bolognese 0.5.3 → 0.6.1

data/bolognese.gemspec

@@ -19,7 +19,7 @@ Gem::Specification.new do |s|
   s.add_dependency 'nokogiri', '~> 1.6', '>= 1.6.8'
   s.add_dependency 'builder', '~> 3.2', '>= 3.2.2'
   s.add_dependency 'activesupport', '~> 4.2', '>= 4.2.5'
-  s.add_dependency 'dotenv', '~> 2.1', '>= 2.1.1'
+  s.add_dependency 'bibtex-ruby', '~> 4.1'
   s.add_dependency 'thor', '~> 0.19'
   s.add_dependency 'namae', '~> 0.10.2'
   s.add_dependency 'postrank-uri', '~> 1.0', '>= 1.0.18'

data/lib/bolognese.rb

@@ -2,12 +2,14 @@ require 'active_support/all'
 require 'nokogiri'
 require 'maremma'
 require 'postrank-uri'
+require 'bibtex'
 
 require "bolognese/version"
 require "bolognese/metadata"
 require "bolognese/crossref"
 require "bolognese/datacite"
 require "bolognese/schema_org"
+require "bolognese/bibtex"
 require "bolognese/orcid"
 require "bolognese/cli"
 require "bolognese/string"

data/lib/bolognese/bibtex.rb

@@ -0,0 +1,101 @@
+module Bolognese
+  class Bibtex < Metadata
+
+    BIB_TO_SO_TRANSLATIONS = {
+      "article" => "ScholarlyArticle"
+    }
+
+    SO_TO_BIB_TRANSLATIONS = {
+      "Article" => "article",
+      "AudioObject" => "misc",
+      "Blog" => "misc",
+      "BlogPosting" => "article",
+      "Collection" => "misc",
+      "CreativeWork" => "misc",
+      "DataCatalog" => "misc",
+      "Dataset" => "misc",
+      "Event" => "misc",
+      "ImageObject" => "misc",
+      "Movie" => "misc",
+      "PublicationIssue" => "misc",
+      "ScholarlyArticle" => "article",
+      "Service" => "misc",
+      "SoftwareSourceCode" => "misc",
+      "VideoObject" => "misc",
+      "WebPage" => "misc",
+      "WebSite" => "misc"
+    }
+
+    def initialize(string: nil)
+      @raw = string
+    end
+
+    alias_method :schema_org, :as_schema_org
+
+    def metadata
+      @metadata ||= raw.present? ? BibTeX.parse(raw).first : {}
+    end
+
+    def exists?
+      metadata.present?
+    end
+
+    def type
+      BIB_TO_SO_TRANSLATIONS[metadata.type.to_s] || "ScholarlyArticle"
+    end
+
+    def resource_type_general
+      SO_TO_DC_TRANSLATIONS[type]
+    end
+
+    def doi
+      metadata.doi
+    end
+
+    def url
+      metadata.url
+    end
+
+    def id
+      normalize_doi(doi)
+    end
+
+    def author
+      Array(metadata.author).map do |a|
+        { "@type" => "Person",
+          "givenName" => a.first,
+          "familyName" => a.last }.compact
+      end
+    end
+
+    def name
+      metadata.title
+    end
+
+    def container_title
+      metadata.journal.to_s.presence || metadata.pubisher.to_s
+    end
+
+    def date_published
+      metadata.date.to_s.presence
+    end
+
+    def publication_year
+      metadata.year.to_s.presence
+    end
+
+    def is_part_of
+      if metadata.journal.present?
+        { "@type" => "Periodical",
+          "name" => metadata.journal.to_s,
+          "issn" => metadata.issn.to_s.presence }.compact
+      else
+        nil
+      end
+    end
+
+    def description
+      metadata.note.to_s.presence
+    end
+  end
+end

data/lib/bolognese/cli.rb

@@ -10,6 +10,8 @@ module Bolognese
     include Bolognese::DoiUtils
     include Bolognese::Utils
 
+    default_task :open
+
     def self.exit_on_failure?
       true
     end
@@ -42,5 +44,32 @@ module Bolognese
         puts "not implemented"
       end
     end
+
+    desc "open file", "read metadata from file"
+    method_option :as, default: "schema_org"
+    method_option :schema_version
+    def open(file)
+      ext = File.extname(file)
+      unless %w(.bib).include? ext
+        $stderr.puts "File type #{ext} not supported"
+        exit 1
+      end
+      string = IO.read(file)
+      provider = "bibtex"
+      output = options[:as] || "schema_org"
+
+      if provider.present?
+        p = case provider
+            when "crossref" then Crossref.new(id: id)
+            when "datacite" then Datacite.new(id: id, schema_version: options[:schema_version])
+            when "bibtex" then Bibtex.new(string: string)
+            else SchemaOrg.new(id: id)
+            end
+
+        puts p.send(output)
+      else
+        puts "not implemented"
+      end
+    end
   end
 end

data/lib/bolognese/crossref.rb

@@ -31,6 +31,35 @@ module Bolognese
       "PostedContent" => "ScholarlyArticle"
     }
 
+    CR_TO_BIB_TRANSLATIONS = {
+      "Proceedings" => "proceedings",
+      "ReferenceBook" => "book",
+      "JournalIssue" => nil,
+      "ProceedingsArticle" => nil,
+      "Other" => nil,
+      "Dissertation" => "phdthesis",
+      "Dataset" => nil,
+      "EditedBook" => "book",
+      "JournalArticle" => "article",
+      "Journal" => nil,
+      "Report" => nil,
+      "BookSeries" => nil,
+      "ReportSeries" => nil,
+      "BookTrack" => nil,
+      "Standard" => nil,
+      "BookSection" => "inbook",
+      "BookPart" => nil,
+      "Book" => "book",
+      "BookChapter" => "inbook",
+      "StandardSeries" => nil,
+      "Monograph" => "book",
+      "Component" => nil,
+      "ReferenceEntry" => nil,
+      "JournalVolume" => nil,
+      "BookSet" => nil,
+      "PostedContent" => "article"
+    }
+
     def initialize(id: nil, string: nil)
       id = normalize_doi(id) if id.present?
 
@@ -45,6 +74,7 @@ module Bolognese
     alias_method :crossref, :raw
     alias_method :as_crossref, :raw
     alias_method :schema_org, :as_schema_org
+    alias_method :bibtex, :as_bibtex
 
     def metadata
       @metadata ||= raw.present? ? Maremma.from_xml(raw).fetch("doi_records", {}).fetch("doi_record", {}) : {}
@@ -100,6 +130,10 @@ module Bolognese
       end
     end
 
+    def bibtex_type
+      CR_TO_BIB_TRANSLATIONS[additional_type] || "misc"
+    end
+
     def name
       parse_attribute(bibliographic_metadata.dig("titles", "title"))
     end
@@ -190,9 +224,11 @@ module Bolognese
     end
 
     def container_title
-      is_part_of.fetch("name", nil)
+      is_part_of.to_h.fetch("name", nil)
     end
 
+    alias_method :journal, :container_title
+
     def citation
        citations = bibliographic_metadata.dig("citation_list", "citation")
        Array.wrap(citations).map do |c|

data/lib/bolognese/datacite.rb

@@ -32,6 +32,7 @@ module Bolognese
     end
 
     alias_method :schema_org, :as_schema_org
+    alias_method :bibtex, :as_bibtex
 
     def schema_version
       @schema_version ||= metadata.fetch("xsi:schemaLocation", "").split(" ").first
@@ -76,6 +77,10 @@ module Bolognese
       metadata.fetch("resourceType", {}).fetch("resourceTypeGeneral", nil)
     end
 
+    def bibtex_type
+      Bolognese::Bibtex::SO_TO_BIB_TRANSLATIONS[type] || "misc"
+    end
+
     def name
       metadata.dig("titles", "title")
     end
@@ -157,13 +162,17 @@ module Bolognese
     end
 
     def date_published
-      date("Issued") || metadata.fetch("publicationYear")
+      date("Issued") || publication_year
     end
 
     def date_modified
       date("Updated")
     end
 
+    def publication_year
+      metadata.fetch("publicationYear")
+    end
+
     def language
       metadata.fetch("language", nil)
     end

data/lib/bolognese/datacite_utils.rb

@@ -123,7 +123,7 @@ module Bolognese
     end
 
     def insert_publication_year(xml)
-      xml.publicationYear(date_published && date_published[0..3])
+      xml.publicationYear(publication_year)
     end
 
     def resource_type

data/lib/bolognese/metadata.rb

@@ -12,32 +12,23 @@ module Bolognese
     include Bolognese::DataciteUtils
     include Bolognese::Utils
 
-    attr_reader :id, :raw, :provider, :schema_version
+    attr_reader :id, :raw, :provider, :schema_version, :license, :citation,
+      :additional_type, :alternate_name, :url, :version, :keywords, :editor,
+      :page_start, :page_end, :date_modified, :language, :spatial_coverage,
+      :content_size, :funder, :journal, :bibtex_type
 
     alias_method :datacite, :as_datacite
 
-    def url
-
-    end
-
-    def version
-
-    end
-
-    def keywords
-
+    def publication_year
+      date_published && date_published[0..3]
     end
 
     def date_created
 
     end
 
-    def page_start
-
-    end
-
-    def page_end
-
+    def pagination
+      [page_start, page_end].compact.join("-").presence
     end
 
     def has_part
@@ -52,28 +43,21 @@ module Bolognese
 
     end
 
-    def language
-
-    end
-
-    def spatial_coverage
-
-    end
-
-    def content_size
-
-    end
-
     def schema_version
 
     end
 
-    def funder
+    def author_string
+      author.map { |a| [a["familyName"], a["givenName"]].join(", ") }
+            .join(" and ").presence
+    end
 
+    def publisher_string
+      publisher.to_h.fetch("name", nil)
     end
 
     def as_schema_org
-      { "@context" => "http://schema.org",
+      { "@context" => id.present? ? "http://schema.org" : nil,
         "@type" => type,
         "@id" => id,
         "url" => url,
@@ -103,5 +87,23 @@ module Bolognese
         "provider" => provider
       }.compact.to_json
     end
+
+    def as_bibtex
+      bib = {
+        bibtex_type: bibtex_type.to_sym,
+        bibtex_key: id,
+        doi: doi,
+        url: url,
+        author: author_string,
+        keywords: keywords,
+        title: name,
+        journal: journal,
+        pages: pagination,
+        publisher: publisher_string,
+        year: publication_year
+      }.compact
+
+      BibTeX::Entry.new(bib).to_s
+    end
   end
 end

data/lib/bolognese/schema_org.rb

@@ -14,6 +14,7 @@ module Bolognese
     end
 
     alias_method :schema_org, :as_schema_org
+    alias_method :bibtex, :as_bibtex
 
     def metadata
       @metadata ||= raw.present? ? Maremma.from_json(raw) : {}
@@ -47,6 +48,10 @@ module Bolognese
       metadata.fetch("additionalType", nil)
     end
 
+    def bibtex_type
+      Bolognese::Bibtex::SO_TO_BIB_TRANSLATIONS[type] || "misc"
+    end
+
     def name
       metadata.fetch("name", nil)
     end

data/lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "0.5.3"
+  VERSION = "0.6.1"
 end

data/spec/bibtex_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Bolognese::Bibtex, vcr: true do
+  let(:string) { IO.read(fixture_path + "crossref.bib") }
+
+  subject { Bolognese::Bibtex.new(string: string) }
+
+  context "get metadata" do
+    it "Crossref DOI" do
+      expect(subject.id).to eq("https://doi.org/10.7554/elife.01567")
+      expect(subject.type).to eq("ScholarlyArticle")
+      expect(subject.url).to eq("http://elifesciences.org/lookup/doi/10.7554/eLife.01567")
+      expect(subject.resource_type_general).to eq("Text")
+      expect(subject.author).to eq([{"@type"=>"Person", "givenName"=>"Martial", "familyName"=>"Sankar"},
+                                    {"@type"=>"Person", "givenName"=>"Kaisa", "familyName"=>"Nieminen"},
+                                    {"@type"=>"Person", "givenName"=>"Laura", "familyName"=>"Ragni"},
+                                    {"@type"=>"Person", "givenName"=>"Ioannis", "familyName"=>"Xenarios"},
+                                    {"@type"=>"Person", "givenName"=>"Christian S", "familyName"=>"Hardtke"}])
+      expect(subject.name).to eq("Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth")
+      expect(subject.date_published).to eq("2014")
+      expect(subject.is_part_of).to eq("@type"=>"Periodical", "name"=>"eLife", "issn"=>"2050-084X")
+    end
+  end
+
+  context "get metadata as datacite xml" do
+    it "Crossref DOI" do
+      expect(subject.validation_errors).to be_empty
+      datacite = Maremma.from_xml(subject.as_datacite).fetch("resource", {})
+      expect(datacite.dig("resourceType", "resourceTypeGeneral")).to eq("Text")
+      expect(datacite.dig("titles", "title")).to eq("Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth")
+      expect(datacite.dig("creators", "creator").count).to eq(5)
+      expect(datacite.dig("creators", "creator").first).to eq("creatorName"=>"Sankar, Martial", "givenName"=>"Martial", "familyName"=>"Sankar")
+    end
+  end
+end

data/spec/cli_spec.rb

@@ -28,6 +28,11 @@ describe Bolognese::CLI do
         subject.options = { as: "datacite" }
         expect { subject.read id }.to output(/http:\/\/datacite.org\/schema\/kernel-4/).to_stdout
       end
+
+      it 'as bibtex' do
+        subject.options = { as: "bibtex" }
+        expect { subject.read id }.to output(/@article{https:\/\/doi.org\/10.7554\/elife.01567/).to_stdout
+      end
     end
 
     context "datacite" do
@@ -42,6 +47,11 @@ describe Bolognese::CLI do
         expect { subject.read id }.to output(/Phylogeny, Malaria, Parasites, Taxonomy, Mitochondrial genome, Africa, Plasmodium/).to_stdout
       end
 
+      it 'as bibtex' do
+        subject.options = { as: "bibtex" }
+        expect { subject.read id }.to output(/@misc{https:\/\/doi.org\/10.5061\/dryad.8515/).to_stdout
+      end
+
       it 'as datacite' do
         subject.options = { as: "datacite" }
         expect { subject.read id }.to output(/http:\/\/datacite.org\/schema\/kernel-3/).to_stdout
@@ -64,6 +74,39 @@ describe Bolognese::CLI do
         subject.options = { as: "datacite" }
         expect { subject.read id }.to output(/http:\/\/datacite.org\/schema\/kernel-4/).to_stdout
       end
+
+      it 'as bibtex' do
+        subject.options = { as: "bibtex" }
+        expect { subject.read id }.to output(/@article{https:\/\/doi.org\/10.5438\/4k3m-nyvg/).to_stdout
+      end
     end
   end
+
+  describe "open" do
+    context "bibtex" do
+      let(:file) { fixture_path + "crossref.bib" }
+
+      it 'default' do
+        expect { subject.open file }.to output(/datePublished/).to_stdout
+      end
+
+      it 'as schema_org' do
+        subject.options = { as: "schema_org" }
+        expect { subject.open file }.to output(/datePublished/).to_stdout
+      end
+
+      it 'as datacite' do
+        subject.options = { as: "datacite" }
+        expect { subject.open file }.to output(/http:\/\/datacite.org\/schema\/kernel-4/).to_stdout
+      end
+    end
+
+    # context "unsupported format" do
+    #   let(:file) { fixture_path + "crossref.xxx" }
+    #
+    #   it 'error' do
+    #     expect { subject.open file }.to output(/datePublished/).to_stderr
+    #   end
+    # end
+  end
 end