bolognese 1.0.2 → 1.0.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: acbbf45161f598000d9ff25bb9377c5332eb1482830a2e2b4137aab2fcc33043
-  data.tar.gz: 3a6c992a04626c916ec2f579aad4f9ab303756071db6a8178499a996834416cd
+  metadata.gz: 101bd48af59219c102771f24b66e6b19d1be7a533cbdf4a1f1ce49919769be35
+  data.tar.gz: 2e1ec013d5ab433476452390f535c0bb675e611b4c8e0a94d13d6cb9984150f7
 SHA512:
-  metadata.gz: 65dba0019f3401ae22dbd8d05da1cda01900ab0f865a3b37c8b4d7983e7d8616779e0ad7b7a123815eef252db8c5734982321bcf5cc2cc599e75a70dec436417
-  data.tar.gz: d856dea89a299b900acd9462cf01b2cb1c88fe0d8619cd003b6f2fb4da4e54624e86bb580db50488c7d31a256c9fbede5802b185b126c253328c714bc3dd1191
+  metadata.gz: 92561676aae45f01727f68080a316343b8dc8eb4112562965504d571add5337024a882da16b6370e58c89b491c9bd82fb24a6e2bd161cdd67c5247073650b8a5
+  data.tar.gz: 490cadbec75e93b049b067ab6297aa5ffde7f9965c3d08e79a3752b39d1a913c268130c817a2d74d03a3c4ca5b5bae2470cb989fe823fbf2d095bd1716d92e1c

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (1.0.2)
+    bolognese (1.0.3)
       activesupport (>= 4.2.5, < 6)
       benchmark_methods (~> 0.7)
       bibtex-ruby (~> 4.1)

data/lib/bolognese/datacite_utils.rb

@@ -93,12 +93,12 @@ module Bolognese
     end
 
     def res_type
-      { "resource_type_general" => resource_type_general || Metadata::SO_TO_DC_TRANSLATIONS[type] || "Other",
-        "__content__" => additional_type || type }
+      { "resource_type_general" => types["resource_type_general"] || Metadata::SO_TO_DC_TRANSLATIONS[types["type"]] || "Other",
+        "__content__" => types["resource_type"] || types["type"] }
     end
 
     def insert_resource_type(xml)
-      return xml unless type.present?
+      return xml unless types["type"].present?
 
       xml.resourceType(res_type["__content__"],
         'resourceTypeGeneral' => res_type["resource_type_general"])
@@ -115,16 +115,15 @@ module Bolognese
     end
 
     def insert_dates(xml)
+      return xml unless Array.wrap(dates).present?
+
       xml.dates do
-        insert_date(xml, date_published, 'Issued') if date_published.present?
-        insert_date(xml, date_modified, 'Updated') if date_modified.present?
+        Array.wrap(dates).each do |date|
+          xml.date(date["date"], 'dateType' => date["date_type"] || "Issued")
+        end
       end
     end
 
-    def insert_date(xml, date, date_type)
-      xml.date(date, 'dateType' => date_type)
-    end
-
     def insert_funding_references(xml)
       return xml unless Array.wrap(funding_references).present?
 

data/lib/bolognese/metadata.rb

@@ -6,13 +6,12 @@ module Bolognese
   class Metadata
     include Bolognese::MetadataUtils
 
-    attr_writer :id, :provider_id, :client_id, :doi, :identifier, :creator, :title, :publisher, :rights, :dates, :date_published, :date_modified,
-                :date_updated, :journal, :volume, :first_page, :last_page, :url, :version, :resource_type_general,
+    attr_writer :id, :provider_id, :client_id, :doi, :identifier, :creator, :title, :publisher, :rights, :dates, :publication_year,
+                :volume, :first_page, :last_page, :url, :version,
                 :keywords, :editor, :description, :alternate_identifiers, :language, :size,
-                :formats, :schema_version, :resource_type_general, :meta, :periodical,
+                :formats, :schema_version, :meta, :periodical,
                 :format, :funding_references, :style, :locale, :state, :geo_location,
-                :type, :additional_type, :citeproc_type, :bibtex_type, :ris_type, :content_url, :related_identifiers
-
+                :types, :content_url, :related_identifiers
 
     def initialize(input: nil, from: nil, **options)
       id = normalize_id(input, options)
@@ -150,17 +149,13 @@ module Bolognese
     end
 
     def publication_year
-      date_published.present? ? date_published[0..3].to_i.presence : nil
+      @publication_year ||= meta.fetch("publication_year", nil)
     end
 
     def periodical
       @periodical ||= meta.fetch("periodical", nil)
     end
 
-    def descriptions
-      Array.wrap(description)
-    end
-
     def geo_location
       @geo_location ||= meta.fetch("geo_location", nil)
     end
@@ -189,28 +184,8 @@ module Bolognese
       @identifier ||= meta.fetch("id", nil)
     end
 
-    def type
-      @type ||= meta.fetch("type", nil)
-    end
-
-    def additional_type
-      @additional_type ||= meta.fetch("additional_type", nil)
-    end
-
-    def citeproc_type
-      @citeproc_type ||= meta.fetch("citeproc_type", nil)
-    end
-
-    def bibtex_type
-      @bibtex_type ||= meta.fetch("bibtex_type", nil)
-    end
-
-    def ris_type
-      @ris_type ||= meta.fetch("ris_type", nil)
-    end
-
-    def resource_type_general
-      @resource_type_general ||= meta.fetch("resource_type_general", nil)
+    def types
+      @types ||= meta.fetch("types", nil)
     end
 
     def title
@@ -220,17 +195,5 @@ module Bolognese
     def creator
       @creator ||= meta.fetch("creator", nil)
     end
-
-    def date_published
-      @date_published ||= meta.fetch("date_published", nil)
-    end
-
-    def date_modified
-      @date_modified ||= meta.fetch("date_modified", nil)
-    end
-
-    def date_updated
-      @date_updated ||= meta.fetch("date_updated", nil)
-    end
   end
 end

data/lib/bolognese/metadata_utils.rb

@@ -123,13 +123,13 @@ module Bolognese
 
     def citeproc_hsh
       {
-        "type" => citeproc_type,
+        "type" => types["citeproc"],
         "id" => identifier,
         "categories" => Array.wrap(keywords).map { |k| parse_attributes(k, content: "text", first: true) }.presence,
         "language" => language,
         "author" => to_citeproc(creator),
         "editor" => to_citeproc(editor),
-        "issued" => date_published ? get_date_parts(date_published) : nil,
+        "issued" => get_date(dates, "Issued") ? get_date_parts(get_date(dates, "Issued")) : nil,
         "submitted" => Array.wrap(dates).find { |d| d["type"] == "Submitted" }.to_h.fetch("__content__", nil),
         "abstract" => parse_attributes(description, content: "text", first: true),
         "container-title" => periodical && periodical["title"],

data/lib/bolognese/readers/bibtex_reader.rb

@@ -30,6 +30,14 @@ module Bolognese
         meta = string.present? ? BibTeX.parse(string).first : OpenStruct.new
 
         type = BIB_TO_SO_TRANSLATIONS[meta.try(:type).to_s] || "ScholarlyArticle"
+        types = {
+          "type" => type,
+          "resource_type_general" => Metadata::SO_TO_DC_TRANSLATIONS[type],
+          "resource_type" => Bolognese::Utils::BIB_TO_CR_TRANSLATIONS[meta.try(:type).to_s] || meta.try(:type).to_s,
+          "bibtex" => meta.type.to_s,
+          "citeproc" => BIB_TO_CP_TRANSLATIONS[meta.try(:type).to_s] || "misc",
+          "ris" => BIB_TO_RIS_TRANSLATIONS[meta.try(:type).to_s] || "GEN"
+        }.compact
         doi = meta.try(:doi).to_s.presence
 
         author = Array(meta.try(:author)).map do |a|
@@ -59,14 +67,16 @@ module Bolognese
 
         page_first, page_last = meta.try(:pages).to_s.split("-")
         state = doi.present? ? "findable" : "not_found"
+        dates = if meta.try(:date).present?
+          [{ "date" => meta.date.to_s,
+             "date_type" => "Issued" }]
+        else
+          nil
+        end
+        publication_year =  meta.try(:date).present? ? meta.date.to_s[0..3] : nil
 
         { "id" => normalize_doi(doi),
-          "type" => type,
-          "bibtex_type" => meta.type.to_s,
-          "citeproc_type" => BIB_TO_CP_TRANSLATIONS[meta.try(:type).to_s] || "misc",
-          "ris_type" => BIB_TO_RIS_TRANSLATIONS[meta.try(:type).to_s] || "GEN",
-          "resource_type_general" => Metadata::SO_TO_DC_TRANSLATIONS[type],
-          "additional_type" => Bolognese::Utils::BIB_TO_CR_TRANSLATIONS[meta.try(:type).to_s] || meta.try(:type).to_s,
+          "types" => types,
           "doi" => doi,
           "url" => meta.try(:url).to_s,
           "title" => meta.try(:title).to_s,
@@ -74,7 +84,8 @@ module Bolognese
           "periodical" => periodical,
           "publisher" => meta.try(:publisher).to_s.presence,
           "related_identifiers" => related_identifiers,
-          "date_published" => meta.try(:date).to_s.presence,
+          "dates" => dates,
+          "publication_year" => publication_year,
           "volume" => meta.try(:volume).to_s.presence,
           "page_first" => page_first,
           "page_last" => page_last,

data/lib/bolognese/readers/citeproc_reader.rb

@@ -35,10 +35,29 @@ module Bolognese
 
         citeproc_type = meta.fetch("type", nil)
         type = CP_TO_SO_TRANSLATIONS[citeproc_type] || "CreativeWork"
+        types = {
+          "type" => type,
+          "resource_type_general" => Bolognese::Utils::SO_TO_DC_TRANSLATIONS[type],
+          "reource_type" => meta.fetch("additionalType", nil),
+          "citeproc" => citeproc_type,
+          "bibtex" => Bolognese::Utils::SO_TO_BIB_TRANSLATIONS[type] || "misc",
+          "ris" => CP_TO_RIS_TRANSLATIONS[type] || "GEN"
+        }.compact
         doi = normalize_doi(meta.fetch("DOI", nil))
-        author = get_authors(from_citeproc(Array.wrap(meta.fetch("author", nil))))
+        creator = get_authors(from_citeproc(Array.wrap(meta.fetch("author", nil))))
         editor = get_authors(from_citeproc(Array.wrap(meta.fetch("editor", nil))))
-        date_published = get_date_from_date_parts(meta.fetch("issued", nil))
+        dates = if meta.fetch("issued", nil).present?
+          [{ "date" => get_date_from_date_parts(meta.fetch("issued", nil)),
+             "date_type" => "Issued" }]
+        else
+          nil
+        end
+        publication_year = get_date_from_date_parts(meta.fetch("issued", nil)).to_s[0..3]
+        rights = if meta.fetch("copyright", nil)
+          { "id" => normalize_url(meta.fetch("copyright")) }.compact
+        else
+          nil
+        end
         related_identifiers = if meta.fetch("container-title", nil).present? && meta.fetch("ISSN", nil).present?
           [{ "type" => "Periodical",
              "relation_type" => "IsPartOf",
@@ -59,24 +78,20 @@ module Bolognese
         state = id.present? ? "findable" : "not_found"
 
         { "id" => id,
-          "type" => type,
-          "additional_type" => meta.fetch("additionalType", nil),
-          "citeproc_type" => citeproc_type,
-          "bibtex_type" => Bolognese::Utils::SO_TO_BIB_TRANSLATIONS[type] || "misc",
-          "ris_type" => CP_TO_RIS_TRANSLATIONS[type] || "GEN",
-          "resource_type_general" => Bolognese::Utils::SO_TO_DC_TRANSLATIONS[type],
+          "types" => types,
           "doi" => doi_from_url(doi),
           "url" => normalize_id(meta.fetch("URL", nil)),
           "title" => meta.fetch("title", nil),
-          "creator" => author,
+          "creator" => creator,
           "periodical" => periodical,
           "publisher" => meta.fetch("publisher", nil),
           "related_identifiers" => related_identifiers,
-          "date_published" => date_published,
+          "dates" => dates,
+          "publication_year" => publication_year,
           "volume" => meta.fetch("volume", nil),
           #{}"pagination" => meta.pages.to_s.presence,
           "description" => meta.fetch("abstract", nil).present? ? { "text" => sanitize(meta.fetch("abstract")) } : nil,
-          #{ }"rights" => { "id" => meta.field?(:copyright) && meta.copyright.to_s.presence },
+          "rights" => rights,
           "version" => meta.fetch("version", nil),
           "keywords" => meta.fetch("categories", nil),
           "state" => state

data/lib/bolognese/readers/codemeta_reader.rb

@@ -21,25 +21,27 @@ module Bolognese
         meta = string.present? ? Maremma.from_json(string) : {}
         identifier = meta.fetch("identifier", nil)
         id = normalize_id(meta.fetch("@id", nil) || identifier)
-        type = meta.fetch("@type", nil)
         author = get_authors(from_schema_org(Array.wrap(meta.fetch("agents", nil))))
         editor = get_authors(from_schema_org(Array.wrap(meta.fetch("editor", nil))))
-        date_published = meta.fetch("datePublished", nil)
+        dates = []
+        dates << { "date" => meta.fetch("datePublished"), "date_type" => "Issued" } if meta.fetch("datePublished", nil).present?
+        dates << { "date" => meta.fetch("dateCreated"), "date_type" => "Created" } if meta.fetch("dateCreated", nil).present?
+        dates << { "date" => meta.fetch("dateModified"), "date_type" => "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)
         state = meta.present? ? "findable" : "not_found"
-        dates = [
-          { "type" => "Issued", "__content__" => "date_published" },
-          { "type" => "Updated", "__content__" => "date_modified" },
-          { "type" => "Created", "__content__" => meta.fetch("datePublished", nil) }
-        ]
-
-        { "id" => id,
+        type = meta.fetch("@type", nil)
+        types = {
           "type" => type,
-          "additional_type" => meta.fetch("additionalType", nil),
-          "citeproc_type" => Bolognese::Utils::SO_TO_CP_TRANSLATIONS[type] || "article-journal",
-          "bibtex_type" => Bolognese::Utils::SO_TO_BIB_TRANSLATIONS[type] || "misc",
-          "ris_type" => Bolognese::Utils::SO_TO_RIS_TRANSLATIONS[type] || "GEN",
           "resource_type_general" => Bolognese::Utils::SO_TO_DC_TRANSLATIONS[type],
+          "resource_type" => meta.fetch("additionalType", nil),
+          "citeproc" => Bolognese::Utils::SO_TO_CP_TRANSLATIONS[type] || "article-journal",
+          "bibtex" => Bolognese::Utils::SO_TO_BIB_TRANSLATIONS[type] || "misc",
+          "ris" => Bolognese::Utils::SO_TO_RIS_TRANSLATIONS[type] || "GEN"
+        }.compact
+
+        { "id" => id,
+          "types" => types,
           "identifier" => identifier,
           "doi" => validate_doi(id),
           "url" => normalize_id(meta.fetch("codeRepository", nil)),
@@ -48,9 +50,8 @@ module Bolognese
           "editor" => editor,
           "publisher" => publisher,
           #{}"is_part_of" => is_part_of,
-          "dates" => meta.fetch("dateCreated", nil),
-          "date_published" => date_published,
-          "date_modified" => meta.fetch("dateModified", nil),
+          "dates" => dates,
+          "publication_year" => publication_year,
           "description" => meta.fetch("description", nil).present? ? { "text" => sanitize(meta.fetch("description")) } : nil,
           "rights" => { "id" => meta.fetch("license", nil) },
           "version" => meta.fetch("version", nil),

data/lib/bolognese/readers/crossref_reader.rb

@@ -31,7 +31,7 @@ module Bolognese
         # report-paper, sa_component, standard
         model = meta.dig("crossref").to_h.keys.first
 
-        additional_type = nil
+        resource_type = nil
         bibliographic_metadata = {}
         program_metadata = {}
         journal_metadata = nil
@@ -44,7 +44,7 @@ module Bolognese
           book_series_metadata = meta.dig("crossref", "book", "book_series_metadata")
           book_set_metadata = meta.dig("crossref", "book", "book_set_metadata")
           bibliographic_metadata = meta.dig("crossref", "book", "content_item").to_h
-          additional_type = bibliographic_metadata.fetch("component_type", nil) ? "book-" + bibliographic_metadata.fetch("component_type") : "book"
+          resource_type = bibliographic_metadata.fetch("component_type", nil) ? "book-" + bibliographic_metadata.fetch("component_type") : "book"
           publisher = book_metadata.dig("publisher", "publisher_name")
         when "conference"
           event_metadata = meta.dig("crossref", "conference", "event_metadata") || {}
@@ -56,7 +56,7 @@ module Bolognese
           journal_issue = meta.dig("crossref", "journal", "journal_issue") || {}
           journal_article = meta.dig("crossref", "journal", "journal_article") || {}
 
-          additional_type = if journal_article.present?
+          resource_type = if journal_article.present?
                               "journal_article"
                             elsif journal_issue.present?
                               "journal_issue"
@@ -70,14 +70,27 @@ module Bolognese
           bibliographic_metadata = meta.dig("crossref", "sa_component", "component_list", "component").to_h
         end
 
-        additional_type = (additional_type || model).to_s.underscore.camelize.presence
-        type = Bolognese::Utils::CR_TO_SO_TRANSLATIONS[additional_type] || "ScholarlyArticle"
-
+        resource_type = (resource_type || model).to_s.underscore.camelize.presence
+        type = Bolognese::Utils::CR_TO_SO_TRANSLATIONS[resource_type] || "ScholarlyArticle"
+        types = {
+          "type" => type,
+          "resource_type_general" => Bolognese::Utils::SO_TO_DC_TRANSLATIONS[type],
+          "resource_type" => resource_type,
+          "citeproc" => Bolognese::Utils::CR_TO_CP_TRANSLATIONS[resource_type] || "article-journal",
+          "bibtex" => Bolognese::Utils::CR_TO_BIB_TRANSLATIONS[resource_type] || "misc",
+          "ris" => Bolognese::Utils::CR_TO_RIS_TRANSLATIONS[resource_type] || "JOUR"
+        }.compact
         doi = bibliographic_metadata.dig("doi_data", "doi").to_s.downcase.presence #|| doi_from_url(options[:id])
 
         # Crossref servers run on Eastern Time
         Time.zone = 'Eastern Time (US & Canada)'
-        date_modified = Time.zone.parse(meta.fetch("timestamp", "2018-01-01")).utc.iso8601
+        dates = [
+          { "date" => crossref_date_published(bibliographic_metadata),
+            "date_type" => "Issued" },
+          { "date" => Time.zone.parse(meta.fetch("timestamp", "2018-01-01")).utc.iso8601,
+            "date_type" => "Updated" }
+        ]
+        publication_year = crossref_date_published(bibliographic_metadata).present? ? crossref_date_published(bibliographic_metadata)[0..3] : nil
         state = meta.present? ? "findable" : "not_found"
 
         related_identifiers = Array.wrap(crossref_is_part_of(journal_metadata)) + Array.wrap(crossref_references(bibliographic_metadata))
@@ -90,12 +103,7 @@ module Bolognese
         end
 
         { "id" => normalize_doi(doi),
-          "type" => type,
-          "additional_type" => additional_type,
-          "citeproc_type" => Bolognese::Utils::CR_TO_CP_TRANSLATIONS[additional_type] || "article-journal",
-          "bibtex_type" => Bolognese::Utils::CR_TO_BIB_TRANSLATIONS[additional_type] || "misc",
-          "ris_type" => Bolognese::Utils::CR_TO_RIS_TRANSLATIONS[additional_type] || "JOUR",
-          "resource_type_general" => Bolognese::Utils::SO_TO_DC_TRANSLATIONS[type],
+          "types" => types,
           "doi" => doi,
           "url" => bibliographic_metadata.dig("doi_data", "resource"),
           "title" => parse_attributes(bibliographic_metadata.dig("titles", "title")),
@@ -107,8 +115,8 @@ module Bolognese
           "periodical" => periodical,
           "service_provider" => "Crossref",
           "related_identifiers" => related_identifiers,
-          "date_published" => crossref_date_published(bibliographic_metadata),
-          "date_modified" => date_modified,
+          "dates" => dates,
+          "publication_year" => publication_year,
           "volume" => journal_issue.dig("journal_volume", "volume"),
           "issue" => journal_issue.dig("issue"),
           "first_page" => bibliographic_metadata.dig("pages", "first_page"),

data/lib/bolognese/readers/datacite_json_reader.rb

@@ -9,8 +9,6 @@ module Bolognese
 
         meta = string.present? ? Maremma.from_json(string) : {}
 
-        resource_type_general = meta.fetch("resource-type-general", nil)
-        type = Bolognese::Utils::DC_TO_SO_TRANSLATIONS[resource_type_general.to_s.dasherize] || "CreativeWork"
         state = meta.fetch("doi", nil).present? ? "findable" : "not_found"
         related_identifiers = Array.wrap(meta.fetch("related-identifiers", nil)).map do |ri|
           { "id" => ri["id"],
@@ -18,14 +16,25 @@ module Bolognese
             "related_identifier_type" => ri["related-identifier-type"],
             "resource_type_general" => ri["resource-type-general"] }.compact
         end
-
-        { "id" => meta.fetch("id", nil),
+        dates = Array.wrap(meta.fetch("dates", nil)).map do |d|
+          { "date" => d["date"],
+            "date_type" => d["date-type"] }.compact
+        end
+        dates << { "date" => meta.fetch("publication-year", nil), "date_type" => "Issued" } if meta.fetch("publication-year", nil).present? && get_date(dates, "Issued").blank?
+        resource_type_general = meta.fetch("resource-type-general", nil)
+        resource_type = meta.fetch("resource-type", nil)
+        type = Bolognese::Utils::CR_TO_SO_TRANSLATIONS[resource_type.to_s.underscore.camelcase] || Bolognese::Utils::DC_TO_SO_TRANSLATIONS[resource_type_general.to_s.dasherize] || "CreativeWork"
+        types = {
           "type" => type,
-          "additional_type" => meta.fetch("resource-type", nil),
-          "citeproc_type" => Bolognese::Utils::DC_TO_CP_TRANSLATIONS[resource_type_general.to_s.dasherize] || "other",
-          "bibtex_type" => Bolognese::Utils::SO_TO_BIB_TRANSLATIONS[type] || "misc",
-          "ris_type" => Bolognese::Utils::SO_TO_RIS_TRANSLATIONS[resource_type_general.to_s.dasherize] || "GEN",
           "resource_type_general" => resource_type_general,
+          "resource_type" => resource_type,
+          "citeproc" => Bolognese::Utils::CR_TO_CP_TRANSLATIONS[resource_type.to_s.underscore.camelcase] || Bolognese::Utils::SO_TO_CP_TRANSLATIONS[type] || "article",
+          "bibtex" => Bolognese::Utils::CR_TO_BIB_TRANSLATIONS[resource_type.to_s.underscore.camelcase] || Bolognese::Utils::SO_TO_BIB_TRANSLATIONS[type] || "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
+
+        { "id" => meta.fetch("id", nil),
+          "types" => types,
           "doi" => validate_doi(meta.fetch("doi", nil)),
           "url" => normalize_id(meta.fetch("url", nil)),
           "title" => meta.fetch("title", nil),
@@ -37,9 +46,8 @@ module Bolognese
           "service_provider" => "DataCite",
           "funding_references" => meta.fetch("funding-references", nil),
           "related_identifiers" => related_identifiers,
-          "dates" => meta.fetch("dates", nil),
-          "date_published" => meta.fetch("date-published", nil) || meta.fetch("publication-year", nil),
-          "date_modified" => meta.fetch("date-modified", nil),
+          "dates" => dates,
+          "publication_year" => meta.fetch("publication-year", nil),
           "description" => meta.fetch("description", nil),
           "rights" => meta.fetch("rights", nil),
           "version" => meta.fetch("version", nil),