bolognese 1.3.19 → 1.3.20

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 623b331e65c7846df6428bb0f6fab2bacbd5914d2db60ee2844103d366bb14af
-  data.tar.gz: 3b70980e0300a2ffbc080fd6885d23b4b081d7e15c82350fbb38f8ae49e8bc49
+  metadata.gz: d2386c22e800464e8a91916067a08a205859d4f7a2adbc7ffa4d7c3082ea7b23
+  data.tar.gz: 95ee5f47012025a176d2c4eda3230f2dfc1c847544c8e953c1dfd2e0e074e779
 SHA512:
-  metadata.gz: b7bdaa8dbe6ba2238dc632836d8449cbbe8a96b89d21a94fd0514028b7d11e8f953edd66c640c11d60b2ebeb6eeeb1cb7db191a200334e856f486ba843a36d01
-  data.tar.gz: f682d7c53f065107d29444c9394de6de744475ab538f2a5cf1dab53e26c526bd5e9c6c753f2d0b2e3b66fa1e686dc90a89205068640f6cd7da27bcf08f6290ed
+  metadata.gz: 15c0cc56f0ce1a77d2571f4e015efbc8f4f4aa469636c0071c83d99510eb11e11a45094dd9ea44a5c1db42c0b53038e9231027de9d821818ae171954b1c18152
+  data.tar.gz: 3d93f06ed8eb492aafde1a629be051f5aa0666e23f9629a1f191fef856501c62d321fd61ce53e0f8daacd8bb458733cd3496f02d48b1ce4f44c8ad6a196695ec

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (1.3.19)
+    bolognese (1.3.20)
       activesupport (>= 4.2.5, < 6)
       benchmark_methods (~> 0.7)
       bibtex-ruby (~> 4.1)
@@ -60,14 +60,14 @@ GEM
     ebnf (1.1.3)
       rdf (~> 3.0)
       sxp (~> 1.0)
-    edtf (3.0.4)
-      activesupport (>= 3.0, < 6.0)
-    excon (0.62.0)
-    faraday (0.14.0)
+    edtf (3.0.5)
+      activesupport (>= 3.0, < 7.0)
+    excon (0.67.0)
+    faraday (0.17.0)
       multipart-post (>= 1.2, < 3)
     faraday-encoding (0.0.5)
       faraday
-    faraday_middleware (0.12.2)
+    faraday_middleware (0.13.1)
       faraday (>= 0.7.4, < 1.0)
     gender_detector (0.1.2)
       unicode_utils (>= 1.3.0)
@@ -78,7 +78,7 @@ GEM
       concurrent-ruby (~> 1.0)
     hashdiff (1.0.0)
     htmlentities (4.3.4)
-    i18n (1.6.0)
+    i18n (1.7.0)
       concurrent-ruby (~> 1.0)
     iso8601 (0.9.1)
     json (2.2.0)
@@ -94,23 +94,23 @@ GEM
       trollop (~> 2)
     latex-decode (0.3.1)
     link_header (0.0.8)
-    loofah (2.2.3)
+    loofah (2.3.0)
       crass (~> 1.0.2)
       nokogiri (>= 1.5.9)
-    maremma (4.3)
+    maremma (4.4)
       activesupport (>= 4.2.5, < 6)
       addressable (>= 2.3.6)
       builder (~> 3.2, >= 3.2.2)
-      excon (~> 0.60, < 0.63)
-      faraday (~> 0.14, < 0.15)
+      excon (~> 0.67.0)
+      faraday (~> 0.17.0)
       faraday-encoding (~> 0.0.4)
-      faraday_middleware (~> 0.12.0)
+      faraday_middleware (~> 0.13.1)
       multi_json (~> 1.12)
       nokogiri (~> 1.10.4)
       oj (>= 2.8.3)
     mini_portile2 (2.4.0)
-    minitest (5.11.3)
-    multi_json (1.13.1)
+    minitest (5.12.2)
+    multi_json (1.14.1)
     multipart-post (2.1.1)
     namae (1.0.1)
     nokogiri (1.10.4)
@@ -146,19 +146,19 @@ GEM
       rdf (~> 3.0)
     rdf-xsd (3.0.1)
       rdf (~> 3.0)
-    rspec (3.8.0)
-      rspec-core (~> 3.8.0)
-      rspec-expectations (~> 3.8.0)
-      rspec-mocks (~> 3.8.0)
-    rspec-core (3.8.2)
-      rspec-support (~> 3.8.0)
-    rspec-expectations (3.8.4)
+    rspec (3.9.0)
+      rspec-core (~> 3.9.0)
+      rspec-expectations (~> 3.9.0)
+      rspec-mocks (~> 3.9.0)
+    rspec-core (3.9.0)
+      rspec-support (~> 3.9.0)
+    rspec-expectations (3.9.0)
       diff-lcs (>= 1.2.0, < 2.0)
-      rspec-support (~> 3.8.0)
-    rspec-mocks (3.8.1)
+      rspec-support (~> 3.9.0)
+    rspec-mocks (3.9.0)
       diff-lcs (>= 1.2.0, < 2.0)
-      rspec-support (~> 3.8.0)
-    rspec-support (3.8.2)
+      rspec-support (~> 3.9.0)
+    rspec-support (3.9.0)
     rspec-xsd (0.1.0)
       nokogiri (~> 1.6)
       rspec (~> 3)
@@ -173,13 +173,13 @@ GEM
     temple (0.8.2)
     thor (0.20.3)
     thread_safe (0.3.6)
-    tilt (2.0.9)
+    tilt (2.0.10)
     trollop (2.9.9)
     tzinfo (1.2.5)
       thread_safe (~> 0.1)
     unicode_utils (1.4.0)
     vcr (3.0.3)
-    webmock (3.7.3)
+    webmock (3.7.6)
       addressable (>= 2.3.6)
       crack (>= 0.3.2)
       hashdiff (>= 0.4.0, < 2.0.0)

data/lib/bolognese/author_utils.rb

@@ -13,6 +13,9 @@ module Bolognese
     }
 
     def get_one_author(author)
+      # author is a string
+      author = { "creatorName" => author } if author.is_a?(String)
+      
       # malformed XML
       return nil if author.fetch("creatorName", nil).is_a?(Array)
 

data/lib/bolognese/cli.rb

@@ -29,18 +29,20 @@ module Bolognese
     method_option :regenerate, :type => :boolean, :force => false
     method_option :style, aliases: "-s", default: "apa"
     method_option :locale, aliases: "-l", default: "en-US"
+    method_option :show_errors, :type => :boolean, :force => false
     def convert(input)
       metadata = Metadata.new(input: input,
                               from: options[:from],
                               regenerate: options[:regenerate],
                               style: options[:style],
-                              locale: options[:locale])
+                              locale: options[:locale],
+                              show_errors: options[:show_errors])
       to = options[:to] || "schema_org"
 
-      if metadata.valid?
-        puts metadata.send(to)
-      else
+      if options[:show_errors] && !metadata.valid?
         $stderr.puts metadata.errors
+      else
+        puts metadata.send(to)
       end
     end
 

data/lib/bolognese/metadata.rb

@@ -6,7 +6,7 @@ module Bolognese
   class Metadata
     include Bolognese::MetadataUtils
 
-    attr_accessor :string, :from, :sandbox, :meta, :regenerate, :issue
+    attr_accessor :string, :from, :sandbox, :meta, :regenerate, :issue, :show_errors
     attr_reader :doc, :page_start, :page_end
     attr_writer :id, :provider_id, :client_id, :doi, :identifiers, :creators, :contributors, :titles, :publisher, 
                 :rights_list, :dates, :publication_year, :volume, :url, :version_info,

data/lib/bolognese/metadata_utils.rb

@@ -12,6 +12,7 @@ require_relative 'readers/crosscite_reader'
 require_relative 'readers/crossref_reader'
 require_relative 'readers/datacite_json_reader'
 require_relative 'readers/datacite_reader'
+require_relative 'readers/npm_reader'
 require_relative 'readers/ris_reader'
 require_relative 'readers/schema_org_reader'
 
@@ -45,6 +46,7 @@ module Bolognese
     include Bolognese::Readers::CrossrefReader
     include Bolognese::Readers::DataciteReader
     include Bolognese::Readers::DataciteJsonReader
+    include Bolognese::Readers::NpmReader
     include Bolognese::Readers::RisReader
     include Bolognese::Readers::SchemaOrgReader
 

data/lib/bolognese/readers/npm_reader.rb

@@ -0,0 +1,106 @@
+# frozen_string_literal: true
+
+module Bolognese
+  module Readers
+    module NpmReader
+      def read_npm(string: nil, **options)
+        if string.present?
+          errors = jsonlint(string)
+          return { "errors" => errors } if errors.present?
+        end
+
+        read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:doi, :id, :url, :sandbox, :validate, :ra))
+
+        meta = string.present? ? Maremma.from_json(string) : {}
+
+        types = {
+          "resourceTypeGeneral" => "Software",
+          "reourceType" => "NPM Package",
+          "schemaOrg" => "SoftwareSourceCode",
+          "citeproc" => "article",
+          "bibtex" => "misc",
+          "ris" => "GEN"
+        }.compact
+
+        creators = if meta.fetch("author", nil).present?
+          get_authors(Array.wrap(meta.fetch("author", nil)))
+        else
+          [{ "nameType" => "Organizational", "name" => ":(unav)" }]
+        end
+        # contributors = get_authors(from_citeproc(Array.wrap(meta.fetch("editor", nil))))
+        # dates = if date = get_date_from_date_parts(meta.fetch("issued", nil))
+        #           if Date.edtf(date).present?
+        #             [{ "date" => date,
+        #               "dateType" => "Issued" }]
+        #           end
+        #         end
+        # publication_year = get_date_from_date_parts(meta.fetch("issued", nil)).to_s[0..3]
+        rights_list = if meta.fetch("license", nil)
+                        [{ "rights" => meta.fetch("license") }.compact]
+                      end
+        # related_identifiers = if meta.fetch("container-title", nil).present? && meta.fetch("ISSN", nil).present?
+        #                         [{ "type" => "Periodical",
+        #                           "relationType" => "IsPartOf",
+        #                           "relatedIdentifierType" => "ISSN",
+        #                           "title" => meta.fetch("container-title", nil),
+        #                           "relatedIdentifier" => meta.fetch("ISSN", nil) }.compact]
+        #                       end
+        # container = if meta.fetch("container-title", nil).present?
+        #   first_page = meta.fetch("page", nil).present? ? meta.fetch("page").split("-").map(&:strip)[0] : nil
+        #   last_page = meta.fetch("page", nil).present? ? meta.fetch("page").split("-").map(&:strip)[1] : nil
+
+        #   { "type" => "Periodical",
+        #     "title" => meta.fetch("container-title", nil),
+        #     "identifier" => meta.fetch("ISSN", nil),
+        #     "identifierType" => meta.fetch("ISSN", nil).present? ? "ISSN" : nil,
+        #     "volume" => meta.fetch("volume", nil),
+        #     "issue" => meta.fetch("issue", nil),
+        #     "firstPage" => first_page,
+        #     "lastPage" => last_page
+        #    }.compact
+        # else
+        #   nil
+        # end
+
+        # identifiers = [normalize_id(meta.fetch("id", nil)), normalize_doi(meta.fetch("DOI", nil))].compact.map do |r|
+        #   r = normalize_id(r)
+
+        #   if r.start_with?("https://doi.org")
+        #     { "identifierType" => "DOI", "identifier" => r }
+        #   else
+        #       { "identifierType" => "URL", "identifier" => r }
+        #   end
+        # end.uniq
+
+        # id = Array.wrap(identifiers).first.to_h.fetch("identifier", nil)
+        # doi = Array.wrap(identifiers).find { |r| r["identifierType"] == "DOI" }.to_h.fetch("identifier", nil)
+          
+        # state = id.present? || read_options.present? ? "findable" : "not_found"
+        subjects = Array.wrap(meta.fetch("keywords", nil)).map do |s|
+          { "subject" => s }
+        end
+
+        { 
+          #"id" => id,
+          #"identifiers" => identifiers,
+          "types" => types,
+          #"doi" => doi_from_url(doi),
+          #"url" => normalize_id(meta.fetch("URL", nil)),
+          "titles" => [{ "title" => meta.fetch("name", nil) }],
+          "creators" => creators,
+          #"contributors" => contributors,
+          #"container" => container,
+          #"publisher" => meta.fetch("publisher", nil),
+          #"related_identifiers" => related_identifiers,
+          #"dates" => dates,
+          #"publication_year" => publication_year,
+          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description")), "descriptionType" => "Abstract" }] : [],
+          "rights_list" => rights_list,
+          "version_info" => meta.fetch("version", nil),
+          "subjects" => subjects
+          #"state" => state
+        }.merge(read_options)
+      end
+    end
+  end
+end

data/lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "1.3.19"
+  VERSION = "1.3.20"
 end

data/lib/bolognese/writers/codemeta_writer.rb

@@ -4,7 +4,7 @@ module Bolognese
   module Writers
     module CodemetaWriter
       def codemeta
-        return nil unless valid?
+        return nil unless valid? || show_errors
 
         hsh = {
           "@context" => id.present? ? "https://raw.githubusercontent.com/codemeta/codemeta/master/codemeta.jsonld" : nil,

data/spec/cli_spec.rb

@@ -180,9 +180,15 @@ describe Bolognese::CLI do
 
       it 'to datacite invalid' do
         file = fixture_path + "datacite_missing_creator.xml"
-        subject.options = { to: "datacite" }
+        subject.options = { to: "datacite", show_errors: "true" }
         expect { subject.convert file }.to output("4:0: ERROR: Element '{http://datacite.org/schema/kernel-4}creators': Missing child element(s). Expected is ( {http://datacite.org/schema/kernel-4}creator ).\n").to_stderr
       end
+
+      it 'to datacite invalid ignore errors' do
+        file = fixture_path + "datacite_missing_creator.xml"
+        subject.options = { to: "datacite" }
+        expect { subject.convert file }.to output(/http:\/\/datacite.org\/schema\/kernel-4/).to_stdout
+      end
     end
 
     context "codemeta" do

data/spec/fixtures/cgimp_package.json

@@ -0,0 +1,18 @@
+{
+  "name": "fullstack_app",
+  "version": "1.0.0",
+  "main": "index.js",
+  "scripts": {
+    "start": "concurrently \"sudo -i service elasticsearch start\" \"cd backend && node --max_old_space_size=8000 server.js\" \"cd client && npm start\""
+  },
+  "keywords": [],
+  "author": "",
+  "license": "ISC",
+  "dependencies": {
+    "concurrently": "^4.1.0",
+    "d3": "^5.9.2",
+    "monk": "^6.0.6"
+  },
+  "devDependencies": {},
+  "description": ""
+}

data/spec/fixtures/cit_package.json

@@ -0,0 +1,19 @@
+{
+  "name": "CITapp",
+  "description": "Concealed Information Test app",
+  "version": "1.1.0",
+  "scripts": {
+    "test": "karma start test/karma.conf.js --single-run"
+  },
+  "devDependencies": {
+    "jasmine-core": "*",
+    "karma": "*",
+    "karma-chrome-launcher": "^2.2.0",
+    "karma-cli": "*",
+    "karma-fixture": "^0.2.6",
+    "karma-html2js-preprocessor": "*",
+    "karma-jasmine": "*",
+    "karma-phantomjs-launcher": "*",
+    "phantomjs-prebuilt": "*"
+  }
+}

data/spec/fixtures/edam_package.json

@@ -0,0 +1,12 @@
+{
+  "name": "edam-browser",
+  "description": "The EDAM Browser is a client-side web-based visualization javascript widget. Its goals are to help describing bio-related resources and service with EDAM, and to facilitate and foster community contributions to EDAM.",
+  "author": "Bryan Brancotte",
+  "version": "1.0.0",
+  "scripts": {
+    "test": "jshint js/*js"
+  },
+  "devDependencies": {
+    "jshint": "^2.9.5"
+  }
+}

data/spec/readers/datacite_json_reader_spec.rb

@@ -44,7 +44,7 @@ describe Bolognese::Metadata, vcr: true do
 
     it "SoftwareSourceCode missing_comma" do
       input = fixture_path + "datacite_software_missing_comma.json"
-      subject = Bolognese::Metadata.new(input: input, from: "datacite_json")
+      subject = Bolognese::Metadata.new(input: input, from: "datacite_json", show_errors: true)
       expect(subject.valid?).to be false
       expect(subject.errors).to eq(["expected comma, not a string at line 4, column 11 [parse.c:381]"])
       expect(subject.codemeta).to be_nil
@@ -52,7 +52,7 @@ describe Bolognese::Metadata, vcr: true do
 
     it "SoftwareSourceCode overlapping_keys" do
       input = fixture_path + "datacite_software_overlapping_keys.json"
-      subject = Bolognese::Metadata.new(input: input, from: "datacite_json")
+      subject = Bolognese::Metadata.new(input: input, from: "datacite_json", show_errors: true)
       expect(subject.valid?).to be false
       expect(subject.errors).to eq(["The same key is defined more than once: id"])
       expect(subject.codemeta).to be_nil

data/spec/readers/npm_reader_spec.rb

@@ -0,0 +1,66 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+
+describe Bolognese::Metadata, vcr: true do
+  let(:input) { fixture_path + "cgimp_package.json" }
+
+  subject { Bolognese::Metadata.new(input: input, from: "npm") }
+
+  context "get npm raw" do
+    it "software" do
+      expect(subject.raw).to eq(IO.read(input).strip)
+    end
+  end
+
+  context "get npm metadata" do
+    it "minimal" do
+      expect(subject.valid?).to be false
+      expect(subject.errors.first).to start_with("3:0: ERROR: Element '{http://datacite.org/schema/kernel-4}identifier'")
+      #expect(subject.identifiers).to eq([{"identifier"=>"https://doi.org/10.5438/4k3m-nyvg", "identifierType"=>"DOI"}])
+      #expect(subject.url).to eq("https://blog.datacite.org/eating-your-own-dog-food")
+      expect(subject.types).to eq("bibtex"=>"misc", "citeproc"=>"article", "reourceType"=>"NPM Package", "resourceTypeGeneral"=>"Software", "ris"=>"GEN", "schemaOrg"=>"SoftwareSourceCode")
+      expect(subject.creators).to eq([{"name"=>":(unav)", "nameType"=>"Organizational"}])
+      expect(subject.titles).to eq([{"title"=>"fullstack_app"}])
+      expect(subject.descriptions).to be_empty
+      expect(subject.rights_list).to eq([{"rights"=>"ISC"}])
+      expect(subject.version_info).to eq("1.0.0")
+      #expect(subject.dates).to eq([{"date"=>"2016-12-20", "dateType"=>"Issued"}])
+      #expect(subject.publication_year).to eq("2016")
+    end
+
+    it "minimal with description" do
+      input = fixture_path + "cit_package.json"
+      subject = Bolognese::Metadata.new(input: input, from: "npm") 
+      expect(subject.valid?).to be false
+      expect(subject.errors.first).to start_with("3:0: ERROR: Element '{http://datacite.org/schema/kernel-4}identifier'")
+      #expect(subject.identifiers).to eq([{"identifier"=>"https://doi.org/10.5438/4k3m-nyvg", "identifierType"=>"DOI"}])
+      #expect(subject.url).to eq("https://blog.datacite.org/eating-your-own-dog-food")
+      expect(subject.types).to eq("bibtex"=>"misc", "citeproc"=>"article", "reourceType"=>"NPM Package", "resourceTypeGeneral"=>"Software", "ris"=>"GEN", "schemaOrg"=>"SoftwareSourceCode")
+      expect(subject.creators).to eq([{"name"=>":(unav)", "nameType"=>"Organizational"}])
+      expect(subject.titles).to eq([{"title"=>"CITapp"}])
+      expect(subject.descriptions).to eq([{"description"=>"Concealed Information Test app", "descriptionType"=>"Abstract"}])
+      expect(subject.version_info).to eq("1.1.0")
+      #expect(subject.dates).to eq([{"date"=>"2016-12-20", "dateType"=>"Issued"}])
+      #expect(subject.publication_year).to eq("2016")
+    end
+
+    it "minimal with description" do
+      input = fixture_path + "edam_package.json"
+      subject = Bolognese::Metadata.new(input: input, from: "npm") 
+      expect(subject.valid?).to be false
+      expect(subject.errors.first).to start_with("3:0: ERROR: Element '{http://datacite.org/schema/kernel-4}identifier'")
+      #expect(subject.identifiers).to eq([{"identifier"=>"https://doi.org/10.5438/4k3m-nyvg", "identifierType"=>"DOI"}])
+      #expect(subject.url).to eq("https://blog.datacite.org/eating-your-own-dog-food")
+      expect(subject.types).to eq("bibtex"=>"misc", "citeproc"=>"article", "reourceType"=>"NPM Package", "resourceTypeGeneral"=>"Software", "ris"=>"GEN", "schemaOrg"=>"SoftwareSourceCode")
+      expect(subject.creators).to eq([{"affiliation"=>[], "familyName"=>"Brancotte", "givenName"=>"Bryan", "name"=>"Brancotte, Bryan", "nameIdentifiers"=>[], "nameType"=>"Personal"}])
+      expect(subject.titles).to eq([{"title"=>"edam-browser"}])
+      expect(subject.descriptions).to eq([{"description"=>
+        +   "The EDAM Browser is a client-side web-based visualization javascript widget. Its goals are to help describing bio-related resources and service with EDAM, and to facilitate and foster community contributions to EDAM.",
+           "descriptionType"=>"Abstract"}])
+      expect(subject.version_info).to eq("1.0.0")
+      #expect(subject.dates).to eq([{"date"=>"2016-12-20", "dateType"=>"Issued"}])
+      #expect(subject.publication_year).to eq("2016")
+    end
+  end
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: bolognese
 version: !ruby/object:Gem::Version
-  version: 1.3.19
+  version: 1.3.20
 platform: ruby
 authors:
 - Martin Fenner
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-09-14 00:00:00.000000000 Z
+date: 2019-10-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: maremma
@@ -571,6 +571,7 @@ files:
 - lib/bolognese/readers/crossref_reader.rb
 - lib/bolognese/readers/datacite_json_reader.rb
 - lib/bolognese/readers/datacite_reader.rb
+- lib/bolognese/readers/npm_reader.rb
 - lib/bolognese/readers/ris_reader.rb
 - lib/bolognese/readers/schema_org_reader.rb
 - lib/bolognese/string.rb
@@ -714,6 +715,8 @@ files:
 - spec/doi_utils_spec.rb
 - spec/find_from_format_spec.rb
 - spec/fixtures/aida.json
+- spec/fixtures/cgimp_package.json
+- spec/fixtures/cit_package.json
 - spec/fixtures/citeproc-no-author.json
 - spec/fixtures/citeproc-no-categories.json
 - spec/fixtures/citeproc.json
@@ -748,6 +751,7 @@ files:
 - spec/fixtures/datacite_software.json
 - spec/fixtures/datacite_software_missing_comma.json
 - spec/fixtures/datacite_software_overlapping_keys.json
+- spec/fixtures/edam_package.json
 - spec/fixtures/funding_reference.xml
 - spec/fixtures/gtex.xml
 - spec/fixtures/maremma/codemeta.json
@@ -1131,6 +1135,7 @@ files:
 - spec/readers/crossref_reader_spec.rb
 - spec/readers/datacite_json_reader_spec.rb
 - spec/readers/datacite_reader_spec.rb
+- spec/readers/npm_reader_spec.rb
 - spec/readers/ris_reader_spec.rb
 - spec/readers/schema_org_reader_spec.rb
 - spec/spec_helper.rb