bolognese 0.6.6 → 0.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 4cf9b774c122a31d80d29bac90783498458f5aff
-  data.tar.gz: 970db12ff440f5907609eaabbac07676b06fe1e6
+  metadata.gz: 510d880a02cea7933abd7de597f9a4061dad981b
+  data.tar.gz: 754fa0b49a23494efa1cac66c99688313bbaafca
 SHA512:
-  metadata.gz: 14b38895e972a549842ec6a1f7388f4f2bd5dff742860689d189a15702843ef3babb1b6b9ff181faabd626391f2e41abf8825aff596c67b8aa9bc422dcde547d
-  data.tar.gz: 639783c23011fc7567b9c8397bf214d0dcc89604660297cc77da29fd9a899f17bffd7bd8f142b50ee4c24804246b201f5278fc5bd4f9d2938262038d239f8b28
+  metadata.gz: 9899dd154362fea725898b2895664704b1f6cdb7a981fd10225094b29273e467f45c45e545fd966ff06451df107c9625e906438172fac0fc44e55608e96165bb
+  data.tar.gz: ba1af6c6cdc58219234ea03e7a989a8eb464fee854fe581ad280465d942db5212113a99e65c23d876e6642c3345d79ff5bdb7ec82c94bbc9a8709acb594eb803

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (0.6.6)
+    bolognese (0.7)
       activesupport (~> 4.2, >= 4.2.5)
       bibtex-ruby (~> 4.1)
       builder (~> 3.2, >= 3.2.2)

data/README.md

@@ -34,7 +34,7 @@ Bolognese reads and/or writes these metadata formats:
       <td>application/vnd.datacite.datacite+xml</td>
       <td>Yes</td>
       <td>Yes</td>
-  </tr>
+    </tr>
     <tr>
         <td><a href='http://schema.org/'>Schema.org in JSON-LD</a></td>
         <td>schema_org</td>
@@ -42,6 +42,13 @@ Bolognese reads and/or writes these metadata formats:
         <td>Yes</td>
         <td>Yes</td>
     </tr>
+    <tr>
+        <td><a href='https://codemeta.github.io/'>Codemeta</a></td>
+        <td>codemeta</td>
+        <td>application/ld+json</td>
+        <td>Yes</td>
+        <td>Yes</td>
+    </tr>
     <tr>
         <td><a href='http://en.wikipedia.org/wiki/BibTeX'>BibTeX</a></td>
         <td>bibtex</td>
@@ -754,6 +761,31 @@ bolognese 10.5061/DRYAD.8515 -t datacite --schema_version http://datacite.org/sc
 </resource>
 ```
 
+
+Convert DataCite XML to Codemeta:
+
+```
+bolognese https://doi.org/10.5063/f1m61h5x -t codemeta
+
+{  
+   "@context":"https://raw.githubusercontent.com/codemeta/codemeta/master/codemeta.jsonld",
+   "@type":"SoftwareSourceCode",
+   "@id":"https://doi.org/10.5063/f1m61h5x",
+   "identifier":"https://doi.org/10.5063/f1m61h5x",
+   "title":"dataone: R interface to the DataONE network of data repositories",
+   "agents":{  
+      "@type":"Person",
+      "givenName":"Matthew B.",
+      "familyName":"Jones"
+   },
+   "datePublished":"2016",
+   "publisher":{  
+      "@type":"Organization",
+      "name":"KNB Data Repository"
+   }
+}
+```
+
 Convert DataCite XML to BibTeX:
 
 ```

data/lib/bolognese.rb

@@ -9,6 +9,7 @@ require "bolognese/metadata"
 require "bolognese/crossref"
 require "bolognese/datacite"
 require "bolognese/schema_org"
+require "bolognese/codemeta"
 require "bolognese/bibtex"
 require "bolognese/orcid"
 require "bolognese/cli"

data/lib/bolognese/codemeta.rb

@@ -0,0 +1,145 @@
+module Bolognese
+  class Codemeta < Metadata
+
+    def initialize(id: nil, string: nil)
+      id = normalize_id(id) if id.present?
+
+      if string.present?
+        @raw = string
+      elsif id.present?
+        response = Maremma.get(id, accept: "application/ld+json", raw: true)
+        @raw = response.body.fetch("data", nil)
+      end
+    end
+
+    alias_method :schema_org, :as_schema_org
+    alias_method :bibtex, :as_bibtex
+
+    def metadata
+      @metadata ||= raw.present? ? Maremma.from_json(raw) : {}
+    end
+
+    def exists?
+      metadata.present?
+    end
+
+    def doi
+      doi_from_url(id)
+    end
+
+    def id
+      normalize_id(metadata.fetch("@id", nil) || identifier)
+    end
+
+    def identifier
+      metadata.fetch("identifier", nil)
+    end
+
+    def url
+      normalize_id(metadata.fetch("codeRepository", nil))
+    end
+
+    def resource_type_general
+      SO_TO_DC_TRANSLATIONS[type]
+    end
+
+    def type
+      metadata.fetch("@type", nil)
+    end
+
+    def additional_type
+      metadata.fetch("additionalType", nil)
+    end
+
+    def bibtex_type
+      Bolognese::Bibtex::SO_TO_BIB_TRANSLATIONS[type] || "misc"
+    end
+
+    def name
+      metadata.fetch("title", nil)
+    end
+
+    def alternate_name
+      metadata.fetch("alternateName", nil)
+    end
+
+    def author
+      a = Array.wrap(metadata.fetch("agents", nil)).map { |a| a.extract!("@type", "@id", "name") }
+      array_unwrap(a)
+    end
+
+    def editor
+      Array(metadata.fetch("editor", nil)).map { |a| a.except("name") }.presence
+    end
+
+    def description
+      metadata.fetch("description", nil)
+    end
+
+    def license
+      metadata.fetch("license", nil)
+    end
+
+    def version
+      metadata.fetch("version", nil)
+    end
+
+    def keywords
+      Array(metadata.fetch("tags", nil)).join(", ").presence
+    end
+
+    def date_created
+      metadata.fetch("dateCreated", nil)
+    end
+
+    def date_published
+      metadata.fetch("datePublished", nil)
+    end
+
+    def date_modified
+      metadata.fetch("dateModified", nil)
+    end
+
+    def related_identifiers(relation_type)
+      normalize_ids(metadata.fetch(relation_type, nil))
+    end
+
+    def same_as
+      related_identifiers("isIdenticalTo")
+    end
+
+    def is_part_of
+      related_identifiers("isPartOf")
+    end
+
+    def has_part
+      related_identifiers("hasPart")
+    end
+
+    def citation
+      related_identifiers("citation")
+    end
+
+    def publisher
+      p = metadata.fetch("publisher", nil)
+      if p.is_a?(Hash)
+        p
+      elsif p.is_a?(String)
+        { "@type" => "Organization",
+          "name" => p }
+      end
+    end
+
+    def container_title
+      if publisher.is_a?(Hash)
+        publisher.fetch("name", nil)
+      elsif publisher.is_a?(String)
+        publisher
+      end
+    end
+
+    def provider
+      metadata.fetch("provider", nil)
+    end
+  end
+end

data/lib/bolognese/crossref.rb

@@ -74,6 +74,7 @@ module Bolognese
     alias_method :crossref, :raw
     alias_method :as_crossref, :raw
     alias_method :schema_org, :as_schema_org
+    alias_method :codemeta, :as_codemeta
     alias_method :bibtex, :as_bibtex
 
     def metadata

data/lib/bolognese/datacite.rb

@@ -32,6 +32,7 @@ module Bolognese
     end
 
     alias_method :schema_org, :as_schema_org
+    alias_method :codemeta, :as_codemeta
     alias_method :bibtex, :as_bibtex
 
     def schema_version

data/lib/bolognese/metadata.rb

@@ -72,6 +72,24 @@ module Bolognese
       }.compact.to_json
     end
 
+    def as_codemeta
+      { "@context" => id.present? ? "https://raw.githubusercontent.com/codemeta/codemeta/master/codemeta.jsonld" : nil,
+        "@type" => type,
+        "@id" => id,
+        "identifier" => id,
+        "codeRepository" => url,
+        "title" => name,
+        "agents" => author,
+        "description" => description,
+        "version" => version,
+        "tags" => keywords.to_s.split(", ").presence,
+        "dateCreated" => date_created,
+        "datePublished" => date_published,
+        "dateModified" => date_modified,
+        "publisher" => publisher
+      }.compact.to_json
+    end
+
     def as_bibtex
       bib = {
         bibtex_type: bibtex_type.to_sym,

data/lib/bolognese/schema_org.rb

@@ -14,6 +14,7 @@ module Bolognese
     end
 
     alias_method :schema_org, :as_schema_org
+    alias_method :codemeta, :as_codemeta
     alias_method :bibtex, :as_bibtex
 
     def metadata
@@ -61,11 +62,13 @@ module Bolognese
     end
 
     def author
-      Array(metadata.fetch("author", nil)).map { |a| a.except("name") }.presence
+      a = Array.wrap(metadata.fetch("author", nil)).map { |a| a.except("name") }
+      array_unwrap(a)
     end
 
     def editor
-      Array(metadata.fetch("editor", nil)).map { |a| a.except("name") }.presence
+      a = Array.wrap(metadata.fetch("editor", nil)).map { |a| a.except("name") }
+      array_unwrap(a)
     end
 
     def description

data/lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "0.6.6"
+  VERSION = "0.7"
 end

data/spec/codemeta_spec.rb

@@ -0,0 +1,81 @@
+require 'spec_helper'
+
+describe Bolognese::Codemeta, vcr: true do
+  # let(:id) { "https://blog.datacite.org/eating-your-own-dog-food" }
+  let(:fixture_path) { "spec/fixtures/" }
+  #
+  # subject { Bolognese::SchemaOrg.new(id: id) }
+
+  # context "get metadata" do
+  #   it "BlogPosting" do
+  #     expect(subject.id).to eq("https://doi.org/10.5438/4k3m-nyvg")
+  #     expect(subject.url).to eq("https://blog.datacite.org/eating-your-own-dog-food")
+  #     expect(subject.type).to eq("BlogPosting")
+  #     expect(subject.author).to eq([{"@type"=>"Person", "@id"=>"http://orcid.org/0000-0003-1419-2405", "givenName"=>"Martin", "familyName"=>"Fenner"}])
+  #     expect(subject.name).to eq("Eating your own Dog Food")
+  #     expect(subject.alternate_name).to eq("MS-49-3632-5083")
+  #     expect(subject.description).to start_with("Eating your own dog food")
+  #     expect(subject.keywords).to eq("datacite, doi, metadata, featured")
+  #     expect(subject.date_published).to eq("2016-12-20")
+  #     expect(subject.date_modified).to eq("2016-12-20")
+  #     expect(subject.is_part_of).to eq("@type"=>"Blog", "@id"=>"https://doi.org/10.5438/0000-00ss", "name"=>"DataCite Blog")
+  #     expect(subject.citation).to eq([{"@type"=>"CreativeWork", "@id"=>"https://doi.org/10.5438/0012"},
+  #                                     {"@type"=>"CreativeWork", "@id"=>"https://doi.org/10.5438/55e5-t5c0"}])
+  #     expect(subject.publisher).to eq("@type"=>"Organization", "name"=>"DataCite")
+  #   end
+  #
+  #   it "BlogPosting schema.org JSON" do
+  #     json = JSON.parse(subject.as_schema_org)
+  #     expect(json["@id"]).to eq("https://doi.org/10.5438/4k3m-nyvg")
+  #   end
+  #
+  #   it "not found error" do
+  #     id = "https://doi.org/10.5438/4K3M-NYVGx"
+  #     subject = Bolognese::SchemaOrg.new(id: id)
+  #     expect(subject.id).to be_nil
+  #     expect(subject.exists?).to be false
+  #   end
+  # end
+
+  context "get metadata as string" do
+    let(:string) { IO.read(fixture_path + 'codemeta.json') }
+
+    subject { Bolognese::Codemeta.new(string: string) }
+
+    it "SoftwareSourceCode" do
+      expect(subject.id).to eq("https://doi.org/10.5063/f1m61h5x")
+      expect(subject.url).to eq("https://github.com/DataONEorg/rdataone")
+      expect(subject.type).to eq("SoftwareSourceCode")
+      expect(subject.author).to eq([{"@type"=>"person",
+                                     "@id"=>"http://orcid.org/0000-0003-0077-4738",
+                                     "name"=>"Matt Jones"},
+                                    {"@type"=>"person",
+                                     "@id"=>"http://orcid.org/0000-0002-2192-403X",
+                                     "name"=>"Peter Slaughter"},
+                                    {"@type"=>"organization",
+                                     "@id"=>"http://orcid.org/0000-0002-3957-2474",
+                                     "name"=>"University of California, Santa Barbara"}])
+      expect(subject.name).to eq("R Interface to the DataONE REST API")
+      expect(subject.description).to start_with("Provides read and write access to data and metadata")
+      expect(subject.keywords).to eq("data sharing, data repository, DataONE")
+      expect(subject.version).to eq("2.0.0")
+      expect(subject.date_created).to eq("2016-05-27")
+      expect(subject.date_published).to eq("2016-05-27")
+      expect(subject.date_modified).to eq("2016-05-27")
+      expect(subject.publisher).to eq("@type"=>"Organization", "name"=>"https://cran.r-project.org")
+    end
+  end
+
+  context "get metadata as datacite xml" do
+    let(:string) { IO.read(fixture_path + 'codemeta.json') }
+
+    subject { Bolognese::Codemeta.new(string: string) }
+
+    it "SoftwareSourceCode" do
+      expect(subject.validation_errors).to be_empty
+      datacite = Maremma.from_xml(subject.as_datacite).fetch("resource", {})
+      expect(datacite.dig("titles", "title")).to eq("R Interface to the DataONE REST API")
+      expect(datacite.fetch("version")).to eq("2.0.0")
+    end
+  end
+end

data/spec/datacite_spec.rb

@@ -227,6 +227,22 @@ describe Bolognese::Datacite, vcr: true do
     end
   end
 
+  context "get metadata as codemeta" do
+    it "SoftwareSourceCode" do
+      id = "https://doi.org/10.5063/f1m61h5x"
+      subject = Bolognese::Datacite.new(id: id)
+      json = JSON.parse(subject.as_codemeta)
+      expect(json["@context"]).to eq("https://raw.githubusercontent.com/codemeta/codemeta/master/codemeta.jsonld")
+      expect(json["@id"]).to eq("https://doi.org/10.5063/f1m61h5x")
+      expect(json["@type"]).to eq("SoftwareSourceCode")
+      expect(json["identifier"]).to eq("https://doi.org/10.5063/f1m61h5x")
+      expect(json["agents"]).to eq("@type"=>"Person", "givenName"=>"Matthew B.", "familyName"=>"Jones")
+      expect(json["title"]).to eq("dataone: R interface to the DataONE network of data repositories")
+      expect(json["datePublished"]).to eq("2016")
+      expect(json["publisher"]).to eq("@type"=>"Organization", "name"=>"KNB Data Repository")
+    end
+  end
+
   context "get metadata as bibtex" do
     it "Dataset" do
       bibtex = BibTeX.parse(subject.as_bibtex).to_a(quotes: '').first

data/spec/fixtures/codemeta.json

@@ -0,0 +1,87 @@
+{
+	"@context": "https://raw.githubusercontent.com/codemeta/codemeta/master/codemeta.jsonld",
+	"@type": "SoftwareSourceCode",
+	"agents": [{
+		"@id": "http://orcid.org/0000-0003-0077-4738",
+		"@type": "person",
+		"email": "jones@nceas.ucsb.edu",
+		"name": "Matt Jones",
+		"affiliation": "NCEAS",
+		"mustBeCited": true,
+		"isMaintainer": true,
+		"isRightsHolder": true,
+		"role": {
+			"namespace": "http://www.ngdc.noaa.gov/metadata/published/xsd/schema/resources/Codelist/gmxCodelists.xml#CI_RoleCode",
+			"roleCode": [
+				"originator",
+				"resourceProvider"
+			]
+		}
+	}, {
+		"@id": "http://orcid.org/0000-0002-2192-403X",
+		"@type": "person",
+		"email": "slaughter@nceas.ucsb.edu",
+		"name": "Peter Slaughter",
+		"affiliation": "NCEAS",
+		"mustBeCited": true,
+		"isMaintainer": false,
+		"isRightsHolder": false,
+		"role": {
+			"namespace": "http://www.ngdc.noaa.gov/metadata/published/xsd/schema/resources/Codelist/gmxCodelists.xml#CI_RoleCode",
+			"roleCode": "contributor"
+		}
+	}, {
+		"@id": "http://orcid.org/0000-0002-3957-2474",
+		"@type": "organization",
+		"email": "info@ucop.edu",
+		"name": "University of California, Santa Barbara",
+		"role": {
+			"namespace": "http://www.ngdc.noaa.gov/metadata/published/xsd/schema/resources/Codelist/gmxCodelists.xml#CI_RoleCode",
+			"roleCode": "copyrightHolder"
+		}
+	}],
+	"depends": {
+		"@type": "URL",
+		"@value": "https://github.com/ropensci/datapack"
+	},
+	"identifier": "http://dx.doi.org/10.5063/F1M61H5X",
+	"codeRepository": "https://github.com/DataONEorg/rdataone",
+	"controlledTem": "software",
+	"datePublished": "2016-05-27",
+	"dateModified": "2016-05-27",
+	"dateCreated": "2016-05-27",
+	"description": "Provides read and write access to data and metadata from the DataONE network of data repositories.Each DataONE repository implements a consistent repository application programming interface. Users call methods in R to access these remote repository functions, such as methods to query the metadata catalog, get access to metadata for particular data packages, and read the data objects from the data repository.Users can also insert and update data objects on repositories that support these methods.",
+	"downloadLink": "https://cran.r-project.org/src/contrib/dataone_2.0.0.tar.gz",
+	"funding": "National Science Foundation grant #012345678",
+	"isAutomatedBuild": false,
+	"issueTracker": "https://github.com/DataONEorg/rdataone/issues",
+	"licenseId": "Apache-2.0",
+	"publisher": "https://cran.r-project.org",
+	"tags": [
+		"data sharing",
+		"data repository",
+		"DataONE"
+	],
+	"title": "R Interface to the DataONE REST API",
+	"version": "2.0.0",
+	"uploadedBy": {
+		"@id": "http://orcid.org/0000-0003-0077-4738",
+		"@type": "person",
+		"email": "mbjones@nceas.ucsb.edu",
+		"name": "Matt Jones"
+	},
+	"programmingLanguage": {
+		"name": "R",
+		"version": "> 3.1.1",
+		"URL": "https://www.r-project.org"
+	},
+	"readme": "https://github.com/DataONEorg/rdataone/README.md",
+	"relatedPublications": "ISBN:0201703726",
+	"relationship": {
+		"relationshipType": "isPartOf",
+		"relatedIdentifier": "urn:uuid:F1A0A7AF-ECF3-4C7D-B675-7C6949963995",
+		"relatedIdentifierType": "UUID"
+	},
+	"softwarePaperCitationIdentifiers": "http://doi.org/0000/0000",
+	"zippedCode": "https://cran.r-project.org/src/contrib/dataone_2.0.0.tar.gz"
+}