lib/bolognese/readers/schema_org_reader.rb in bolognese-1.0.29 vs lib/bolognese/readers/schema_org_reader.rb in bolognese-1.0.30

@@ -35,25 +35,23 @@
 
         read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:doi, :id, :url, :sandbox, :validate))
 
         meta = string.present? ? Maremma.from_json(string) : {}
 
-        identifier = Array.wrap(meta.fetch("identifier", nil))
-        if identifier.length > 1
-          alternate_identifiers = identifier[1..-1].map do |r|
-            if r.is_a?(String)
-              { "alternateIdentifierType" => "URL", "alternateIdentifier" => r }
-            elsif r.is_a?(Hash)
-              { "alternateIdentifierType" => r["propertyID"], "alternateIdentifier" => r["value"] }
-            end
+        identifiers = ([options[:doi] || meta.fetch("@id", nil)] + Array.wrap(meta.fetch("identifier", nil))).map do |r|
+          r = normalize_id(r) if r.is_a?(String)
+          if r.is_a?(String) && r.start_with?("https://doi.org")
+            { "identifierType" => "DOI", "identifier" => r }
+          elsif r.is_a?(String)
+              { "identifierType" => "URL", "identifier" => r }
+          elsif r.is_a?(Hash)
+            { "identifierType" => get_identifier_type(r["propertyID"]), "identifier" => r["value"] }
           end
-        else
-          alternate_identifiers = nil
-        end
-        identifier = identifier.first
+        end.compact.uniq
 
-        id = normalize_id(meta.fetch("@id", nil) || meta.fetch("identifier", nil))
+        id = Array.wrap(identifiers).first.to_h.fetch("identifier", nil)
+
         schema_org = meta.fetch("@type", nil) && meta.fetch("@type").camelcase
         resource_type_general = Bolognese::Utils::SO_TO_DC_TRANSLATIONS[schema_org]
         types = {
           "resourceTypeGeneral" => resource_type_general,
           "resourceType" => meta.fetch("additionalType", nil),
@@ -61,23 +59,30 @@
           "citeproc" => Bolognese::Utils::SO_TO_CP_TRANSLATIONS[schema_org] || "article-journal",
           "bibtex" => Bolognese::Utils::SO_TO_BIB_TRANSLATIONS[schema_org] || "misc",
           "ris" => Bolognese::Utils::SO_TO_RIS_TRANSLATIONS[resource_type_general.to_s.dasherize] || "GEN"
         }.compact
         authors = meta.fetch("author", nil) || meta.fetch("creator", nil)
-        creators = get_authors(from_schema_org(Array.wrap(authors)))
-        contributors = get_authors(from_schema_org(Array.wrap(meta.fetch("editor", nil))))
+        creators = get_authors(from_schema_org_creators(Array.wrap(authors)))
+        contributors = get_authors(from_schema_org_contributors(Array.wrap(meta.fetch("editor", nil))))
         publisher = parse_attributes(meta.fetch("publisher", nil), content: "name", first: true)
 
         ct = (schema_org == "Dataset") ? "includedInDataCatalog" : "Periodical"
-        periodical = if meta.fetch(ct, nil).present?
+        container = if meta.fetch(ct, nil).present?
+          url =  parse_attributes(from_schema_org(meta.fetch(ct, nil)), content: "url", first: true)
+          
           {
-            "type" => (schema_org == "Dataset") ? "DataCatalog" : "Periodical",
+            "type" => (schema_org == "Dataset") ? "DataRepository" : "Periodical",
             "title" => parse_attributes(from_schema_org(meta.fetch(ct, nil)), content: "name", first: true),
-            "url" => parse_attributes(from_schema_org(meta.fetch(ct, nil)), content: "url", first: true)
+            "identifier" => url,
+            "identifierType" => url.present? ? "URL" : nil,
+            "volume" => meta.fetch("volumeNumber", nil),
+            "issue" => meta.fetch("issueNumber", nil),
+            "firstPage" => meta.fetch("pageStart", nil),
+            "lastPage" => meta.fetch("pageEnd", nil)
           }.compact
         else
-          nil
+          {}
         end
 
         related_identifiers = Array.wrap(schema_org_is_identical_to(meta)) +
           Array.wrap(schema_org_is_part_of(meta)) +
           Array.wrap(schema_org_has_part(meta)) +
@@ -136,21 +141,20 @@
         end
 
         { "id" => id,
           "types" => types,
           "doi" => validate_doi(id),
-          "identifier" => identifier,
-          "alternate_identifiers" => alternate_identifiers,
+          "identifiers" => identifiers,
           "url" => normalize_id(meta.fetch("url", nil)),
           "content_url" => Array.wrap(meta.fetch("contentUrl", nil)),
           "sizes" => Array.wrap(meta.fetch("contenSize", nil)).presence,
           "formats" => Array.wrap(meta.fetch("encodingFormat", nil) || meta.fetch("fileFormat", nil)),
           "titles" => meta.fetch("name", nil).present? ? [{ "title" => meta.fetch("name", nil) }] : nil,
           "creators" => creators,
           "contributors" => contributors,
           "publisher" => publisher,
           "agency" => parse_attributes(meta.fetch("provider", nil), content: "name", first: true),
-          "periodical" => periodical,
+          "container" => container,
           "related_identifiers" => related_identifiers,
           "publication_year" => publication_year,
           "dates" => dates,
           "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description")), "descriptionType" => "Abstract" }] : nil,
           "rights_list" => rights_list,