bolognese 0.3 → 0.4.1

data/lib/bolognese/utils.rb

@@ -1,12 +1,32 @@
 module Bolognese
   module Utils
+    def normalize_orcid(orcid)
+      orcid = validate_orcid(orcid)
+      return nil unless orcid.present?
+
+      # turn ORCID ID into URL
+      "http://orcid.org/" + Addressable::URI.encode(orcid)
+    end
+
+    def orcid_from_url(url)
+      Array(/\Ahttp:\/\/orcid\.org\/(.+)/.match(url)).last
+    end
+
+    def orcid_as_url(orcid)
+      "http://orcid.org/#{orcid}" if orcid.present?
+    end
+
+    def validate_orcid(orcid)
+      Array(/\A(?:http:\/\/orcid\.org\/)?(\d{4}-\d{4}-\d{4}-\d{3}[0-9X]+)\z/.match(orcid)).last
+    end
+
     def parse_attributes(element)
       if element.is_a?(String)
         element
       elsif element.is_a?(Hash)
         element.fetch("text", nil)
       elsif element.is_a?(Array)
-        element.map { |e| e.fetch("text", nil) }
+        element.map { |e| e.fetch("text", nil) }
       else
         nil
       end
@@ -23,5 +43,15 @@ module Bolognese
         nil
       end
     end
+
+    def normalize_url(url)
+      return nil unless url.present?
+
+      normalize_doi(url) || PostRank::URI.clean(url)
+    end
+
+    def normalize_ids(list)
+      Array.wrap(list).map { |url| url.merge("@id" => normalize_url(url["@id"])) }
+    end
   end
 end

data/lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "0.3"
+  VERSION = "0.4.1"
 end

data/spec/crossref_spec.rb

@@ -3,12 +3,31 @@ require 'spec_helper'
 describe Bolognese::Crossref, vcr: true do
   let(:id) { "10.7554/eLife.01567" }
 
-  subject { Bolognese::Crossref.new(id) }
+  subject { Bolognese::Crossref.new(id: id) }
 
   context "get metadata" do
+    it "DOI with data citation" do
+      expect(subject.id).to eq("https://doi.org/10.7554/elife.01567")
+      expect(subject.type).to eq("ScholarlyArticle")
+      expect(subject.additional_type).to eq("JournalArticle")
+      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.license).to eq("http://creativecommons.org/licenses/by/3.0/")
+      expect(subject.name).to eq("Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth")
+      expect(subject.date_published).to eq("2014-02-11")
+      expect(subject.date_modified).to eq("2015-08-11T05:35:02Z")
+      expect(subject.is_part_of).to eq("@type"=>"Periodical", "name"=>"eLife", "issn"=>"2050-084X")
+      expect(subject.citation.count).to eq(27)
+      expect(subject.citation[21]).to eq("@type"=>"CreativeWork", "@id"=>"https://doi.org/10.5061/dryad.b835k", "position"=>"22", "datePublished"=>"2014")
+      expect(subject.provider).to eq("@type"=>"Organization", "name"=>"Crossref")
+    end
+
     it "journal article" do
       id = "https://doi.org/10.1371/journal.pone.0000030"
-      subject = Bolognese::Crossref.new(id)
+      subject = Bolognese::Crossref.new(id: id)
       expect(subject.id).to eq(id)
       expect(subject.type).to eq("ScholarlyArticle")
       expect(subject.additional_type).to eq("JournalArticle")
@@ -29,7 +48,7 @@ describe Bolognese::Crossref, vcr: true do
 
     it "posted_content" do
       id = "https://doi.org/10.1101/097196"
-      subject = Bolognese::Crossref.new(id)
+      subject = Bolognese::Crossref.new(id: id)
       expect(subject.id).to eq(id)
       expect(subject.type).to eq("CreativeWork")
       expect(subject.additional_type).to eq("PostedContent")
@@ -44,28 +63,9 @@ describe Bolognese::Crossref, vcr: true do
       expect(subject.provider).to eq("@type"=>"Organization", "name"=>"Crossref")
     end
 
-    it "DOI with data citation" do
-      expect(subject.id).to eq("https://doi.org/10.7554/elife.01567")
-      expect(subject.type).to eq("ScholarlyArticle")
-      expect(subject.additional_type).to eq("JournalArticle")
-      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.license).to eq("http://creativecommons.org/licenses/by/3.0/")
-      expect(subject.name).to eq("Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth")
-      expect(subject.date_published).to eq("2014-02-11")
-      expect(subject.date_modified).to eq("2015-08-11T05:35:02Z")
-      expect(subject.is_part_of).to eq("@type"=>"Periodical", "name"=>"eLife", "issn"=>"2050-084X")
-      expect(subject.citation.count).to eq(27)
-      expect(subject.citation[21]).to eq("@type"=>"CreativeWork", "@id"=>"https://doi.org/10.5061/dryad.b835k", "position"=>"22", "datePublished"=>"2014")
-      expect(subject.provider).to eq("@type"=>"Organization", "name"=>"Crossref")
-    end
-
     it "DOI with SICI DOI" do
       id = "https://doi.org/10.1890/0012-9658(2006)87[2832:tiopma]2.0.co;2"
-      subject = Bolognese::Crossref.new(id)
+      subject = Bolognese::Crossref.new(id: id)
       expect(subject.id).to eq("https://doi.org/10.1890/0012-9658(2006)87%5B2832:tiopma%5D2.0.co;2")
       expect(subject.type).to eq("ScholarlyArticle")
       expect(subject.additional_type).to eq("JournalArticle")
@@ -80,9 +80,35 @@ describe Bolognese::Crossref, vcr: true do
       expect(subject.provider).to eq("@type"=>"Organization", "name"=>"Crossref")
     end
 
+    it "DOI with ORCID ID" do
+      id = "https://doi.org/10.1155/2012/291294"
+      subject = Bolognese::Crossref.new(id: id)
+      expect(subject.id).to eq("https://doi.org/10.1155/2012/291294")
+      expect(subject.type).to eq("ScholarlyArticle")
+      expect(subject.additional_type).to eq("JournalArticle")
+      expect(subject.author).to eq([{"@type"=>"Person", "givenName"=>"Wendy", "familyName"=>"Thanassi"},
+                                    {"@type"=>"Person", "givenName"=>"Art", "familyName"=>"Noda"},
+                                    {"@type"=>"Person",
+                                     "@id"=>"http://orcid.org/0000-0003-2043-4925",
+                                     "givenName"=>"Beatriz",
+                                     "familyName"=>"Hernandez"},
+                                    {"@type"=>"Person", "givenName"=>"Jeffery", "familyName"=>"Newell"},
+                                    {"@type"=>"Person", "givenName"=>"Paul", "familyName"=>"Terpeluk"},
+                                    {"@type"=>"Person", "givenName"=>"David", "familyName"=>"Marder"},
+                                    {"@type"=>"Person", "givenName"=>"Jerome A.", "familyName"=>"Yesavage"}])
+      expect(subject.license).to eq("http://creativecommons.org/licenses/by/3.0/")
+      expect(subject.name).to eq("Delineating a Retesting Zone Using Receiver Operating Characteristic Analysis on Serial QuantiFERON Tuberculosis Test Results in US Healthcare Workers")
+      expect(subject.date_published).to eq("2012")
+      expect(subject.date_modified).to eq("2016-08-02T12:42:41Z")
+      expect(subject.page_start).to eq("1")
+      expect(subject.page_end).to eq("7")
+      expect(subject.is_part_of).to eq("@type"=>"Periodical", "name"=>"Pulmonary Medicine", "issn"=>"2090-1836")
+      expect(subject.provider).to eq("@type"=>"Organization", "name"=>"Crossref")
+    end
+
     it "date in future" do
       id = "https://doi.org/10.1016/j.ejphar.2015.03.018"
-      subject = Bolognese::Crossref.new(id)
+      subject = Bolognese::Crossref.new(id: id)
       expect(subject.id).to eq(id)
       expect(subject.type).to eq("ScholarlyArticle")
       expect(subject.additional_type).to eq("JournalArticle")
@@ -105,17 +131,51 @@ describe Bolognese::Crossref, vcr: true do
 
     it "not found error" do
       id = "https://doi.org/10.7554/elife.01567x"
-      subject = Bolognese::Crossref.new(id)
-      expect(subject.id).to eq(id)
+      subject = Bolognese::Crossref.new(id: id)
+      expect(subject.id).to be_nil
       expect(subject.exists?).to be false
     end
   end
 
+  context "get metadata as string" do
+    it "DOI with data citation" do
+      id = "10.7554/eLife.01567"
+      string = Bolognese::Crossref.new(id: id).raw
+
+      subject = Bolognese::Crossref.new(string: string)
+      expect(subject.id).to eq("https://doi.org/10.7554/elife.01567")
+      expect(subject.type).to eq("ScholarlyArticle")
+      expect(subject.additional_type).to eq("JournalArticle")
+      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.license).to eq("http://creativecommons.org/licenses/by/3.0/")
+      expect(subject.name).to eq("Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth")
+      expect(subject.date_published).to eq("2014-02-11")
+      expect(subject.date_modified).to eq("2015-08-11T05:35:02Z")
+      expect(subject.is_part_of).to eq("@type"=>"Periodical", "name"=>"eLife", "issn"=>"2050-084X")
+      expect(subject.citation.count).to eq(27)
+      expect(subject.citation[21]).to eq("@type"=>"CreativeWork", "@id"=>"https://doi.org/10.5061/dryad.b835k", "position"=>"22", "datePublished"=>"2014")
+      expect(subject.provider).to eq("@type"=>"Organization", "name"=>"Crossref")
+    end
+  end
+
   context "get metadata as datacite xml" do
     it "DOI with data citation" do
       expect(subject.validation_errors).to be_empty
       datacite = Maremma.from_xml(subject.as_datacite).fetch("resource", {})
       expect(datacite.dig("titles", "title")).to eq("Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth")
     end
+
+    it "DOI with ORCID ID" do
+      id = "https://doi.org/10.1155/2012/291294"
+      subject = Bolognese::Crossref.new(id: id)
+      expect(subject.validation_errors).to be_empty
+      datacite = Maremma.from_xml(subject.as_datacite).fetch("resource", {})
+      expect(datacite.dig("creators", "creator").count).to eq(7)
+      expect(datacite.dig("creators", "creator").first).to eq("creatorName"=>"Thanassi, Wendy", "givenName"=>"Wendy", "familyName"=>"Thanassi")
+    end
   end
 end

data/spec/datacite_spec.rb

@@ -4,7 +4,7 @@ describe Bolognese::Datacite, vcr: true do
   context "get metadata" do
     let(:id) { "https://doi.org/10.5061/DRYAD.8515" }
 
-    subject { Bolognese::Datacite.new(id) }
+    subject { Bolognese::Datacite.new(id: id) }
 
     it "Dataset" do
       expect(subject.id).to eq("https://doi.org/10.5061/dryad.8515")
@@ -30,7 +30,7 @@ describe Bolognese::Datacite, vcr: true do
 
     it "BlogPosting" do
       id = "https://doi.org/10.5438/4K3M-NYVG"
-      subject = Bolognese::Datacite.new(id)
+      subject = Bolognese::Datacite.new(id: id)
       expect(subject.id).to eq("https://doi.org/10.5438/4k3m-nyvg")
       expect(subject.type).to eq("ScholarlyArticle")
       expect(subject.additional_type).to eq("BlogPosting")
@@ -45,5 +45,46 @@ describe Bolognese::Datacite, vcr: true do
                                       {"@type"=>"CreativeWork", "@id"=>"https://doi.org/10.5438/55e5-t5c0"}])
       expect(subject.provider).to eq("@type"=>"Organization", "name"=>"DataCite")
     end
+
+    it "Date" do
+      id = "https://doi.org/10.4230/lipics.tqc.2013.93"
+      subject = Bolognese::Datacite.new(id: id)
+      expect(subject.id).to eq("https://doi.org/10.4230/lipics.tqc.2013.93")
+      expect(subject.type).to eq("ScholarlyArticle")
+      expect(subject.additional_type).to eq("ConferencePaper")
+      expect(subject.author).to eq([{"@type"=>"Person", "givenName"=>"Nathaniel", "familyName"=>"Johnston"}])
+      expect(subject.name).to eq("The Minimum Size of Qubit Unextendible Product Bases")
+      expect(subject.description).to start_with("We investigate the problem of constructing unextendible product bases in the qubit case")
+      expect(subject.date_published).to eq("2013")
+      expect(subject.provider).to eq("@type"=>"Organization", "name"=>"DataCite")
+    end
+  end
+
+  context "get metadata as string" do
+    it "Dataset" do
+      id = "https://doi.org/10.5061/DRYAD.8515"
+      string = Bolognese::Datacite.new(id: id).raw
+
+      subject = Bolognese::Datacite.new(string: string)
+      expect(subject.id).to eq("https://doi.org/10.5061/dryad.8515")
+      expect(subject.type).to eq("Dataset")
+      expect(subject.additional_type).to eq("DataPackage")
+      expect(subject.author).to eq([{"@type"=>"Person", "givenName"=>"Benjamin", "familyName"=>"Ollomo"},
+                                    {"@type"=>"Person", "givenName"=>"Patrick", "familyName"=>"Durand"},
+                                    {"@type"=>"Person", "givenName"=>"Franck", "familyName"=>"Prugnolle"},
+                                    {"@type"=>"Person", "givenName"=>"Emmanuel J. P.", "familyName"=>"Douzery"},
+                                    {"@type"=>"Person", "givenName"=>"Céline", "familyName"=>"Arnathau"},
+                                    {"@type"=>"Person", "givenName"=>"Dieudonné", "familyName"=>"Nkoghe"},
+                                    {"@type"=>"Person", "givenName"=>"Eric", "familyName"=>"Leroy"},
+                                    {"@type"=>"Person", "givenName"=>"François", "familyName"=>"Renaud"}])
+      expect(subject.name).to eq("Data from: A new malaria agent in African hominids.")
+      expect(subject.alternate_name).to eq("Ollomo B, Durand P, Prugnolle F, Douzery EJP, Arnathau C, Nkoghe D, Leroy E, Renaud F (2009) A new malaria agent in African hominids. PLoS Pathogens 5(5): e1000446.")
+      expect(subject.license).to eq("http://creativecommons.org/publicdomain/zero/1.0/")
+      expect(subject.date_published).to eq("2011")
+      expect(subject.has_part).to eq([{"@type"=>"CreativeWork", "@id"=>"https://doi.org/10.5061/dryad.8515/1"},
+                                      {"@type"=>"CreativeWork", "@id"=>"https://doi.org/10.5061/dryad.8515/2"}])
+      expect(subject.citation).to eq([{"@type"=>"CreativeWork", "@id"=>"https://doi.org/10.1371/journal.ppat.1000446"}])
+      expect(subject.provider).to eq("@type"=>"Organization", "name"=>"DataCite")
+    end
   end
 end

data/spec/{doi_spec.rb → doi_utils_spec.rb}

@@ -3,7 +3,7 @@ require 'spec_helper'
 describe Bolognese::Crossref, vcr: true do
   let(:id) { "https://doi.org/10.1371/journal.pone.0000030" }
 
-  subject { Bolognese::Crossref.new(id) }
+  subject { Bolognese::Crossref.new(id: id) }
 
   context "normalize doi" do
     it "doi" do
@@ -83,7 +83,7 @@ describe Bolognese::Crossref, vcr: true do
     it "not found" do
       doi = "https://doi.org/10.5061/dryad.8515x"
       response = subject.get_doi_ra(doi)
-      expect(response["errors"]).to eq([{"DOI"=>"10.5061/DRYAD.8515X", "status"=>"Invalid DOI"}])
+      expect(response["errors"]).to eq([{"DOI"=>"10.5061/dryad.8515x", "status"=>"Invalid DOI"}])
     end
   end
 end

data/spec/fixtures/vcr_cassettes/Bolognese_CLI/read/crossref/as_crossref.yml

@@ -2,7 +2,7 @@
 http_interactions:
 - request:
     method: get
-    uri: https://doi.crossref.org/doiRA/10.7554/ELIFE.01567
+    uri: https://doi.crossref.org/doiRA/10.7554/elife.01567
     body:
       encoding: US-ASCII
       string: ''
@@ -27,7 +27,7 @@ http_interactions:
       Content-Length:
       - '57'
       Date:
-      - Fri, 17 Feb 2017 07:46:57 GMT
+      - Fri, 17 Feb 2017 08:06:57 GMT
       Connection:
       - close
     body:
@@ -35,12 +35,12 @@ http_interactions:
       string: |-
         [
          {
-         "DOI": "10.7554/ELIFE.01567",
+         "DOI": "10.7554/elife.01567",
          "RA": "Crossref"
         }
         ]
     http_version: 
-  recorded_at: Fri, 17 Feb 2017 07:47:00 GMT
+  recorded_at: Fri, 17 Feb 2017 08:06:58 GMT
 - request:
     method: get
     uri: https://doi.org/10.7554/elife.01567
@@ -70,14 +70,14 @@ http_interactions:
       Content-Length:
       - '177'
       Date:
-      - Fri, 17 Feb 2017 07:47:07 GMT
+      - Fri, 17 Feb 2017 08:06:58 GMT
     body:
       encoding: UTF-8
       string: |-
         <html><head><title>Handle Redirect</title></head>
         <body><a href="http://data.crossref.org/10.7554%2Felife.01567">http://data.crossref.org/10.7554%2Felife.01567</a></body></html>
     http_version: 
-  recorded_at: Fri, 17 Feb 2017 07:47:08 GMT
+  recorded_at: Fri, 17 Feb 2017 08:06:58 GMT
 - request:
     method: get
     uri: http://data.crossref.org/10.7554%2Felife.01567
@@ -113,7 +113,7 @@ http_interactions:
       Server:
       - http-kit
       Date:
-      - Fri, 17 Feb 2017 07:47:07 GMT
+      - Fri, 17 Feb 2017 08:06:58 GMT
       Connection:
       - close
     body:
@@ -756,5 +756,5 @@ http_interactions:
         aWNsZT4NCiAgICAgIDwvam91cm5hbD4NCiAgICA8L2Nyb3NzcmVmPg0KICA8
         L2RvaV9yZWNvcmQ+DQo8L2RvaV9yZWNvcmRzPg==
     http_version: 
-  recorded_at: Fri, 17 Feb 2017 07:47:09 GMT
+  recorded_at: Fri, 17 Feb 2017 08:06:58 GMT
 recorded_with: VCR 3.0.3

data/spec/fixtures/vcr_cassettes/Bolognese_CLI/read/crossref/as_datacite.yml

@@ -2,7 +2,7 @@
 http_interactions:
 - request:
     method: get
-    uri: https://doi.crossref.org/doiRA/10.7554/ELIFE.01567
+    uri: https://doi.crossref.org/doiRA/10.7554/elife.01567
     body:
       encoding: US-ASCII
       string: ''
@@ -27,7 +27,7 @@ http_interactions:
       Content-Length:
       - '57'
       Date:
-      - Fri, 17 Feb 2017 07:47:08 GMT
+      - Fri, 17 Feb 2017 08:06:54 GMT
       Connection:
       - close
     body:
@@ -35,12 +35,12 @@ http_interactions:
       string: |-
         [
          {
-         "DOI": "10.7554/ELIFE.01567",
+         "DOI": "10.7554/elife.01567",
          "RA": "Crossref"
         }
         ]
     http_version: 
-  recorded_at: Fri, 17 Feb 2017 07:47:09 GMT
+  recorded_at: Fri, 17 Feb 2017 08:06:54 GMT
 - request:
     method: get
     uri: https://doi.org/10.7554/elife.01567
@@ -70,14 +70,14 @@ http_interactions:
       Content-Length:
       - '177'
       Date:
-      - Fri, 17 Feb 2017 07:47:16 GMT
+      - Fri, 17 Feb 2017 08:06:53 GMT
     body:
       encoding: UTF-8
       string: |-
         <html><head><title>Handle Redirect</title></head>
         <body><a href="http://data.crossref.org/10.7554%2Felife.01567">http://data.crossref.org/10.7554%2Felife.01567</a></body></html>
     http_version: 
-  recorded_at: Fri, 17 Feb 2017 07:47:18 GMT
+  recorded_at: Fri, 17 Feb 2017 08:06:54 GMT
 - request:
     method: get
     uri: http://data.crossref.org/10.7554%2Felife.01567
@@ -113,7 +113,7 @@ http_interactions:
       Server:
       - http-kit
       Date:
-      - Fri, 17 Feb 2017 07:47:17 GMT
+      - Fri, 17 Feb 2017 08:06:54 GMT
       Connection:
       - close
     body:
@@ -756,7 +756,7 @@ http_interactions:
         aWNsZT4NCiAgICAgIDwvam91cm5hbD4NCiAgICA8L2Nyb3NzcmVmPg0KICA8
         L2RvaV9yZWNvcmQ+DQo8L2RvaV9yZWNvcmRzPg==
     http_version: 
-  recorded_at: Fri, 17 Feb 2017 07:47:18 GMT
+  recorded_at: Fri, 17 Feb 2017 08:06:55 GMT
 - request:
     method: get
     uri: https://doi.org/10.7554/elife.01567
@@ -786,14 +786,14 @@ http_interactions:
       Content-Length:
       - '177'
       Date:
-      - Fri, 17 Feb 2017 07:47:25 GMT
+      - Fri, 17 Feb 2017 08:06:54 GMT
     body:
       encoding: UTF-8
       string: |-
         <html><head><title>Handle Redirect</title></head>
         <body><a href="http://data.crossref.org/10.7554%2Felife.01567">http://data.crossref.org/10.7554%2Felife.01567</a></body></html>
     http_version: 
-  recorded_at: Fri, 17 Feb 2017 07:47:26 GMT
+  recorded_at: Fri, 17 Feb 2017 08:06:55 GMT
 - request:
     method: get
     uri: http://data.crossref.org/10.7554%2Felife.01567
@@ -829,7 +829,7 @@ http_interactions:
       Server:
       - http-kit
       Date:
-      - Fri, 17 Feb 2017 07:47:26 GMT
+      - Fri, 17 Feb 2017 08:06:55 GMT
       Connection:
       - close
     body:
@@ -1472,5 +1472,5 @@ http_interactions:
         aWNsZT4NCiAgICAgIDwvam91cm5hbD4NCiAgICA8L2Nyb3NzcmVmPg0KICA8
         L2RvaV9yZWNvcmQ+DQo8L2RvaV9yZWNvcmRzPg==
     http_version: 
-  recorded_at: Fri, 17 Feb 2017 07:47:27 GMT
+  recorded_at: Fri, 17 Feb 2017 08:06:55 GMT
 recorded_with: VCR 3.0.3