bolognese 0.8.13 → 0.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 2369ba9f739cdf61b879479f95a5732f967eb37e
-  data.tar.gz: 6d707dfe7018625eec5fcb0717562be17cabebdf
+  metadata.gz: 290c7cb9287a0b5e018ff3392821d45d41af4360
+  data.tar.gz: 9e687bbe98ed2bafc223bb49044b8fc2ea2a44d3
 SHA512:
-  metadata.gz: 3f634c7a9ca25143d6d0af60e147a753de70471d55016f98bf2f5f75fbd2cc509c19c731ddca33ce25823ae37674763c674689a316491363c5af1bf73886ef5a
-  data.tar.gz: 1958c1d8d9938e8c26ed04b3f39100f3f0a95822ff3959f3ad6ac501ee3f0c85654f47f5c1deb579e73d07f4c657ae4c67c6d9ed470a8a9c7961d6a9405c01a3
+  metadata.gz: 2dcaac0f7918491a15d02f08154e94130cdd2c5579c65669e04bce2ec27643f3a4376c70537a700ea61ce3bc18d634bab3de152555d6c68ac5110ccfa43bf96e
+  data.tar.gz: 2add919c630f4c2f211337d36366a556b3e9e00c6b41e5a82cf145064c8bf7ca1ad815eeaae155fd20d7fcee3d431af34fd4ed024726cf43a64e3278cc89bc55

data/Gemfile.lock

@@ -1,16 +1,20 @@
 PATH
   remote: .
   specs:
-    bolognese (0.8.13)
+    bolognese (0.9)
       activesupport (~> 4.2, >= 4.2.5)
       bibtex-ruby (~> 4.1)
       builder (~> 3.2, >= 3.2.2)
       colorize (~> 0.8.1)
+      json-ld (~> 2.1, >= 2.1.2)
+      jsonlint (~> 0.2.0)
       loofah (~> 2.0, >= 2.0.3)
       maremma (~> 3.5)
       namae (~> 0.10.2)
-      nokogiri (~> 1.7, >= 1.7.1)
+      nokogiri (~> 1.6, >= 1.6.8)
       postrank-uri (~> 1.0, >= 1.0.18)
+      rdf-rdfxml (~> 2.0)
+      rdf-turtle (~> 2.2)
       thor (~> 0.19)
 
 GEM
@@ -29,10 +33,14 @@ GEM
     codeclimate-test-reporter (1.0.8)
       simplecov (<= 0.13)
     colorize (0.8.1)
+    concurrent-ruby (1.0.5)
     crack (0.4.3)
       safe_yaml (~> 1.0.0)
     diff-lcs (1.3)
     docile (1.1.5)
+    ebnf (1.1.0)
+      rdf (~> 2.0)
+      sxp (~> 1.0)
     excon (0.45.4)
     faraday (0.9.2)
       multipart-post (>= 1.2, < 3)
@@ -40,14 +48,26 @@ GEM
       faraday
     faraday_middleware (0.10.1)
       faraday (>= 0.7.4, < 1.0)
+    haml (4.0.7)
+      tilt
+    hamster (3.0.0)
+      concurrent-ruby (~> 1.0)
     hashdiff (0.3.2)
+    htmlentities (4.3.4)
     i18n (0.8.1)
-    json (2.0.3)
+    json (2.0.4)
+    json-ld (2.1.2)
+      multi_json (~> 1.12)
+      rdf (~> 2.1)
+    jsonlint (0.2.0)
+      oj (~> 2)
+      trollop (~> 2)
     latex-decode (0.2.2)
       unicode (~> 0.4)
+    link_header (0.0.8)
     loofah (2.0.3)
       nokogiri (>= 1.5.9)
-    maremma (3.5.4)
+    maremma (3.5.6)
       activesupport (~> 4.2, >= 4.2.5)
       addressable (>= 2.3.6)
       builder (~> 3.2, >= 3.2.2)
@@ -55,12 +75,12 @@ GEM
       faraday (~> 0.9.2)
       faraday-encoding (~> 0.0.1)
       faraday_middleware (~> 0.10.0)
-      multi_json (~> 1.11.2)
-      nokogiri (~> 1.7, >= 1.7.1)
+      multi_json (~> 1.12)
+      nokogiri (~> 1.6, >= 1.6.8)
       oj (~> 2.18, >= 2.18.1)
     mini_portile2 (2.1.0)
     minitest (5.10.1)
-    multi_json (1.11.3)
+    multi_json (1.12.1)
     multipart-post (2.0.0)
     namae (0.10.2)
     nokogiri (1.7.1)
@@ -75,6 +95,27 @@ GEM
     rack-test (0.6.3)
       rack (>= 1.0)
     rake (12.0.0)
+    rdf (2.2.5)
+      hamster (~> 3.0)
+      link_header (~> 0.0, >= 0.0.8)
+    rdf-aggregate-repo (2.2.0)
+      rdf (~> 2.0)
+    rdf-rdfa (2.1.1)
+      haml (~> 4.0)
+      htmlentities (~> 4.3)
+      rdf (~> 2.0)
+      rdf-aggregate-repo (~> 2.0)
+      rdf-xsd (~> 2.0)
+    rdf-rdfxml (2.0.0)
+      htmlentities (~> 4.3)
+      rdf (~> 2.0)
+      rdf-rdfa (~> 2.0)
+      rdf-xsd (~> 2.0)
+    rdf-turtle (2.2.0)
+      ebnf (~> 1.1)
+      rdf (~> 2.2)
+    rdf-xsd (2.1.0)
+      rdf (~> 2.1)
     rspec (3.5.0)
       rspec-core (~> 3.5.0)
       rspec-expectations (~> 3.5.0)
@@ -97,8 +138,12 @@ GEM
       json (>= 1.8, < 3)
       simplecov-html (~> 0.10.0)
     simplecov-html (0.10.0)
+    sxp (1.0.0)
+      rdf (~> 2.0)
     thor (0.19.4)
     thread_safe (0.3.6)
+    tilt (2.0.7)
+    trollop (2.1.2)
     tzinfo (1.2.3)
       thread_safe (~> 0.1)
     unicode (0.4.4.4)
@@ -124,4 +169,4 @@ DEPENDENCIES
   webmock (~> 1.22, >= 1.22.3)
 
 BUNDLED WITH
-   1.14.6
+   1.12.5

data/README.md

@@ -51,6 +51,20 @@ Bolognese reads and/or writes these metadata formats:
       <td>Yes</td>
       <td>Yes</td>
     </tr>
+    <tr>
+      <td><a href='http://www.w3.org/TR/rdf-syntax-grammar/'>RDF XML</a></td>
+      <td>rdf_xml</td>
+      <td>application/rdf+xml</td>
+      <td>No</td>
+      <td>Yes</td>
+    </tr>
+    <tr>
+      <td><a href='http://www.w3.org/TeamSubmission/turtle/'>RDF Turtle</a></td>
+      <td>turtle</td>
+      <td>text/turtle</td>
+      <td>No</td>
+      <td>Yes</td>
+    </tr>
     <tr>
       <td><a href='http://gsl-nagoya-u.net/http/pub/citeproc-doc.html'>Citeproc JSON</a></td>
       <td>citeproc</td>
@@ -61,8 +75,15 @@ Bolognese reads and/or writes these metadata formats:
     <tr>
       <td><a href='https://codemeta.github.io/'>Codemeta</a></td>
       <td>codemeta</td>
-      <td>application/ld+json</td>
+      <td>application/vnd.codemeta.ld+json</td>
+      <td>Yes</td>
       <td>Yes</td>
+    </tr>
+    <tr>
+      <td><a href='https://github.com/scholix/schema'>Scholix</a></td>
+      <td>scholix</td>
+      <td>application/vnd.scholix+json</td>
+      <td>No</td>
       <td>Yes</td>
     </tr>
     <tr>

data/bolognese.gemspec

@@ -16,7 +16,7 @@ Gem::Specification.new do |s|
 
   # Declary dependencies here, rather than in the Gemfile
   s.add_dependency 'maremma', '~> 3.5'
-  s.add_dependency 'nokogiri', '~> 1.7', '>= 1.7.1'
+  s.add_dependency 'nokogiri', '~> 1.6', '>= 1.6.8'
   s.add_dependency 'loofah', '~> 2.0', '>= 2.0.3'
   s.add_dependency 'builder', '~> 3.2', '>= 3.2.2'
   s.add_dependency 'activesupport', '~> 4.2', '>= 4.2.5'
@@ -25,6 +25,10 @@ Gem::Specification.new do |s|
   s.add_dependency 'colorize', '~> 0.8.1'
   s.add_dependency 'namae', '~> 0.10.2'
   s.add_dependency 'postrank-uri', '~> 1.0', '>= 1.0.18'
+  s.add_dependency 'json-ld', '~> 2.1', '>= 2.1.2'
+  s.add_dependency 'jsonlint', '~> 0.2.0'
+  s.add_dependency 'rdf-turtle', '~> 2.2'
+  s.add_dependency 'rdf-rdfxml', '~> 2.0'
   s.add_development_dependency 'bundler', '~> 1.0'
   s.add_development_dependency 'rspec', '~> 3.4'
   s.add_development_dependency 'rspec-xsd', '~> 0.1.0'

data/lib/bolognese.rb

@@ -5,6 +5,11 @@ require 'postrank-uri'
 require 'bibtex'
 require 'colorize'
 require 'loofah'
+require 'json/ld'
+require 'rdf/turtle'
+require 'rdf/rdfxml'
+require 'logger'
+require 'jsonlint'
 
 require "bolognese/version"
 require "bolognese/metadata"

data/lib/bolognese/citeproc.rb

@@ -15,15 +15,19 @@ module Bolognese
     end
 
     def metadata
-      @metadata ||= raw.present? ? Maremma.from_json(raw) : {}
+      @metadata ||= raw.present? && valid? ? Maremma.from_json(raw) : {}
     end
 
     def exists?
       metadata.present?
     end
 
+    def errors
+      jsonlint(raw)
+    end
+
     def valid?
-      citeproc.present?
+      errors.empty?
     end
 
     def doi

data/lib/bolognese/codemeta.rb

@@ -13,15 +13,19 @@ module Bolognese
     end
 
     def metadata
-      @metadata ||= raw.present? ? Maremma.from_json(raw) : {}
+      @metadata ||= raw.present? && valid? ? Maremma.from_json(raw) : {}
     end
 
     def exists?
       metadata.present?
     end
 
+    def errors
+      jsonlint(raw)
+    end
+
     def valid?
-      codemeta.present?
+      errors.empty?
     end
 
     def doi

data/lib/bolognese/crossref.rb

@@ -68,7 +68,7 @@ module Bolognese
       elsif id.present?
         response = Maremma.get(id, accept: "application/vnd.crossref.unixref+xml", host: true, raw: true)
         @raw = response.body.fetch("data", nil)
-        @raw = Nokogiri::XML(@raw, &:noblanks).to_s if @raw.present?
+        @raw = Nokogiri::XML(@raw, nil, 'UTF-8', &:noblanks).to_s if @raw.present?
       end
     end
 

data/lib/bolognese/datacite.rb

@@ -11,7 +11,7 @@ module Bolognese
       elsif id.present?
         response = Maremma.get(id, accept: "application/vnd.datacite.datacite+xml", raw: true)
         @raw = response.body.fetch("data", nil)
-        @raw = Nokogiri::XML(@raw, &:noblanks).to_s if @raw.present?
+        @raw = Nokogiri::XML(@raw, nil, 'UTF-8', &:noblanks).to_s if @raw.present?
       end
 
       @should_passthru = !regenerate
@@ -35,7 +35,7 @@ module Bolognese
     end
 
     def errors
-      schema.validate(Nokogiri::XML(raw)).map { |error| error.to_s }.unwrap
+      schema.validate(Nokogiri::XML(raw, nil, 'UTF-8')).map { |error| error.to_s }.unwrap
     rescue Nokogiri::XML::SyntaxError => e
       e.message
     end

data/lib/bolognese/datacite_json.rb

@@ -8,11 +8,15 @@ module Bolognese
     end
 
     def metadata
-      @metadata ||= raw.present? ? Maremma.from_json(raw) : {}
+      @metadata ||= raw.present? && valid? ? Maremma.from_json(raw) : {}
+    end
+
+    def errors
+      jsonlint(raw)
     end
 
     def valid?
-      datacite_json.present?
+      errors.empty?
     end
 
     def datacite
@@ -32,11 +36,11 @@ module Bolognese
     end
 
     def type
-      DC_TO_SO_TRANSLATIONS[resource_type_general.to_s.dasherize] || "CreativeWork"
+      DC_TO_SO_TRANSLATIONS[resource_type_general.to_s.dasherize] || valid?.presence && "CreativeWork"
     end
 
     def citeproc_type
-      DC_TO_CP_TRANSLATIONS[resource_type_general.to_s.dasherize] || "other"
+      DC_TO_CP_TRANSLATIONS[resource_type_general.to_s.dasherize] || valid?.presence && "other"
     end
 
     def ris_type
@@ -204,7 +208,7 @@ module Bolognese
     end
 
     def publisher
-      metadata.fetch("publisher")
+      metadata.fetch("publisher", nil)
     end
 
     alias_method :container_title, :publisher

data/lib/bolognese/datacite_utils.rb

@@ -19,7 +19,7 @@ module Bolognese
     end
 
     def datacite_errors
-      @datacite_errors ||= schema.validate(Nokogiri::XML(datacite)).map { |error| error.to_s }
+      @datacite_errors ||= schema.validate(Nokogiri::XML(datacite, nil, 'UTF-8')).map { |error| error.to_s }
     end
 
     def insert_work(xml)

data/lib/bolognese/metadata.rb

@@ -237,8 +237,8 @@ module Bolognese
         "isBasedOn" => Array.wrap(is_supplement_to).map { |r| { "@id" => r["id"] }}.unwrap }.compact
     end
 
-    def schema_org
-      hsh = {
+    def schema_hsh
+      {
         "@context" => id.present? ? "http://schema.org" : nil,
         "@type" => type,
         "@id" => id,
@@ -266,13 +266,28 @@ module Bolognese
         "predecessor_of" => is_previous_version_of,
         "successor_of" => is_new_version_of,
         "citation" => Array.wrap(references).map { |r| r.except("relationType").merge("@type" => "CreativeWork") }.unwrap,
-        "@reverse" => reverse,
+        "@reverse" => reverse.presence,
         "schemaVersion" => schema_version,
-        "publisher" => { "@type" => "Organization", "name" => publisher },
+        "publisher" => publisher.present? ? { "@type" => "Organization", "name" => publisher } : nil,
         "funder" => funder,
-        "provider" => { "@type" => "Organization", "name" => provider }
-      }.compact
-      JSON.pretty_generate hsh
+        "provider" => provider.present? ? { "@type" => "Organization", "name" => provider } : nil
+      }.compact.presence
+    end
+
+    def schema_org
+      JSON.pretty_generate schema_hsh
+    end
+
+    def graph
+      RDF::Graph.new << JSON::LD::API.toRdf(schema_hsh)
+    end
+
+    def turtle
+      graph.dump(:ttl, prefixes: { schema: "http://schema.org/" })
+    end
+
+    def rdf_xml
+      graph.dump(:rdfxml, prefixes: { schema: "http://schema.org/" })
     end
 
     def datacite_json
@@ -309,7 +324,7 @@ module Bolognese
         "schemaVersion" => schema_version,
         "provider" => provider
       }.compact
-      JSON.pretty_generate hsh
+      JSON.pretty_generate hsh.presence
     end
 
     def citeproc
@@ -333,7 +348,7 @@ module Bolognese
         "version" => version,
         "volume" => volume
       }.compact
-      JSON.pretty_generate hsh
+      JSON.pretty_generate hsh.presence
     end
 
     def codemeta
@@ -353,7 +368,7 @@ module Bolognese
         "dateModified" => date_modified,
         "publisher" => publisher
       }.compact
-      JSON.pretty_generate hsh
+      JSON.pretty_generate hsh.presence
     end
 
     def bibtex
@@ -394,5 +409,12 @@ module Bolognese
         "ER" => ""
       }.compact.map { |k, v| v.is_a?(Array) ? v.map { |vi| "#{k} - #{vi}" }.join("\r\n") : "#{k} - #{v}" }.join("\r\n")
     end
+
+    def jsonlint(json)
+      error_array = []
+      linter = JsonLint::Linter.new
+      linter.send(:check_data, json, error_array)
+      error_array
+    end
   end
 end

data/lib/bolognese/schema_org.rb

@@ -17,21 +17,26 @@ module Bolognese
         @raw = string
       elsif id.present?
         response = Maremma.get(id)
-        doc = Nokogiri::XML(response.body.fetch("data", nil))
+        doc = Nokogiri::XML(response.body.fetch("data", nil), nil, 'UTF-8')
         @raw = doc.at_xpath('//script[@type="application/ld+json"]')
+        @raw = @raw.text if @raw.present?
       end
     end
 
     def metadata
-      @metadata ||= raw.present? ? Maremma.from_json(raw) : {}
+      @metadata ||= raw.present? && valid? ? Maremma.from_json(raw) : {}
     end
 
     def exists?
       metadata.present?
     end
 
+    def errors
+      jsonlint(raw)
+    end
+
     def valid?
-      schema_org.present?
+      errors.empty?
     end
 
     def doi
@@ -61,7 +66,7 @@ module Bolognese
     def ris_type
       SO_TO_RIS_TRANSLATIONS[resource_type_general.to_s.dasherize] || "GEN"
     end
-    
+
     def additional_type
       metadata.fetch("additionalType", nil)
     end