", nil))
                   if Date.edtf(date).present?
                     [{ "date" => date,
@@ -107,7 +113,7 @@ 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

@@ -4,6 +4,12 @@ module Bolognese
   module Readers
     module CrossrefReader
       # CrossRef types from https://api.crossref.org/types
+
+      CR_TO_DC_CONTRIBUTOR_TYPES = {
+        "editor" => "Editor",
+        "translator" => "Translator",
+      }
+
       def get_crossref(id: nil, **options)
         return { "string" => nil, "state" => "not_found" } unless id.present?
 
@@ -40,8 +46,7 @@ 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"
@@ -139,7 +144,7 @@ module Bolognese
 
         state = meta.present? || read_options.present? ? "findable" : "not_found"
 
-        related_identifiers = Array.wrap(crossref_is_part_of(journal_metadata)) + Array.wrap(crossref_references(bibliographic_metadata))
+        related_identifiers = Array.wrap(crossref_is_part_of(journal_metadata)) + Array.wrap(crossref_references(bibliographic_metadata)) + Array.wrap(crossref_has_translation(program_metadata)) + Array.wrap(crossref_is_translation_of(program_metadata))
 
         container = if journal_metadata.present?
                       issn = normalize_issn(journal_metadata.to_h.fetch("issn", nil))
@@ -188,7 +193,7 @@ module Bolognese
           "titles" => titles,
           "identifiers" => identifiers,
           "creators" => crossref_people(bibliographic_metadata, "author"),
-          "contributors" => crossref_people(bibliographic_metadata, "editor"),
+          "contributors" => crossref_people(bibliographic_metadata, "editor") + crossref_people(bibliographic_metadata, "translator"),
           "funding_references" => crossref_funding_reference(program_metadata),
           "publisher" => publisher,
           "container" => container,
@@ -232,11 +237,11 @@ 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)
@@ -283,10 +288,11 @@ module Bolognese
               "givenName" => given_name,
               "familyName" => family_name,
               "affiliation" => affiliation.presence,
-              "contributorType" => contributor_role == "editor" ? "Editor" : nil }.compact
+              "contributorType" => contributor_role == 'author' ? nil : CR_TO_DC_CONTRIBUTOR_TYPES[a["contributor_role"]] }.compact
           else
             { "nameType" => "Organizational",
-              "name" => a["name"] || a["__content__"] }
+              "name" => a["name"] || a["__content__"],
+              "contributorType" => contributor_role == 'author' ? nil : CR_TO_DC_CONTRIBUTOR_TYPES[a["contributor_role"]] }.compact
           end
         end
       end
@@ -363,6 +369,32 @@ module Bolognese
           end
         end.compact.unwrap
       end
+    
+      def crossref_has_translation(program_metadata)
+        refs = program_metadata.dig("related_item") if program_metadata.is_a?(Hash)
+        Array.wrap(refs).select { |a| a.dig("intra_work_relation", "relationship_type") == "hasTranslation" }.map do |c|
+          if c.dig("intra_work_relation", "identifier_type") == "doi"
+            { "relatedIdentifier" => parse_attributes(c["intra_work_relation"]).downcase,
+              "relationType" => "HasTranslation",
+              "relatedIdentifierType" => "DOI" }.compact
+          else
+            nil
+          end
+        end.compact.unwrap
+      end
+
+      def crossref_is_translation_of(program_metadata)
+        refs = program_metadata.dig("related_item") if program_metadata.is_a?(Hash)
+        Array.wrap(refs).select { |a| a.dig("intra_work_relation", "relationship_type") == "isTranslationOf" }.map do |c|
+          if c.dig("intra_work_relation", "identifier_type") == "doi"
+            { "relatedIdentifier" => parse_attributes(c["intra_work_relation"]).downcase,
+              "relationType" => "IsTranslationOf",
+              "relatedIdentifierType" => "DOI" }.compact
+          else
+            nil
+          end
+        end.compact.unwrap
+      end
     end
   end
 end

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|
@@ -230,14 +246,14 @@ module Bolognese
             "creators" => get_authors(Array.wrap(ri.dig("creators", "creator"))),
             "titles" => get_titles(ri),
             "publicationYear" => ri["publicationYear"],
-            "volume" => ri["volume"],
-            "issue" => ri["issue"],
+            "volume" => parse_attributes(ri["volume"]),
+            "issue" => parse_attributes(ri["issue"]),
             "number" => number,
             "numberType" => numberType,
-            "firstPage" => ri["firstPage"],
-            "lastPage" => ri["lastPage"],
-            "publisher" => ri["publisher"],
-            "edition" => ri["edition"],
+            "firstPage" => parse_attributes(ri["firstPage"]),
+            "lastPage" => parse_attributes(ri["lastPage"]),
+            "publisher" => parse_attributes(ri["publisher"]),
+            "edition" => parse_attributes(ri["edition"]),
             "contributors" => get_authors(Array.wrap(ri.dig("contributors", "contributor"))),
           }.compact
         end
@@ -246,17 +262,13 @@ module Bolognese
           if !gl.is_a?(Hash) || gl["geoLocationPoint"].is_a?(String) || gl["geoLocationBox"].is_a?(String) || gl["geoLocationPolygon"].is_a?(String)
             nil
           else
-
-            # Handle scenario where multiple geoLocationPolygons are allowed within a single geoLocation
-            # we want to return an array if it's already an array (i.e. multiple geoLocationPolygons)
-            # vs if it's singular just return the object
-            # This is for backwards compatability to allow both scenarios.
+            ## Array of geoLocationPolygons within a geoLocation is Unsupported in JSON.
             if gl.dig("geoLocationPolygon").kind_of?(Array)
-              geoLocationPolygon = gl.dig("geoLocationPolygon").map do |glp|
-                Array.wrap(glp.dig("polygonPoint")).map { |glpp| { "polygonPoint" => glpp } }.compact.presence
-              end.compact.presence
+              geoLocationPolygon = nil
             else
-              geoLocationPolygon = Array.wrap(gl.dig("geoLocationPolygon", "polygonPoint")).map { |glp| { "polygonPoint" => glp } }.compact.presence
+              geoLocationPolygon = Array.wrap(gl.dig("geoLocationPolygon", "polygonPoint")).map { |glp| { "polygonPoint" => glp } }
+                .push(Array.wrap(gl.dig("geoLocationPolygon", "inPolygonPoint")).map { |glp| { "inPolygonPoint" => glp } }.first)
+                .compact.presence             
             end
 
             {
@@ -287,7 +299,7 @@ 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

@@ -10,7 +10,9 @@ module Bolognese
         "isPartOf" => "IsPartOf",
         "hasPart" => "HasPart",
         "isPredecessor" => "IsPreviousVersionOf",
-        "isSuccessor" => "IsNewVersionOf"
+        "isSuccessor" => "IsNewVersionOf",
+        "workTranslation" => "HasTranslation",
+        "translationOfWork" => "IsTranslationOf"
       }
 
       SO_TO_DC_REVERSE_RELATION_TYPES = {
@@ -74,7 +76,17 @@ 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)
+        translators = get_authors(from_schema_org_contributors(Array.wrap(meta.fetch("translator", nil))))
+        translators.map! do |translator|
+          translator["contributorType"] = "Translator"
+          translator
+        end
+        contributors += translators
+
+        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?
@@ -102,7 +114,9 @@ module Bolognese
           Array.wrap(schema_org_references(meta)) +
           Array.wrap(schema_org_is_referenced_by(meta)) +
           Array.wrap(schema_org_is_supplement_to(meta)) +
-          Array.wrap(schema_org_is_supplemented_by(meta))
+          Array.wrap(schema_org_is_supplemented_by(meta)) +
+          Array.wrap(schema_org_has_translation(meta)) +
+          Array.wrap(schema_org_is_translation_of(meta))
 
         rights_list = Array.wrap(meta.fetch("license", nil)).compact.map do |rl|
           hsh_to_spdx("__content__" => rl["name"], "rightsURI" => rl["id"])
@@ -123,6 +137,7 @@ module Bolognese
         dates << { "date" => meta.fetch("datePublished"), "dateType" => "Issued" } if Date.edtf(meta.fetch("datePublished", nil)).present?
         dates << { "date" => meta.fetch("dateCreated"), "dateType" => "Created" } if Date.edtf(meta.fetch("dateCreated", nil)).present?
         dates << { "date" => meta.fetch("dateModified"), "dateType" => "Updated" } if Date.edtf(meta.fetch("dateModified", nil)).present?
+        dates << { "date" => meta.fetch("temporalCoverage"), "dateType" => "Coverage" } if Date.edtf(meta.fetch("temporalCoverage", nil)).present?
         publication_year = meta.fetch("datePublished")[0..3] if meta.fetch("datePublished", nil).present?
 
         if meta.fetch("inLanguage", nil).is_a?(String)
@@ -180,7 +195,7 @@ 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,
@@ -236,6 +251,14 @@ module Bolognese
         schema_org_related_identifier(meta, relation_type: "isBasedOn")
       end
 
+      def schema_org_has_translation(meta)
+        schema_org_related_identifier(meta, relation_type: "workTranslation", )
+      end
+
+      def schema_org_is_translation_of(meta)
+        schema_org_related_identifier(meta, relation_type: "translationOfWork")
+      end
+
     end
   end
 end