bolognese 1.11.4 → 2.7.0

data/lib/bolognese/readers/datacite_reader.rb

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require 'base64'
+
 module Bolognese
   module Readers
     module DataciteReader
@@ -94,13 +96,29 @@ 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|
@@ -115,9 +133,9 @@ module Bolognese
 
         subjects = Array.wrap(meta.dig("subjects", "subject")).reduce([]) do |sum, subject|
           if subject.is_a?(String)
-            sum += name_to_fos(subject)
+            sum += name_to_subject(subject)
           elsif subject.is_a?(Hash)
-            sum += hsh_to_fos(subject)
+            sum += hsh_to_subject(subject)
           end
 
           sum
@@ -125,15 +143,12 @@ module Bolognese
 
         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)
-              { "date" => date,
-                "dateType" => parse_attributes(r, content: "dateType"),
-                "dateInformation" => parse_attributes(r, content: "dateInformation")
-              }.compact
-            end
+            { "date" => date,
+              "dateType" => parse_attributes(r, content: "dateType"),
+              "dateInformation" => parse_attributes(r, content: "dateInformation")
+            }.compact
           end
         end.compact
-        dates << { "date" => meta.fetch("publicationYear", nil), "dateType" => "Issued" } if meta.fetch("publicationYear", nil).present? && get_date(dates, "Issued").blank?
         sizes = Array.wrap(meta.dig("sizes", "size")).map do |k|
           if k.blank?
             nil
@@ -161,8 +176,7 @@ module Bolognese
           if funder_identifier_type == "Crossref Funder ID"
             funder_identifier = validate_funder_doi(funder_identifier)
           elsif funder_identifier_type == "ROR"
-            funder_identifier =  normalize_ror(funder_identifier)
-            scheme_uri = "https://ror.org"
+            funder_identifier = normalize_ror(funder_identifier)
           else
             funder_identifier = normalize_id(funder_identifier) ? normalize_id(funder_identifier) : funder_identifier
           end
@@ -178,7 +192,8 @@ module Bolognese
         end
         related_identifiers = Array.wrap(meta.dig("relatedIdentifiers", "relatedIdentifier")).map do |ri|
           if ri["relatedIdentifierType"] == "DOI"
-            rid = validate_doi(ri["__content__"].to_s.downcase)
+            doi = ri["__content__"].to_s.downcase
+            rid = validate_doi(doi) || doi
           else
             rid = ri["__content__"]
           end
@@ -190,7 +205,8 @@ module Bolognese
             "resourceTypeGeneral" => ri["resourceTypeGeneral"],
             "relatedMetadataScheme" => ri["relatedMetadataScheme"],
             "schemeUri" => ri["schemeURI"],
-            "schemeType" => ri["schemeType"]
+            "schemeType" => ri["schemeType"],
+            "relationTypeInformation" => ri["relationTypeInformation"]
           }.compact
         end
 
@@ -200,7 +216,8 @@ module Bolognese
           relatedItemIdentifier = nil
           if rii
             if rii["relatedItemIdentifierType"] == "DOI"
-              rid = validate_doi(rii["__content__"].to_s.downcase)
+              doi = rii["__content__"].to_s.downcase
+              rid = validate_doi(doi) || doi
             else
               rid = rii["__content__"]
             end
@@ -210,14 +227,13 @@ module Bolognese
               "relatedItemIdentifierType" => rii["relatedItemIdentifierType"],
               "relatedMetadataScheme" => rii["relatedMetadataScheme"],
               "schemeURI" => rii["schemeURI"],
-              "schemeType" => rii["schemeType"]
+              "schemeType" => rii["schemeType"],
             }.compact
           end
 
           number = ri["number"]
           if number.is_a?(String)
             number = number
-            numberType = nil
           else
             number = ri.dig("number", "__content__")
             numberType = ri.dig("number", "numberType")
@@ -225,6 +241,7 @@ module Bolognese
 
           a = {
             "relationType" => ri["relationType"],
+            "relationTypeInformation" => ri["relationTypeInformation"],
             "relatedItemType" => ri["relatedItemType"],
             "relatedItemIdentifier" => relatedItemIdentifier,
             "creators" => get_authors(Array.wrap(ri.dig("creators", "creator"))),
@@ -246,17 +263,13 @@ 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
 
             {
@@ -286,8 +299,7 @@ module Bolognese
           "titles" => titles,
           "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,
@@ -307,28 +319,6 @@ module Bolognese
         }.merge(read_options)
       end
 
-      def set_container(meta)
-        series_information = Array.wrap(meta.dig("descriptions", "description")).find { |r| r["descriptionType"] == "SeriesInformation" }.to_h.fetch("__content__", nil)
-        si = get_series_information(series_information)
-
-        is_part_of = Array.wrap(meta.dig("relatedIdentifiers", "relatedIdentifier")).find { |ri| ri["relationType"] == "IsPartOf" }.to_h
-
-        if si["title"].present? || is_part_of.present?
-          {
-            "type" => meta.dig("resourceType", "resourceTypeGeneral") == "Dataset" ? "DataRepository" : "Series",
-            "identifier" => is_part_of["__content__"],
-            "identifierType" => is_part_of["relatedIdentifierType"],
-            "title" => si["title"],
-            "volume" => si["volume"],
-            "issue" => si["issue"],
-            "firstPage" => si["firstPage"],
-            "lastPage" => si["lastPage"]
-          }.compact
-        else
-          {}
-        end
-      end
-
       def get_titles(meta)
         titles = Array.wrap(meta.dig("titles", "title")).map do |r|
           if r.blank?

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

@@ -78,7 +78,7 @@ module Bolognese
         end
         state = meta.fetch("DO", nil).present? || read_options.present? ? "findable" : "not_found"
         subjects = Array.wrap(meta.fetch("KW", nil)).reduce([]) do |sum, subject|
-          sum += name_to_fos(subject)
+          sum += name_to_subject(subject)
 
           sum
         end
@@ -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)
@@ -159,7 +174,7 @@ module Bolognese
         subjects = meta.fetch("keywords", nil)
         subjects = subjects.to_s.split(", ") if subjects.is_a?(String)
         subjects = Array.wrap(subjects).reduce([]) do |sum, subject|
-          sum += name_to_fos(subject)
+          sum += name_to_subject(subject)
           sum
         end
 
@@ -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