cirneco 0.8.12 → 0.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: cb4a217d9d1a32ba1d92b3d505e53903ccd2b25b
-  data.tar.gz: c51d52740d43affa17960e3e5382208b93ea0cfb
+  metadata.gz: 57b29c4bbd81d6692811c2e2c4f56887195f3c0a
+  data.tar.gz: 5758d8072bb5e32569859f06c7ad8e8b26d2f71e
 SHA512:
-  metadata.gz: d18feee400aae01050747ff636fb13fd31b61f2c36e3cd9fa5fc4236fbc6d7d5d45e82aa0b64a2d4d2b53f679b115a1398ee7d0146f496a608b90b82e8ce8112
-  data.tar.gz: 0a4a8efc2685ec9460eb9e42ab793ac65dfebab386289d8bd1104202ab0010fc1521e6063595e3b97915d60939d6ef7e10099fa64c2cec5f603f4413ea1ead20
+  metadata.gz: 1e08e2b24e7af2e0b41a37c68fee12b0309b640d61d8dadb6f4efdca8240b6b3185cc8ec877a893c3632c0226dc2130daa5f082df1c761a577cabce26d39cf98
+  data.tar.gz: 187f2c7707e0217fdefb9de490b02290cbaa4d00633c0723315d26e4064106f12d14b09345c9770b91d811d5f88b553c2469523225fe053bdf74670ae3214144

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    cirneco (0.8.12)
+    cirneco (0.9)
       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 (0.3.6)
+    bergamasco (0.3.9)
       activesupport (~> 4.2, >= 4.2.5)
       addressable (~> 2.3.8)
       builder (~> 3.2, >= 3.2.2)
@@ -122,4 +122,4 @@ DEPENDENCIES
   webmock (~> 1.22, >= 1.22.3)
 
 BUNDLED WITH
-   1.13.6
+   1.12.5

data/lib/cirneco/doi.rb

@@ -153,5 +153,18 @@ module Cirneco
       response = hide_dois_for_all_urls(url, options)
       puts response
     end
+
+    desc "write_jats DOCUMENT", "write_jats document"
+    method_option :sitepath, :default => ENV['SITE_SITEPATH']
+    method_option :authorpath, :default => ENV['SITE_AUTHORPATH']
+    method_option :referencespath, :default => ENV['SITE_REFERENCESPATH']
+    method_option :source_dir, :default => ENV['SOURCE_DIR']
+    method_option :build_dir, :default => ENV['BUILD_DIR']
+    method_option :posts_dir, :default => ENV['POSTS_DIR']
+    method_option :csl, :default => ENV['SITE_CSLPATH']
+    def write_jats(url)
+      response = generate_jats_for_all_urls(url, options)
+      puts response
+    end
   end
 end

data/lib/cirneco/utils.rb

@@ -279,8 +279,82 @@ module Cirneco
       work.delete_metadata(metadata["doi"], options)
     end
 
-    def generate_jats(filepath, options={})
-      Bergamasco::Pandoc.write_jats(filepath, options)
+    def generate_metadata_for_jats(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 = ActiveSupport::JSON.decode(json.text)
+
+      return "Error: required metadata missing" unless ["name", "author", "publisher", "datePublished", "@type"].all? { |k| metadata.key? k }
+
+      # required metadata
+      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["title"] = metadata["name"]
+      metadata["author"] = format_authors(metadata["author"]).map do |a|
+        { "given_name" => a[:given_name],
+          "family_name" => a[:family_name],
+          "name" => a[:name],
+          "orcid" => a[:orcid] }.compact
+      end
+
+      metadata["publisher"] = metadata.fetch("publisher", {}).fetch("name", nil)
+      metadata["tags"] = metadata["keywords"].to_s.split(", ").select { |k| k != "featured" }
+      metadata["date"] = metadata.fetch("datePublished", "")
+      metadata["publication_year"] = metadata.fetch("date", "")[0..3].to_i
+      metadata["publication_month"] = metadata.fetch("date", "")[5..6].to_i
+      metadata["publication_day"] = metadata.fetch("date", "")[8..9].to_i
+
+      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"] = get_related_identifiers(metadata)
+
+      if metadata["license"].present?
+        metadata["license_name"] = LICENSES.fetch(metadata["license"], nil)
+        metadata["license_url"] = metadata["license"]
+      end
+
+      metadata = metadata.extract!(*%w(publisher doi tags title author date
+        publication_year publication_month publication_day license_name
+        license_url))
+    end
+
+    def generate_jats_for_url(url, options={})
+      filename, build_path, source_path = filepath_from_url(url, options)
+      metadata = generate_metadata_for_jats(build_path, options)
+      file = IO.read(source_path)
+      content = Bergamasco::Markdown.split_yaml_frontmatter(file).last
+      text = Bergamasco::Markdown.join_yaml_frontmatter(metadata, content)
+
+      xml = Bergamasco::Pandoc.convert_to_jats(text, options)
+
+      if metadata["doi"].present?
+        xmlname = metadata["doi"].split('/', 2).last
+      else
+        xmlname = filename.gsub(/\.html\.(erb|md)/, ".xml")
+      end
+
+      xmlpath = build_path.gsub("index.html", xmlname)
+      IO.write(xmlpath, xml)
+
+      "JATS XML written for #{filename}"
+    end
+
+    def generate_jats_for_all_urls(url, options={})
+      urls = get_urls_for_works(url)
+      urls.map do |u|
+        generate_jats_for_url(u, options)
+      end.join("\n")
     end
 
     def url_from_path(site_url, filepath)
@@ -292,10 +366,10 @@ module Cirneco
         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
+        { :given_name => author["givenName"],
+          :family_name => author["familyName"],
+          :name => name,
+          :orcid => orcid }.compact
       end
     end
 

data/lib/cirneco/version.rb

@@ -1,3 +1,3 @@
 module Cirneco
-  VERSION = "0.8.12"
+  VERSION = "0.9"
 end

data/spec/doi_spec.rb

@@ -15,7 +15,7 @@ describe Cirneco::Doi do
   let(:username) { ENV['MDS_USERNAME'] }
   let(:password) { ENV['MDS_PASSWORD'] }
   let(:csl) { "spec/fixtures/apa.csl" }
-  let(:bibliography) { "spec/fixtures/bibliography.yaml" }
+  let(:bibliography) { "spec/fixtures/references.yaml" }
   let(:api_options) { { username: username, password: password, sandbox: true } }
   let(:mint_options) { { username: username, password: password, sandbox: true, source_dir: "/spec/fixtures/", build_dir: "/spec/fixtures/", csl: csl, bibliography: bibliography } }
 
@@ -122,6 +122,14 @@ describe Cirneco::Doi do
       expect { subject.mint_and_hide filepath }.to output("DOI 10.5072/0000-03VC minted and hidden for cool-dois.html.md\nDOI 10.5072/0000-00SS minted and hidden for index.html.erb\n").to_stdout
     end
   end
+
+  context "jats" do
+    it 'writes jats for list of urls' do
+      filepath = fixture_path + 'index.html'
+      subject.options = mint_options
+      expect { subject.write_jats filepath }.to output("JATS XML written for cool-dois.html.md\nJATS XML written for index.html.erb\n").to_stdout
+    end
+  end
 end
 
 

data/spec/utils_spec.rb

@@ -194,14 +194,26 @@ describe Cirneco::DataCenter, vcr: true, :order => :defined do
       expect(response.body["data"]).to eq("OK")
       expect(response.status).to eq(200)
     end
+  end
 
-    # it 'should generate jats xml' do
-    #   filepath = fixture_path + 'cool-dois.html.md'
-    #   number = 123
-    #   metadata = subject.generate_metadata_for_work(filepath)
-    #   xml_path = subject.generate_jats(filepath, options.merge(metadata: metadata))
-    #   expect(xml_path).to eq(fixture_path + 'cool-dois.xml')
-    # end
+  context "jats" do
+    it 'should generate metadata for jats' do
+      filepath = fixture_path + 'cool-dois/index.html'
+      metadata = subject.generate_metadata_for_jats(filepath)
+      expect(metadata["author"]).to eq([{"given_name"=>"Martin", "family_name"=>"Fenner", "orcid"=>"0000-0003-1419-2405"}])
+      expect(metadata["license_url"]).to eq("https://creativecommons.org/licenses/by/4.0/")
+    end
+
+    it 'should generate jats xml' do
+      filepath = fixture_path + 'cool-dois/index.html'
+      expect(subject.generate_jats_for_url(filepath, options)).to eq("JATS XML written for cool-dois.html.md")
+    end
+
+    it 'should generate jats for all urls' do
+      filepath = fixture_path + 'index.html'
+      response = subject.generate_jats_for_all_urls(filepath, options)
+      expect(response).to eq("JATS XML written for cool-dois.html.md\nJATS XML written for index.html.erb")
+    end
   end
 
   context "get_related_identifiers" do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: cirneco
 version: !ruby/object:Gem::Version
-  version: 0.8.12
+  version: '0.9'
 platform: ruby
 authors:
 - Martin Fenner