cirneco 0.7.4 → 0.8.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: a82c5b7e10ab13eabc1913ab86f856bf1d28ee68
-  data.tar.gz: 38c61337cd19f6623ad835182baccaf2607f260f
+  metadata.gz: 0b25496604d7050477dbcd531561118874f76705
+  data.tar.gz: f3e38823524d96cd5382a5a54d3838daaf8c3b42
 SHA512:
-  metadata.gz: 2dbc92ffc86523a271cb1c69e2af2778f56bf2b0ca4c61714f3825fcbc84ee52ce7af2b42cd9edd1ead906e9a2620feda76de8dfec622698b74b2ab8a7910cf1
-  data.tar.gz: 00c3050b139c651521e9e1d0c0a454002f0fff2ff5f5b1a018a0e8aefcc4a258a009c2c7a03c332c1bb762dbc3176a941d8bdce961b7680a056f5c2d6b8967e1
+  metadata.gz: 8b521bfb2aceebdb52d8248dccd6de030976413720d392748358620e7527c33398224a4a33aea7d7a4dca234c8506da408aceb00f5ac1a47e82177e034392c70
+  data.tar.gz: ada9b59941a9c2f3182736cb3ba54b2c44d3b16226a20abc51be56dc0a5b6f0e6949311c6b5ca9a20e396fe624fd872c2d152abe6fd99f823d0f164d27f46b1c

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    cirneco (0.7.4)
+    cirneco (0.8.1)
       activesupport (~> 4.2, >= 4.2.5)
       base32-crockford-checksum (~> 0.2.2)
       bergamasco (~> 0.3)
@@ -22,7 +22,7 @@ GEM
       tzinfo (~> 1.1)
     addressable (2.3.8)
     base32-crockford-checksum (0.2.3)
-    bergamasco (0.3.4)
+    bergamasco (0.3.5)
       activesupport (~> 4.2, >= 4.2.5)
       addressable (~> 2.3.8)
       builder (~> 3.2, >= 3.2.2)
@@ -33,7 +33,7 @@ GEM
       pandoc-ruby (~> 2.0, >= 2.0.0)
       safe_yaml (~> 1.0, >= 1.0.4)
     builder (3.2.2)
-    codeclimate-test-reporter (1.0.3)
+    codeclimate-test-reporter (1.0.4)
       simplecov
     crack (0.4.3)
       safe_yaml (~> 1.0.0)
@@ -47,7 +47,7 @@ GEM
       faraday
     faraday_middleware (0.10.1)
       faraday (>= 0.7.4, < 1.0)
-    hashdiff (0.3.1)
+    hashdiff (0.3.2)
     i18n (0.7.0)
     json (1.8.3)
     loofah (2.0.3)

data/lib/cirneco/api.rb

@@ -6,7 +6,7 @@ module Cirneco
     def post_metadata(data, options={})
       return OpenStruct.new(body: { "errors" => [{ "title" => "Username or password missing" }] }) unless options[:username].present? && options[:password].present?
 
-      mds_url = options[:sandbox] ? 'https://mds.test.datacite.org' : 'https://mds.datacite.org'
+      mds_url = options[:sandbox] ? 'https://mds-sandbox.datacite.org' : 'https://mds.datacite.org'
 
       url = "#{mds_url}/metadata"
       Maremma.post(url, content_type: 'application/xml;charset=UTF-8', data: data, username: options[:username], password: options[:password])
@@ -15,7 +15,7 @@ module Cirneco
     def get_metadata(doi, options={})
       return OpenStruct.new(body: { "errors" => [{ "title" => "Username or password missing" }] }) unless options[:username].present? && options[:password].present?
 
-      mds_url = options[:sandbox] ? 'https://mds.test.datacite.org' : 'https://mds.datacite.org'
+      mds_url = options[:sandbox] ? 'https://mds-sandbox.datacite.org' : 'https://mds.datacite.org'
 
       url = "#{mds_url}/metadata/#{doi}"
       Maremma.get(url, accept: 'application/xml', username: options[:username], password: options[:password], raw: true)
@@ -24,7 +24,7 @@ module Cirneco
     def delete_metadata(doi, options={})
       return OpenStruct.new(body: { "errors" => [{ "title" => "Username or password missing" }] }) unless options[:username].present? && options[:password].present?
 
-      mds_url = options[:sandbox] ? 'https://mds.test.datacite.org' : 'https://mds.datacite.org'
+      mds_url = options[:sandbox] ? 'https://mds-sandbox.datacite.org' : 'https://mds.datacite.org'
 
       url = "#{mds_url}/metadata/#{doi}"
       Maremma.delete(url, username: options[:username], password: options[:password])
@@ -35,7 +35,7 @@ module Cirneco
 
       payload = "doi=#{doi}\nurl=#{options[:url]}"
 
-      mds_url = options[:sandbox] ? 'https://mds.test.datacite.org' : 'https://mds.datacite.org'
+      mds_url = options[:sandbox] ? 'https://mds-sandbox.datacite.org' : 'https://mds.datacite.org'
 
       url = "#{mds_url}/doi/#{doi}"
       Maremma.put(url, content_type: 'text/plain;charset=UTF-8', data: payload, username: options[:username], password: options[:password])
@@ -44,7 +44,7 @@ module Cirneco
     def get_doi(doi, options={})
       return OpenStruct.new(body: { "errors" => [{ "title" => "Username or password missing" }] }) unless options[:username].present? && options[:password].present?
 
-      mds_url = options[:sandbox] ? 'https://mds.test.datacite.org' : 'https://mds.datacite.org'
+      mds_url = options[:sandbox] ? 'https://mds-sandbox.datacite.org' : 'https://mds.datacite.org'
 
       url = "#{mds_url}/doi/#{doi}"
       Maremma.get(url, username: options[:username], password: options[:password])
@@ -53,7 +53,7 @@ module Cirneco
     def get_dois(options={})
       return OpenStruct.new(body: { "errors" => [{ "title" => "Username or password missing" }] }) unless options[:username].present? && options[:password].present?
 
-      mds_url = options[:sandbox] ? 'https://mds.test.datacite.org' : 'https://mds.datacite.org'
+      mds_url = options[:sandbox] ? 'https://mds-sandbox.datacite.org' : 'https://mds.datacite.org'
 
       url = "#{mds_url}/doi"
       response = Maremma.get(url, username: options[:username], password: options[:password])
@@ -66,7 +66,7 @@ module Cirneco
 
       payload = options[:raw] ? options[:media] : options[:media].map { |m| "#{m[:mime_type]}=#{m[:url]}" }.join("\n")
 
-      mds_url = options[:sandbox] ? 'https://mds.test.datacite.org' : 'https://mds.datacite.org'
+      mds_url = options[:sandbox] ? 'https://mds-sandbox.datacite.org' : 'https://mds.datacite.org'
 
       url = "#{mds_url}/media/#{doi}"
       Maremma.post(url, content_type: 'text/plain;charset=UTF-8', data: payload, username: options[:username], password: options[:password])
@@ -75,7 +75,7 @@ module Cirneco
     def get_media(doi, options={})
       return OpenStruct.new(body: { "errors" => [{ "title" => "Username or password missing" }] }) unless options[:username].present? && options[:password].present?
 
-      mds_url = options[:sandbox] ? 'https://mds.test.datacite.org' : 'https://mds.datacite.org'
+      mds_url = options[:sandbox] ? 'https://mds-sandbox.datacite.org' : 'https://mds.datacite.org'
 
       url = "#{mds_url}/media/#{doi}"
       response = Maremma.get(url, accept: 'application/xml', username: options[:username], password: options[:password])

data/lib/cirneco/base.rb

@@ -16,6 +16,6 @@ module Cirneco
     end
 
     # default values for some ENV variables
-    ENV['MDS_URL'] ||= "https://mds.test.datacite.org"
+    ENV['MDS_URL'] ||= "https://mds-sandbox.datacite.org"
   end
 end

data/lib/cirneco/doi.rb

@@ -14,6 +14,7 @@ module Cirneco
     method_option :username, :default => ENV['MDS_USERNAME']
     method_option :password, :default => ENV['MDS_PASSWORD']
     method_option :sandbox, :type => :boolean, :force => false
+    method_option :limit, :type => :numeric, :default => 25
     def get(doi)
       if doi == "all"
         response = get_dois(options)
@@ -23,6 +24,8 @@ module Cirneco
 
       if response.body["errors"]
         puts "Error: " + response.body["errors"].first.fetch("title", "An error occured")
+      elsif doi == "all"
+        puts response.body["data"][0...options[:limit]]
       else
         puts response.body["data"]
       end
@@ -54,6 +57,14 @@ module Cirneco
       end
     end
 
+    desc "generate DOI", "generate a DOI name"
+    method_option :lower_limit, :type => :numeric, :default => 0
+    method_option :namespace, :default => 'MS-'
+    method_option :number, :type => :numeric, :aliases => '-n'
+    def accession_number
+      puts generate_accession_number(options)
+    end
+
     desc "decode DOI", "decode DOI encoded using Crockford base32 algorithm"
     def decode(doi)
       number = decode_doi(doi)
@@ -86,10 +97,10 @@ module Cirneco
     method_option :sandbox, :type => :boolean, :force => false
     def mint(filepath)
 
-      if File.directory?(filepath)
-        response = mint_dois_for_all_files(filepath, options)
+      if filepath.is_a?(Array)
+        response = mint_dois_for_all_urls(filepath, options)
       else
-        response = mint_doi_for_file(filepath, options)
+        response = mint_doi_for_url(filepath, options)
       end
 
       puts response
@@ -107,10 +118,10 @@ module Cirneco
     method_option :sandbox, :type => :boolean, :force => false
     def mint_and_hide(filepath)
 
-      if File.directory?(filepath)
-        response = mint_and_hide_dois_for_all_files(filepath, options)
+      if filepath.is_a?(Array)
+        response = mint_and_hide_dois_for_all_urls(filepath, options)
       else
-        response = mint_and_hide_doi_for_file(filepath, options)
+        response = mint_and_hide_doi_for_url(filepath, options)
       end
 
       puts response
@@ -127,10 +138,10 @@ module Cirneco
     method_option :sandbox, :type => :boolean, :force => false
     def hide(filepath)
 
-      if File.directory?(filepath)
-        response = hide_dois_for_all_files(filepath, options)
+      if filepath.is_a?(Array)
+        response = hide_dois_for_all_urls(filepath, options)
       else
-        response = hide_doi_for_file(filepath, options)
+        response = hide_doi_for_url(filepath, options)
       end
 
       puts response

data/lib/cirneco/utils.rb

@@ -1,6 +1,7 @@
 require 'base32/crockford'
 require 'securerandom'
 require 'bergamasco'
+require 'open-uri'
 require 'time'
 
 module Cirneco
@@ -8,6 +9,11 @@ module Cirneco
     # "ZZZZZZZ" decoded as number
     UPPER_LIMIT = 34359738367
 
+    LICENSES = {
+      "https://creativecommons.org/licenses/by/4.0/" => "Creative Commons Attribution (CC-BY 4.0)",
+      "https://creativecommons.org/publicdomain/zero/1.0/" => "Creative Commons Public Domain Dedication (CC0 1.0)"
+    }
+
     def get_dois_by_prefix(prefix, options={})
       response = get_dois(options)
 
@@ -23,193 +29,207 @@ module Cirneco
     end
 
     def encode_doi(prefix, options={})
-      number = options[:number] || SecureRandom.random_number(UPPER_LIMIT)
+      number = options[:number].to_s.scan(/\d+/).first.to_i
+      number = SecureRandom.random_number(UPPER_LIMIT) unless number > 0
       prefix.to_s + "/" + Base32::Crockford.encode(number, split: 4, length: 8, checksum: true)
     end
 
-    # currently only supports markdown files with YAML header
-    def mint_doi_for_file(filepath, options={})
+    def generate_accession_number(options={})
+      lower_limit = options[:lower_limit] || 0
+      namespace = options[:namespace] || 'MS-'
+      registered_numbers = options[:registered_numbers] || []
+
+      if options[:number]
+        number = options[:number].to_s
+      else
+        begin
+          number = SecureRandom.random_number(1000000) + lower_limit
+        end while registered_numbers.include? number
+        number = number.to_s
+      end
+
+      number = number.to_s.rjust(options[:length], '0') if options[:length]
+
+      if options[:split]
+        number = number.reverse
+        number = number.scan(/.{1,#{options[:split]}}/).map { |x| x.reverse }
+        number = number.reverse.join("-")
+      end
+
+      namespace + number
+    end
+
+    def get_accession_number(filepath)
+      metadata = Bergamasco::Markdown.read_yaml_for_doi_metadata(filepath)
+      metadata.fetch("accession_number", nil).to_s.scan(/\d+/).first.to_i
+    end
+
+    def get_all_accession_numbers(folderpath)
+      Dir.glob("#{folderpath}/*.md").map do |filepath|
+        get_accession_number(filepath)
+      end.select { |a| a > 0 }.sort
+    end
+
+    def update_accession_number(filepath, options={})
       filename = File.basename(filepath)
       return "File #{filename} ignored: not a markdown file" unless File.extname(filepath) == ".md"
 
       old_metadata = Bergamasco::Markdown.read_yaml_for_doi_metadata(filepath)
-      return "DOI #{old_metadata["doi"]} not changed for #{filename}" if old_metadata["doi"] && old_metadata["published"]
+      return "Accession number #{old_metadata["accession_number"]} not changed for #{filename}" if old_metadata["accession_number"]
+
+      if old_metadata["doi"].present?
+        number = decode_doi(old_metadata["doi"])
+        options[:number] = number if number > 0
+      else
+        folderpath = File.dirname(filepath)
+        options[:registered_numbers] = get_all_accession_numbers(folderpath)
+      end
+
+      accession_number = generate_accession_number(options)
+
+      new_metadata = Bergamasco::Markdown.update_file(filepath, "accession_number" => accession_number)
+      "Accession number #{new_metadata["accession_number"]} generated for #{filename}"
+    end
+
+    def update_all_accession_numbers(folderpath)
+      Dir.glob("#{folderpath}/*.md").map do |filepath|
+        update_accession_number(filepath)
+      end
+    end
+
+    # fetch schema.org metadata in JSON-LD format to mint DOI
+    def mint_doi_for_url(url, options={})
+      filename = File.basename(url)
+      source_path = options[:source_path] || "/"
+      filepath = Dir.pwd + source_path + filename + ".md"
 
-      metadata = generate_metadata_for_work(filepath, options)
-      work = post_metadata_for_work(metadata, options)
+      metadata = generate_metadata_for_work(url, options)
+      return "DOI #{metadata["doi"]} not changed for #{filename}" if metadata["doi"] && metadata["date_issued"]
 
-      # return "Errors for DOI #{metadata["doi"]}:\n#{work.validation_errors}" if work.validation_errors.present?
+      response = post_metadata_for_work(metadata, options)
+      return "Errors for DOI #{metadata["doi"]}: #{response.body['errors'].first['title']}\n" if response.body['errors']
 
       new_metadata = Bergamasco::Markdown.update_file(filepath, "doi" => metadata["doi"], "published" => true)
       "DOI #{new_metadata["doi"]} minted for #{filename}"
     end
 
-    # currently only supports markdown files with YAML header
-    def mint_and_hide_doi_for_file(filepath, options={})
-      filename = File.basename(filepath)
-      return "File #{filename} ignored: not a markdown file" unless File.extname(filepath) == ".md"
-
-      old_metadata = Bergamasco::Markdown.read_yaml_for_doi_metadata(filepath)
-      return "DOI #{old_metadata["doi"]} not changed for #{filename}" if old_metadata["doi"] && old_metadata["published"]
+    # fetch schema.org metadata in JSON-LD format to mint DOI
+    def mint_and_hide_doi_for_url(url, options={})
+      filename = File.basename(url)
+      source_path = options[:source_path] || "/"
+      filepath = Dir.pwd + source_path + filename + ".md"
 
-      metadata = generate_metadata_for_work(filepath, options)
-      work = post_and_hide_metadata_for_work(metadata, options)
+      metadata = generate_metadata_for_work(url, options)
+      return "DOI #{metadata["doi"]} not changed for #{filename}" if metadata["doi"] && metadata["date_issued"]
 
-      # return "Errors for DOI #{metadata["doi"]}:\n#{work.validation_errors}" if work.validation_errors.present?
+      response = post_metadata_for_work(metadata, options)
+      return "Errors for DOI #{metadata["doi"]}: #{response.body['errors'].first['title']}\n" if response.body['errors']
 
       new_metadata = Bergamasco::Markdown.update_file(filepath, "doi" => metadata["doi"], "published" => false)
       "DOI #{new_metadata["doi"]} minted and hidden for #{filename}"
     end
 
-    # currently only supports markdown files with YAML header
+    # fetch schema.org metadata in JSON-LD format to mint DOI
     # DOIs are never deleted, but we can remove the metadata from the DataCite index
-    def hide_doi_for_file(filepath, options={})
-      filename = File.basename(filepath)
-      return "File #{filename} ignored: not a markdown file" unless File.extname(filepath) == ".md"
-
-      old_metadata = Bergamasco::Markdown.read_yaml_for_doi_metadata(filepath)
-      return "DOI #{old_metadata["doi"]} not changed for #{filename}" unless old_metadata["doi"] && old_metadata["published"]
+    def hide_doi_for_url(url, options={})
+      filename = File.basename(url)
+      source_path = options[:source_path] || "/"
+      filepath = Dir.pwd + source_path + filename + ".md"
 
-      metadata = generate_metadata_for_work(filepath, options)
-      work = hide_metadata_for_work(metadata, options)
+      metadata = generate_metadata_for_work(url, options)
+      return "DOI #{metadata["doi"]} not changed for #{filename}" unless metadata["doi"] && metadata["date_issued"]
 
-      # return "Errors for DOI #{old_metadata["doi"]}:\n#{work.validation_errors}" if work.validation_errors.present?
+      response = hide_metadata_for_work(metadata, options)
+      return "Errors for DOI #{metadata["doi"]}: #{response.body['errors'].first['title']}\n" if response.body['errors']
 
       new_metadata = Bergamasco::Markdown.update_file(filepath, "published" => false)
-      "DOI #{old_metadata["doi"]} hidden for #{filename}"
+      "DOI #{metadata["doi"]} hidden for #{filename}"
     end
 
-    def mint_dois_for_all_files(folderpath, options={})
-      Dir.glob("#{folderpath}/*.md").map do |filepath|
-        mint_doi_for_file(filepath, options)
+    def mint_dois_for_all_urls(urls, options={})
+      urls.map do |url|
+        mint_doi_for_url(url, options)
       end.join("\n")
     end
 
-    def mint_and_hide_dois_for_all_files(folderpath, options={})
-      Dir.glob("#{folderpath}/*.md").map do |filepath|
-        mint_and_hide_doi_for_file(filepath, options)
+    def mint_and_hide_dois_for_all_urls(urls, options={})
+      urls.map do |url|
+        mint_and_hide_doi_for_url(url, options)
       end.join("\n")
     end
 
-    def hide_dois_for_all_files(folderpath, options={})
-      Dir.glob("#{folderpath}/*.md").map do |filepath|
-        hide_doi_for_file(filepath, options)
+    def hide_dois_for_all_urls(urls, options={})
+      urls.map do |url|
+        hide_doi_for_url(url, options)
       end.join("\n")
     end
 
-    def generate_metadata_for_work(filepath, options={})
-      sitepath = options[:sitepath] || ENV['SITE_SITEPATH'] || "data/site.yml"
-      authorpath = options[:authorpath] || ENV['SITE_AUTHORPATH'] || "data/authors.yml"
-      referencespath = options[:referencespath] || ENV['SITE_REFERENCESPATH'] || "data/references.yaml"
-      csl = options[:csl] || ENV['SITE_CSLPATH'] || "styles/apa.csl"
-      options = options.merge(csl: csl, bibliography: referencespath)
+    def generate_metadata_for_work(url, options={})
+      doc = Nokogiri::HTML(open(url))
+      json = doc.at_xpath("//script[@type='application/ld+json']")
+      return "Error: no schema.org metadata found" unless json.present?
 
-      metadata = Bergamasco::Markdown.read_yaml_for_doi_metadata(filepath, options.except(:number))
+      metadata = ActiveSupport::JSON.decode(json.text)
 
-      return "Error: required metadata missing" unless ["author", "title", "date", "summary"].all? { |k| metadata.key? k }
-
-      # read in optional yaml configuration files for site, author and references
-      site_options = Bergamasco::Markdown.read_yaml(sitepath) || {}
-      author_options = Bergamasco::Markdown.read_yaml(authorpath) || {}
-      references = Bergamasco::Markdown.read_yaml(referencespath) || {}
+      return "Error: required metadata missing" unless ["name", "author", "publisher", "datePublished", "@type"].all? { |k| metadata.key? k }
 
       # required metadata
-      prefix = options[:prefix] || site_options["prefix"] || ENV['PREFIX']
-      metadata["doi"] ||= encode_doi(prefix, options)
-
-      site_url = site_options["site_url"] || ENV['SITE_URL']
-      metadata["url"] ||= url_from_path(site_url, filepath)
-
-      metadata["creators"] = Array(metadata["author"]).map do |a|
-        author = author_options.fetch(a, {})
-        if author.present?
-          { given_name: author["given"],
-            family_name: author["family"],
-            orcid: author["orcid"] }
-        else
-          { literal: a }
-        end
+      if /(http|https):\/\/(dx\.)?doi\.org\/(\w+)/.match(metadata["@id"])
+        uri = Addressable::URI.parse(metadata["@id"])
+        metadata["doi"] = uri.path[1..-1].upcase
       end
 
-      metadata["publisher"] = site_options["site_title"] || ENV['SITE_TITLE']
-      metadata["publication_year"] = metadata["date"][0..3].to_i
+      metadata["title"] = metadata["name"]
+
+      metadata["creators"] = format_authors(metadata["author"])
+
+      metadata["publisher"] = metadata.fetch("publisher", {}).fetch("name", nil)
+      metadata["publication_year"] = metadata.fetch("datePublished", "")[0..3].to_i
 
-      metadata["type"] ||= site_options["default_type"] || ENV['SITE_DEFAULT_TYPE'] || "BlogPosting"
-      resource_type_general = metadata["type"] == "Dataset" ? "Dataset" : "Text"
+      resource_type_general = metadata["@type"] == "Dataset" ? "Dataset" : "Text"
 
-      metadata["resource_type"] = { value: metadata["type"],
+      metadata["resource_type"] = { value: metadata["@type"],
                                     resource_type_general: resource_type_general }
 
       # recommended metadata
-      metadata["descriptions"] = [{ value: metadata["summary"],
-                                    description_type: "Abstract" }]
+
+      # use alternate_identifier to generate DOI
+      metadata["alternate_identifier"] = metadata["alternateName"]
+
+      if metadata["description"].present?
+        metadata["descriptions"] = [{ value: metadata["description"], description_type: "Abstract" }]
+      end
 
       # use default version 1.0
       metadata["version"] ||= "1.0"
 
       # fetch reference metadata if available
-      metadata["related_identifiers"] = Array(metadata["references"]).map do |r|
-        reference = references.fetch(r, {})
-        if reference.present?
-          if reference["DOI"].present?
-            value = reference["DOI"].upcase
-            type = "DOI"
-          elsif /(http|https):\/\/(dx\.)?doi\.org\/(\w+)/.match(reference["URL"])
-            uri = Addressable::URI.parse(reference["URL"])
-            value = uri.path[1..-1].upcase
-            type = "DOI"
-          elsif reference["URL"].present?
-            value = reference["URL"]
-            type = "URL"
-          else
-            type = nil
-          end
-        else
-          if /(http|https):\/\/(dx\.)?doi\.org\/(\w+)/.match(r)
-            uri = Addressable::URI.parse(r)
-            value = uri.path[1..-1].upcase
-            type = "DOI"
-          elsif /(http|https):\/\//.match(r)
-            uri = Addressable::URI.parse(r)
-            value = uri.normalize.to_s
-            type = "URL"
-          else
-            type = nil
-          end
-        end
-
-        {
-          value: value,
-          related_identifier_type: type,
-          relation_type: "References"
-        }
-      end.select { |t| t[:related_identifier_type].present? }
-
-      metadata["license_name"] = site_options.fetch("license", {}).fetch("name", nil) || ENV['SITE_LICENCE_NAME'] || "Creative Commons Attribution"
-      metadata["license_url"] = site_options.fetch("license", {}).fetch("url", nil) || ENV['SITE_LICENCE_URL'] || "https://creativecommons.org/licenses/by/4.0/"
-      metadata["rights_list"] = [{ value: metadata["license_name"], rights_uri: metadata["license_url"] }]
+      metadata["related_identifiers"] = get_related_identifiers(metadata)
 
-      metadata["subjects"] = Array(metadata["tags"]).select { |t| t != "featured" }
+      if metadata["license"].present?
+        value = LICENSES.fetch(metadata["license_url"], nil)
+        metadata["rights_list"] = [{ value: value, rights_uri: metadata["license"] }] if value.present?
+      end
 
-      contributor = site_options["institution"] || ENV['SITE_INSTITUTION']
-      metadata["contributors"] = [{ literal: contributor, contributor_type: "HostingInstitution" }]
+      if metadata["keywords"].present?
+        metadata["subjects"] = Array(metadata["keywords"].split(", ")).select { |k| k != "featured" }
+      end
 
-      metadata["date_issued"] = metadata["date"]
-      metadata["publication_month"] = metadata["date"][5..6]
-      metadata["publication_day"] = metadata["date"][8..9]
+      metadata["date_created"] = metadata["dateCreated"]
+      metadata["date_issued"] = metadata["datePublished"]
+      metadata["date_updated"] = metadata["dateModified"]
 
-      metadata = metadata.extract!(*%w(doi url creators title publisher
-        publication_year publication_month publication_day resource_type descriptions version license_name license_url rights_list subjects contributors
-        date_issued related_identifiers))
-    end
-
-    def url_from_path(site_url, filepath)
-      site_url.to_s.chomp("\\") + "/" + File.basename(filepath)[0..-9] + "/"
+      metadata = metadata.extract!(*%w(doi alternate_identifier url creators title
+        publisher publication_year resource_type descriptions version rights_list
+        subjects date_issued date_created date_updated related_identifiers))
     end
 
     def post_metadata_for_work(metadata, options={})
+      prefix = options[:prefix] || ENV['PREFIX']
+      metadata["doi"] ||= encode_doi(prefix, number: metadata["alternate_identifier"])
+
       work = Cirneco::Work.new(metadata)
-      return work.validation_errors if work.validation_errors.present?
+      return work.validation_errors if work.validation_errors.body["errors"].present?
 
       response = work.post_metadata(work.data, options)
       return response unless response.status == 201
@@ -218,8 +238,11 @@ module Cirneco
     end
 
     def post_and_hide_metadata_for_work(metadata, options={})
+      prefix = options[:prefix] || ENV['PREFIX']
+      metadata["doi"] ||= encode_doi(prefix, number: metadata["alternate_identifier"])
+
       work = Cirneco::Work.new(metadata)
-      return work.validation_errors if work.validation_errors.present?
+      return work.validation_errors if work.validation_errors.body["errors"].present?
 
       response = work.post_metadata(work.data, options)
       return response unless response.status == 201
@@ -231,8 +254,11 @@ module Cirneco
     end
 
     def hide_metadata_for_work(metadata, options={})
+      prefix = options[:prefix] || ENV['PREFIX']
+      metadata["doi"] ||= encode_doi(prefix, number: metadata["alternate_identifier"])
+
       work = Cirneco::Work.new(metadata)
-      return work.validation_errors if work.validation_errors.present?
+      return work.validation_errors if work.validation_errors.body["errors"].present?
 
       work.delete_metadata(metadata["doi"], options)
     end
@@ -240,5 +266,55 @@ module Cirneco
     def generate_jats(filepath, options={})
       Bergamasco::Pandoc.write_jats(filepath, options)
     end
+
+    def url_from_path(site_url, filepath)
+      site_url.to_s.chomp("\\") + "/" + File.basename(filepath)[0..-9] + "/"
+    end
+
+    def format_authors(authors)
+      Array(authors).map do |author|
+        orcid = orcid_from_url(author["@id"])
+        name = (author["givenName"].present? || author["familyName"].present?) ? nil : author["name"]
+
+        { given_name: author["givenName"],
+          family_name: author["familyName"],
+          name: name,
+          orcid: orcid }.compact
+      end
+    end
+
+    def get_related_identifiers(metadata)
+      citations = Array(metadata["citation"])
+      parts = Array(metadata["IsPartOf"]).map { |r| r["relation_type"] = "IsPartOf" }
+
+      (citations + parts).map do |r|
+        id = r.fetch("@id", "")
+        relation_type = r.fetch("relation_type", "References")
+
+        if /(http|https):\/\/(dx\.)?doi\.org\/(\w+)/.match(id)
+          uri = Addressable::URI.parse(id)
+          value = uri.path[1..-1].upcase
+          type = "DOI"
+        elsif /(http|https):\/\//.match(id)
+          uri = Addressable::URI.parse(id)
+          value = uri.normalize.to_s
+          type = "URL"
+        else
+          type = nil
+        end
+
+        {
+          value: value,
+          related_identifier_type: type,
+          relation_type: relation_type
+        }
+      end.select { |t| t[:related_identifier_type].present? }
+    end
+
+    def orcid_from_url(url)
+      return nil unless url.present?
+
+      Array(/\Ahttp:\/\/orcid\.org\/(.+)/.match(url)).last
+    end
   end
 end

data/lib/cirneco/version.rb

@@ -1,3 +1,3 @@
 module Cirneco
-  VERSION = "0.7.4"
+  VERSION = "0.8.1"
 end