bolognese 0.9.36 → 0.9.37

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 8418e6a4e1153609a947be86602553372b3b2954
-  data.tar.gz: 2cde62ec0b2f271b30f0697f31543b7c78664b3a
+  metadata.gz: a4c14e423a53deb412218f3c855ee1f593296d98
+  data.tar.gz: d05e58c1dbcb061bdc4d468f048a02cd238036a5
 SHA512:
-  metadata.gz: 69dcf1889488893a2f2aaeb5f472bb8afd60cfc4caf3882c7972942ef2b23e96b103c8d6d6a616dc04271113aae374275171f001b1a620942f70f4707e319733
-  data.tar.gz: c505571a6a18693f89711e533a35fc6afdaceff702cd8aacb0c3d889c01c59cb0982be2b075c3a6850bbd463a4089abfa0bdd9f3bb0bde6f9be851907b8b18be
+  metadata.gz: 24551c9b762c597fc87407e926a510c873d42e6650f83e76e9081d8b43d92e4e06067aa18a8f30aca41bacc39a98fa7fda1b6d4ea1dc5c785aa7c971785c95fe
+  data.tar.gz: a66e615373cf75d9fee848a017db2966722fb2754b5985bf58f8cf45b43d3be4c243f25d4ac7377ee26d722fc0b682f02fb1ae308c584f6db5a3b74badfbc3ed

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (0.9.36)
+    bolognese (0.9.37)
       activesupport (>= 4.2.5, < 6)
       benchmark_methods (~> 0.7)
       bibtex-ruby (~> 4.1)
@@ -149,7 +149,7 @@ GEM
     temple (0.8.0)
     thor (0.19.4)
     thread_safe (0.3.6)
-    tilt (2.0.7)
+    tilt (2.0.8)
     trollop (2.1.2)
     tzinfo (1.2.3)
       thread_safe (~> 0.1)

data/lib/bolognese/cli.rb

@@ -27,9 +27,16 @@ module Bolognese
     method_option :to, aliases: "-t", default: "schema_org"
     method_option :regenerate, :type => :boolean, :force => false
     def convert(input)
-      metadata = Metadata.new(input: input, from: options[:from], regenerate: options[:regenerate])
+      metadata = Metadata.new(input: input,
+                              from: options[:from],
+                              regenerate: options[:regenerate])
       to = options[:to] || "schema_org"
-      puts metadata.send(to)
+
+      if metadata.valid?
+        puts metadata.send(to)
+      else
+        $stderr.puts metadata.errors
+      end
     end
 
     default_task :convert

data/lib/bolognese/datacite_utils.rb

@@ -1,11 +1,5 @@
 module Bolognese
   module DataciteUtils
-    def schema
-      kernel = schema_version.split("/").last || "kernel-4.0"
-      filepath = File.expand_path("../../../resources/#{kernel}/metadata.xsd", __FILE__)
-      Nokogiri::XML::Schema(open(filepath))
-    end
-
     def datacite_xml
       @datacite_xml ||= Nokogiri::XML::Builder.new(:encoding => 'UTF-8') do |xml|
         xml.resource(root_attributes) do
@@ -14,12 +8,13 @@ module Bolognese
       end.to_xml
     end
 
-    # def datacite
-    #   datacite_xml
-    # end
+    def datacite_errors(xml: nil, schema_version: nil)
+      schema_version ||= "http://datacite.org/schema/kernel-4"
+      kernel = schema_version.to_s.split("/").last
+      filepath = File.expand_path("../../../resources/#{kernel}/metadata.xsd", __FILE__)
+      schema = Nokogiri::XML::Schema(open(filepath))
 
-    def datacite_errors
-      schema.validate(Nokogiri::XML(datacite, nil, 'UTF-8')).map { |error| error.to_s }.unwrap
+      schema.validate(Nokogiri::XML(xml, nil, 'UTF-8')).map { |error| error.to_s }.unwrap
     rescue Nokogiri::XML::SyntaxError => e
       e.message
     end
@@ -88,7 +83,7 @@ module Bolognese
     end
 
     def insert_publisher(xml)
-      xml.publisher(publisher)
+      xml.publisher(publisher || container_title)
     end
 
     def insert_publication_year(xml)

data/lib/bolognese/doi_utils.rb

@@ -1,7 +1,8 @@
 module Bolognese
   module DoiUtils
     def validate_doi(doi)
-      Array(/\A(?:(http|https):\/(\/)?(dx\.)?(doi.org|doi.test.datacite.org)\/)?(doi:)?(10\.\d{4,5}\/.+)\z/.match(doi)).last
+      doi = Array(/\A(?:(http|https):\/(\/)?(dx\.)?(doi.org|doi.test.datacite.org)\/)?(doi:)?(10\.\d{4,5}\/.+)\z/.match(doi)).last
+      doi = doi.delete("\u200B").downcase if doi.present?
     end
 
     def validate_prefix(doi)

data/lib/bolognese/metadata.rb

@@ -57,20 +57,24 @@ module Bolognese
     include Bolognese::Writers::SchemaOrgWriter
     include Bolognese::Writers::TurtleWriter
 
-    attr_reader :id, :from, :raw, :metadata, :doc, :provider, :schema_version, :license, :citation,
-      :additional_type, :alternate_name, :url, :version, :keywords, :editor,
-      :page_start, :page_end, :date_modified, :language, :spatial_coverage,
-      :content_size, :funding, :journal, :bibtex_type, :date_created, :has_part,
-      :publisher, :contributor, :same_as, :is_previous_version_of, :is_new_version_of,
-      :should_passthru, :errors, :datacite_errors, :date_accepted, :date_available,
-      :date_copyrighted, :date_collected, :date_submitted, :date_valid,
-      :is_cited_by, :cites, :is_supplement_to, :is_supplemented_by,
-      :is_continued_by, :continues, :has_metadata, :is_metadata_for,
-      :is_referenced_by, :references, :is_documented_by, :documents,
-      :is_compiled_by, :compiles, :is_variant_form_of, :is_original_form_of,
-      :is_reviewed_by, :reviews, :is_derived_from, :is_source_of, :format,
-      :related_identifier, :reverse, :citeproc_type, :ris_type, :volume, :issue,
-      :member_id, :data_center_id, :date_registered, :date_updated, :name_detector
+    attr_accessor :doi, :author, :title, :publisher, :contributor, :license,
+      :date_accepted, :date_available, :date_copyrighted, :date_collected,
+      :date_submitted, :date_valid, :date_created, :date_modified,
+      :date_registered, :date_updated, :member_id, :data_center_id, :journal,
+      :volume, :issue, :pagination, :url, :version, :keywords, :editor,
+      :description, :alternate_name, :language, :content_size, :spatial_coverage,
+      :schema_version, :additional_type, :has_part, :same_as,
+      :is_previous_version_of, :is_new_version_of,   :is_cited_by, :cites,
+      :is_supplement_to, :is_supplemented_by, :is_continued_by, :continues,
+      :has_metadata, :is_metadata_for, :is_referenced_by, :references,
+      :is_documented_by, :documents, :is_compiled_by, :compiles,
+      :is_variant_form_of, :is_original_form_of, :is_reviewed_by, :reviews,
+      :is_derived_from, :is_source_of, :format, :funding, :type, :bibtex_type,
+      :citeproc_type, :ris_type
+
+    attr_reader :id, :from, :raw, :metadata, :doc, :provider, :citation,
+      :page_start, :page_end, :should_passthru, :errors,
+      :related_identifier, :reverse, :name_detector
 
     def initialize(input: nil, from: nil, regenerate: false, **options)
       id = normalize_id(input, options)
@@ -100,7 +104,6 @@ module Bolognese
       @raw = string.present? ? string.strip : nil
 
       @should_passthru = (@from == "datacite") && !regenerate
-      @doi = options[:doi].presence
 
       @url = hsh.to_h["url"].presence
       @date_registered = hsh.to_h["date_registered"].presence
@@ -117,40 +120,39 @@ module Bolognese
       exists? && errors.nil?
     end
 
+    # validate against DataCite schema, unless there are already errors in the reader
     def errors
-      metadata.fetch("errors", nil)
+      xml = should_passthru ? raw : datacite_xml
+      metadata.fetch("errors", nil) || datacite_errors(xml: xml,
+                                                       schema_version: schema_version)
     end
 
-    # def errors
-    #   doc && doc.errors.map { |error| error.to_s }.unwrap
-    # end
-
     def id
       @doi.present? ? doi_as_url(@doi) : metadata.fetch("id", nil)
     end
 
     def type
-      metadata.fetch("type", nil)
+      @type ||= metadata.fetch("type", nil)
     end
 
     def additional_type
-      metadata.fetch("additional_type", nil)
+      @additional_type ||= metadata.fetch("additional_type", nil)
     end
 
     def citeproc_type
-      metadata.fetch("citeproc_type", nil)
+      @citeproc_type ||= metadata.fetch("citeproc_type", nil)
     end
 
     def bibtex_type
-      metadata.fetch("bibtex_type", nil)
+      @bibtex_type ||= metadata.fetch("bibtex_type", nil)
     end
 
     def ris_type
-      metadata.fetch("ris_type", nil)
+      @ris_type ||= metadata.fetch("ris_type", nil)
     end
 
     def resource_type_general
-      metadata.fetch("resource_type_general", nil)
+      @resource_type_general ||= metadata.fetch("resource_type_general", nil)
     end
 
     def doi
@@ -162,63 +164,63 @@ module Bolognese
     end
 
     def title
-      metadata.fetch("title", nil)
+      @title ||= metadata.fetch("title", nil)
     end
 
     def alternate_name
-      metadata.fetch("alternate_name", nil)
+      @alternate_name ||= metadata.fetch("alternate_name", nil)
     end
 
     def author
-      metadata.fetch("author", nil)
+      @author ||= metadata.fetch("author", nil)
     end
 
     def editor
-      metadata.fetch("editor", nil)
+      @editor ||= metadata.fetch("editor", nil)
     end
 
     def publisher
-      metadata.fetch("publisher", nil)
+      @publisher ||= metadata.fetch("publisher", nil)
     end
 
     def provider
-      metadata.fetch("provider", nil)
+      @provider ||= metadata.fetch("provider", nil)
     end
 
     def date_created
-      metadata.fetch("date_created", nil)
+      @date_created ||= metadata.fetch("date_created", nil)
     end
 
     def date_accepted
-      metadata.fetch("date_accepted", nil)
+      @date_accepted ||= metadata.fetch("date_accepted", nil)
     end
 
     def date_available
-      metadata.fetch("date_available", nil)
+      @date_available ||= metadata.fetch("date_available", nil)
     end
 
     def date_copyrighted
-      metadata.fetch("date_copyrighted", nil)
+      @date_copyrighted ||= metadata.fetch("date_copyrighted", nil)
     end
 
     def date_collected
-      metadata.fetch("date_collected", nil)
+      @date_collected ||= metadata.fetch("date_collected", nil)
     end
 
     def date_submitted
-      metadata.fetch("date_submitted", nil)
+      @date_submitted ||= metadata.fetch("date_submitted", nil)
     end
 
     def date_valid
-      metadata.fetch("date_valid", nil)
+      @date_valid ||= metadata.fetch("date_valid", nil)
     end
 
     def date_published
-      metadata.fetch("date_published", nil)
+      @date_published ||= metadata.fetch("date_published", nil)
     end
 
     def date_modified
-      metadata.fetch("date_modified", nil)
+      @date_modified ||= metadata.fetch("date_modified", nil)
     end
 
     def date_registered
@@ -230,43 +232,43 @@ module Bolognese
     end
 
     def volume
-      metadata.fetch("volume", nil)
+      @volume ||= metadata.fetch("volume", nil)
     end
 
     def pagination
-      metadata.fetch("pagination", nil)
+      @pagination ||= metadata.fetch("pagination", nil)
     end
 
     def description
-      metadata.fetch("description", nil)
+      @description ||= metadata.fetch("description", nil)
     end
 
     def license
-      metadata.fetch("license", nil)
+      @license ||= metadata.fetch("license", nil)
     end
 
     def version
-      metadata.fetch("version", nil)
+      @version ||= metadata.fetch("version", nil)
     end
 
     def keywords
-      metadata.fetch("keywords", nil)
+      @keywords ||= metadata.fetch("keywords", nil)
     end
 
     def language
-      metadata.fetch("language", nil)
+      @language ||= metadata.fetch("language", nil)
     end
 
     def content_size
-      metadata.fetch("content_size", nil)
+      @content_size ||= metadata.fetch("content_size", nil)
     end
 
     def schema_version
-      metadata.fetch("schema_version", nil)
+      @schema_version ||= metadata.fetch("schema_version", nil)
     end
 
     def funding
-      metadata.fetch("funding", nil)
+      @funding ||= metadata.fetch("funding", nil)
     end
 
     def member_id
@@ -330,7 +332,8 @@ module Bolognese
     end
 
     def related_identifier_hsh(relation_type)
-      Array.wrap(send(relation_type)).map { |r| r.merge("relationType" => relation_type.camelize) }
+      Array.wrap(send(relation_type)).select { |r| r["id"] || r["issn"] }
+        .map { |r| r.merge("relationType" => relation_type.camelize) }
     end
 
     def related_identifier

data/lib/bolognese/readers/crossref_reader.rb

@@ -98,7 +98,7 @@ module Bolognese
                             meta.dig("crossref").keys.last.camelize
                           end
         type = CR_TO_SO_TRANSLATIONS[additional_type] || "ScholarlyArticle"
-        doi = bibliographic_metadata.dig("doi_data", "doi")
+        doi = bibliographic_metadata.dig("doi_data", "doi").to_s.downcase
 
         { "id" => normalize_doi(doi),
           "type" => type,

data/lib/bolognese/readers/datacite_reader.rb

@@ -29,7 +29,18 @@ module Bolognese
       end
 
       def read_datacite(string: nil, **options)
-        meta = string.present? ? Maremma.from_xml(string).fetch("resource", {}) : {}
+        return { "errors" => "no content" } unless string.present?
+
+        meta = Maremma.from_xml(string).fetch("resource", {})
+        schema_version = meta.fetch("xmlns", nil)
+
+        # validate only when option is set, as this step is expensive and
+        # not needed if XML comes from DataCite MDS
+        if options[:validate]
+          errors = datacite_errors(xml: string, schema_version: schema_version)
+          return { "errors" => errors } if errors.present?
+        end
+
         id = normalize_doi(meta.dig("identifier", "__content__"), sandbox: options[:sandbox])
         doi = doi_from_url(id)
         resource_type_general = meta.dig("resourceType", "resourceTypeGeneral")
@@ -104,7 +115,7 @@ module Bolognese
           "keywords" => keywords,
           "language" => meta.fetch("language", nil),
           "content_size" => meta.fetch("size", nil),
-          "schema_version" => meta.fetch("xmlns", nil)
+          "schema_version" => schema_version
         }
       end
 

data/lib/bolognese/readers/ris_reader.rb

@@ -36,7 +36,7 @@ module Bolognese
         ris_type = meta.fetch("TY", nil) || "GEN"
         type = RIS_TO_SO_TRANSLATIONS[ris_type] || "CreativeWork"
 
-        doi = meta.fetch("DO", nil)
+        doi = validate_doi(meta.fetch("DO", nil))
         author = Array.wrap(meta.fetch("AU", nil)).map { |a| { "name" => a } }
         container_title = meta.fetch("T2", nil)
         is_part_of = if container_title.present?

data/lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "0.9.36"
+  VERSION = "0.9.37"
 end

data/lib/bolognese/writers/citeproc_writer.rb

@@ -2,8 +2,6 @@ module Bolognese
   module Writers
     module CiteprocWriter
       def citeproc
-        return nil unless valid?
-
         hsh = {
           "type" => citeproc_type,
           "id" => id,

data/lib/bolognese/writers/codemeta_writer.rb

@@ -3,7 +3,7 @@ module Bolognese
     module CodemetaWriter
       def codemeta
         return nil unless valid?
-        
+
         hsh = {
           "@context" => id.present? ? "https://raw.githubusercontent.com/codemeta/codemeta/master/codemeta.jsonld" : nil,
           "@type" => type,

data/lib/bolognese/writers/crosscite_writer.rb

@@ -2,8 +2,6 @@ module Bolognese
   module Writers
     module CrossciteWriter
       def crosscite
-        return nil unless valid?
-
         hsh = {
           "id" => id,
           "doi" => doi,

data/lib/bolognese/writers/datacite_json_writer.rb

@@ -2,8 +2,6 @@ module Bolognese
   module Writers
     module DataciteJsonWriter
       def datacite_json
-        return nil unless valid?
-
         hsh = {
           "id" => id,
           "doi" => doi,

data/lib/bolognese/writers/rdf_xml_writer.rb

@@ -2,8 +2,6 @@ module Bolognese
   module Writers
     module RdfXmlWriter
       def rdf_xml
-        return nil unless valid?
-        
         graph.dump(:rdfxml, prefixes: { schema: "http://schema.org/" })
       end
     end