bolognese 1.0.19 → 1.0.20

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b7fe45911c1a9129e1524d9191961be4253f63bc02f235285d8337bee8cf32d4
-  data.tar.gz: ed331f04bdc2d3eef3caadc58255db850c7c079c5fd8f4d52381cf4f786ecdd1
+  metadata.gz: cdd3b249d4bb2a0126342f0fe581953de3bad55a6da02eda97129752b7caf5f0
+  data.tar.gz: b54781d9c4c6fdce604e2f4c12cb4a009a67f7c9e72b0034fa3a18829bd10d7c
 SHA512:
-  metadata.gz: 1fa38a18df702115dddc6764f7b33903f3e3d5d6742c8cf69106792eb00962e6b28f35f3f97e6bb55aca1fdabaa8e39db5f898d7545b2a11a2ba9b77bb732b30
-  data.tar.gz: 467ba2c38715b9523b673528e7d06022f3f52d124837e7e2ddadc37e176347e66471823b000214603ffb29b5d14f58f0d954972ec5e5a4939daef9d4cb3a1d1f
+  metadata.gz: 3e587b91a50c7617c450b373fcae22f97f5455c0566bbe33b1fab806b2a426f821e49692dc2a865f8a286b4d5c63f22caf92263c9fb2b576bf1bc5d480e183f5
+  data.tar.gz: 2d4d9c4629aa9787c04b2d5933fe269353de0f95e4f6038fed2ceb7663c113dcbdb010642f1229cfb762b3332a842895b937df7bc9b1edc127c0778a41d4d38c

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (1.0.19)
+    bolognese (1.0.20)
       activesupport (>= 4.2.5, < 6)
       benchmark_methods (~> 0.7)
       bibtex-ruby (~> 4.1)

data/lib/bolognese/metadata.rb

@@ -76,28 +76,31 @@ module Bolognese
       @content_url = hsh.to_h["content_url"].presence
 
       # set attributes directly
-      @creator = options[:creator]
-      @contributor = options[:contributor]
-      @titles = options[:titles]
-      @types = options[:types]
-      @alternate_identifiers = options[:alternate_identifiers]
-      @periodical = options[:periodical]
-      @publisher = options[:publisher]
-      @funding_references = options[:funding_references]
-      @dates = options[:dates]
-      @publication_year = options[:publication_year]
-      @descriptions = options[:descriptions]
-      @rights_list = options[:rights_list]
-      @version = options[:version]
-      @subjects = options[:subjects]
-      @language = options[:language]
-      @geo_locations = options[:geo_locations]
-      @related_identifiers = options[:related_identifiers]
-      @formats = options[:formats]
-      @sizes = options[:sizes]
+      read_options = options.slice(
+        :sandbox,
+        :creator,
+        :contributor,
+        :titles,
+        :types,
+        :alternate_identifiers,
+        :periodical,
+        :publisher,
+        :funding_references,
+        :dates,
+        :publication_year,
+        :descriptions,
+        :rights_list,
+        :version,
+        :subjects,
+        :language,
+        :geo_locations,
+        :related_identifiers,
+        :formats,
+        :sizes
+      ).compact
 
       # generate name for method to call dynamically
-      @meta = @from.present? ? send("read_" + @from, string: string, sandbox: options[:sandbox]) : {}
+      @meta = @from.present? ? send("read_" + @from, { string: string }.merge(read_options)) : {}
       @identifier = normalize_doi(options[:doi] || input, options) || @meta.fetch("id", nil) || @meta.fetch("identifier", nil)
     end
 

data/lib/bolognese/readers/bibtex_reader.rb

@@ -27,6 +27,8 @@ module Bolognese
       }
 
       def read_bibtex(string: nil, **options)
+        read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:string, :sandbox))
+
         meta = string.present? ? BibTeX.parse(string).first : OpenStruct.new
 
         schema_org = BIB_TO_SO_TRANSLATIONS[meta.try(:type).to_s] || "ScholarlyArticle"
@@ -66,7 +68,7 @@ module Bolognese
         end
 
         page_first, page_last = meta.try(:pages).to_s.split("-")
-        state = doi.present? ? "findable" : "not_found"
+        state = doi.present? || read_options.present? ? "findable" : "not_found"
         dates = if meta.try(:date).present?
           [{ "date" => meta.date.to_s,
              "dateType" => "Issued" }]
@@ -92,7 +94,7 @@ module Bolognese
           "descriptions" => meta.try(:abstract).present? ? [{ "description" => meta.try(:abstract) && sanitize(meta.abstract.to_s).presence, "descriptionType" => "Abstract" }] : [],
           "rights_list" =>  meta.try(:copyright).present? ? [{ "rightsUri" => meta.try(:copyright).to_s.presence }.compact] : [],
           "state" => state
-        }
+        }.merge(read_options)
       end
     end
   end

data/lib/bolognese/readers/citeproc_reader.rb

@@ -31,6 +31,8 @@ module Bolognese
           return { "errors" => errors } if errors.present?
         end
 
+        read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:string, :sandbox))
+
         meta = string.present? ? Maremma.from_json(string) : {}
 
         citeproc_type = meta.fetch("type", nil)
@@ -75,7 +77,7 @@ module Bolognese
           nil
         end
         id = normalize_id(meta.fetch("id", nil))
-        state = id.present? ? "findable" : "not_found"
+        state = id.present? || read_options.present? ? "findable" : "not_found"
         subjects = Array.wrap(meta.fetch("categories", nil)).map do |s|
           { "subject" => s }
         end
@@ -99,7 +101,7 @@ module Bolognese
           "version" => meta.fetch("version", nil),
           "subjects" => subjects,
           "state" => state
-        }
+        }.merge(read_options)
       end
     end
   end

data/lib/bolognese/readers/codemeta_reader.rb

@@ -18,6 +18,8 @@ module Bolognese
           return { "errors" => errors } if errors.present?
         end
 
+        read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:string, :sandbox))
+
         meta = string.present? ? Maremma.from_json(string) : {}
         identifier = meta.fetch("identifier", nil)
         id = normalize_id(meta.fetch("@id", nil) || identifier)
@@ -29,7 +31,7 @@ module Bolognese
         dates << { "date" => meta.fetch("dateModified"), "dateType" => "Updated" } if meta.fetch("dateModified", nil).present?
         publication_year = meta.fetch("datePublished")[0..3] if meta.fetch("datePublished", nil).present?
         publisher = meta.fetch("publisher", nil)
-        state = meta.present? ? "findable" : "not_found"
+        state = meta.present? || read_options.present? ? "findable" : "not_found"
         schema_org = meta.fetch("@type", nil)
         types = {
           "resourceTypeGeneral" => Bolognese::Utils::SO_TO_DC_TRANSLATIONS[schema_org],
@@ -60,7 +62,7 @@ module Bolognese
           "version" => meta.fetch("version", nil),
           "subjects" => subjects,
           "state" => state
-        }
+        }.merge(read_options)
       end
 
       # def related_identifiers(relation_type)

data/lib/bolognese/readers/crossref_reader.rb

@@ -20,6 +20,8 @@ module Bolognese
       end
 
       def read_crossref(string: nil, **options)
+        read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:string, :sandbox))
+
         if string.present?
           m = Maremma.from_xml(string).dig("doi_records", "doi_record") || {}
           meta = m.dig("crossref", "error").nil? ? m : {}
@@ -91,7 +93,7 @@ module Bolognese
             "dateType" => "Updated" }
         ]
         publication_year = crossref_date_published(bibliographic_metadata).present? ? crossref_date_published(bibliographic_metadata)[0..3] : nil
-        state = meta.present? ? "findable" : "not_found"
+        state = meta.present? || read_options.present? ? "findable" : "not_found"
 
         related_identifiers = Array.wrap(crossref_is_part_of(journal_metadata)) + Array.wrap(crossref_references(bibliographic_metadata))
         periodical = if journal_metadata.present?
@@ -129,7 +131,7 @@ module Bolognese
           "sizes" => nil,
           "schema_version" => nil,
           "state" => state
-        }
+        }.merge(read_options)
       end
 
       def crossref_alternate_identifiers(bibliographic_metadata)

data/lib/bolognese/readers/datacite_json_reader.rb

@@ -7,9 +7,11 @@ module Bolognese
         errors = jsonlint(string)
         return { "errors" => errors } if errors.present?
 
+        read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:string, :sandbox))
+
         meta = string.present? ? Maremma.from_json(string) : {}
 
-        state = meta.fetch("doi", nil).present? ? "findable" : "not_found"
+        state = meta.fetch("doi", nil).present? || read_options.present? ? "findable" : "not_found"
 
         dates = Array.wrap(meta.fetch("dates", nil)).map do |d|
           { "date" => d["date"],
@@ -51,7 +53,7 @@ module Bolognese
           "geo_locations" => meta.fetch("geoLocations", nil),
           "schema_version" => meta.fetch("schemaVersion", nil),
           "state" => state
-        }
+        }.merge(read_options)
       end
     end
   end

data/lib/bolognese/readers/datacite_reader.rb

@@ -49,6 +49,8 @@ module Bolognese
       end
 
       def read_datacite(string: nil, **options)
+        read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:string, :sandbox))
+
         doc = Nokogiri::XML(string, nil, 'UTF-8', &:noblanks)
         ns = doc.collect_namespaces.find { |k, v| v.start_with?("http://datacite.org/schema/kernel") }
         schema_version = Array.wrap(ns).last || "http://datacite.org/schema/kernel-4"
@@ -180,7 +182,7 @@ module Bolognese
           end
         end.compact
         periodical = set_periodical(meta)
-        state = doi.present? ? "findable" : "not_found"
+        state = doi.present? || read_options.present? ? "findable" : "not_found"
 
         { "id" => id,
           "types" => types,
@@ -207,7 +209,7 @@ module Bolognese
           "sizes" => sizes,
           "schema_version" => schema_version,
           "state" => state
-        }
+        }.merge(read_options)
       end
 
       def set_periodical(meta)

data/lib/bolognese/readers/ris_reader.rb

@@ -34,6 +34,8 @@ module Bolognese
       }
 
       def read_ris(string: nil, **options)
+        read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:string, :sandbox))
+
         meta = ris_meta(string: string)
 
         ris_type = meta.fetch("TY", nil) || "GEN"
@@ -69,7 +71,7 @@ module Bolognese
         else
           nil
         end
-        state = doi.present? ? "findable" : "not_found"
+        state = doi.present? || read_options.present? ? "findable" : "not_found"
         subjects = Array.wrap(meta.fetch("KW", nil)).map do |s|
           { "subject" => s }
         end
@@ -93,7 +95,7 @@ module Bolognese
           "subjects" => subjects,
           "language" => meta.fetch("LA", nil),
           "state" => state
-        }
+        }.merge(read_options)
       end
 
       def ris_meta(string: nil)

data/lib/bolognese/readers/schema_org_reader.rb

@@ -33,6 +33,8 @@ module Bolognese
           return { "errors" => errors } if errors.present?
         end
 
+        read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:string, :sandbox))
+
         meta = string.present? ? Maremma.from_json(string) : {}
 
         identifier = Array.wrap(meta.fetch("identifier", nil))
@@ -108,7 +110,7 @@ module Bolognese
         dates << { "date" => meta.fetch("dateModified"), "dateType" => "Updated" } if meta.fetch("dateModified", nil).present?
         publication_year = meta.fetch("datePublished")[0..3] if meta.fetch("datePublished", nil).present?
         
-        state = meta.present? ? "findable" : "not_found"
+        state = meta.present? || read_options.present? ? "findable" : "not_found"
         geo_locations = Array.wrap(meta.fetch("spatialCoverage", nil)).map do |gl|
           if gl.dig("geo", "box")
             s, w, n, e = gl.dig("geo", "box").split(" ", 4)
@@ -159,7 +161,7 @@ module Bolognese
           "schema_version" => meta.fetch("schemaVersion", nil).to_s.presence,
           "funding_references" => funding_references,
           "geo_locations" => geo_locations
-        }
+        }.merge(read_options)
       end
 
       def schema_org_related_identifier(meta, relation_type: nil)

data/lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "1.0.19"
+  VERSION = "1.0.20"
 end

data/spec/readers/datacite_reader_spec.rb

@@ -185,6 +185,7 @@ describe Bolognese::Metadata, vcr: true do
           "funderName"=>"European Commission"}],
         types: { "resourceTypeGeneral" => "Dataset", "schemaOrg" => "Dataset" })
       
+      expect(subject.valid?).to be true
       expect(subject.doi).to eq("10.5281/zenodo.1239")
       expect(subject.identifier).to eq("https://doi.org/10.5281/zenodo.1239")
       expect(subject.types["schemaOrg"]).to eq("Dataset")
@@ -203,6 +204,7 @@ describe Bolognese::Metadata, vcr: true do
         "funderName"=>"European Commission"}])
       expect(subject.agency).to eq("DataCite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
+      expect(subject.state).to eq("findable")
     end
 
     it "missing resource_type_general" do

data/spec/readers/schema_org_reader_spec.rb

@@ -229,5 +229,42 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.publisher).to eq("TOPMed")
       expect(subject.funding_references).to eq([{"funderIdentifier"=>"https://doi.org/10.13039/100000050", "funderIdentifierType"=>"Crossref Funder ID", "funderName"=>"National Heart, Lung, and Blood Institute (NHLBI)"}])
     end
+
+    it "from attributes" do
+      subject = Bolognese::Metadata.new(input: nil,
+        from: "schema_org",
+        doi: "10.5281/zenodo.1239",
+        creator: [{"type"=>"Person", "name"=>"Jahn, Najko", "givenName"=>"Najko", "familyName"=>"Jahn"}],
+        titles: [{ "title" => "Publication Fp7 Funding Acknowledgment - Plos Openaire" }],
+        descriptions: [{ "description" => "The dataset contains a sample of metadata describing papers" }],
+        publisher: "Zenodo",
+        publication_year: "2013",
+        dates: [{"date"=>"2013-04-03", "dateType"=>"Issued"}],
+        funding_references: [{"awardNumber"=>"246686",
+          "awardTitle"=>"Open Access Infrastructure for Research in Europe",
+          "awardUri"=>"info:eu-repo/grantAgreement/EC/FP7/246686/",
+          "funderIdentifier"=>"https://doi.org/10.13039/501100000780",
+          "funderIdentifierType"=>"Crossref Funder ID",
+          "funderName"=>"European Commission"}],
+        types: { "resourceTypeGeneral" => "Dataset", "schemaOrg" => "Dataset" })
+      
+      expect(subject.valid?).to be true
+      expect(subject.doi).to eq("10.5281/zenodo.1239")
+      expect(subject.identifier).to eq("https://doi.org/10.5281/zenodo.1239")
+      expect(subject.types["schemaOrg"]).to eq("Dataset")
+      expect(subject.types["resourceTypeGeneral"]).to eq("Dataset")
+      expect(subject.creator).to eq([{"familyName"=>"Jahn", "givenName"=>"Najko", "name"=>"Jahn, Najko", "type"=>"Person"}])
+      expect(subject.titles).to eq([{"title"=>"Publication Fp7 Funding Acknowledgment - Plos Openaire"}])
+      expect(subject.descriptions.first["description"]).to start_with("The dataset contains a sample of metadata describing papers")
+      expect(subject.dates).to eq([{"date"=>"2013-04-03", "dateType"=>"Issued"}])
+      expect(subject.publication_year).to eq("2013")
+      expect(subject.publisher).to eq("Zenodo")
+      expect(subject.funding_references).to eq([{"awardNumber"=>"246686",
+        "awardTitle"=>"Open Access Infrastructure for Research in Europe",
+        "awardUri"=>"info:eu-repo/grantAgreement/EC/FP7/246686/",
+        "funderIdentifier"=>"https://doi.org/10.13039/501100000780",
+        "funderIdentifierType"=>"Crossref Funder ID",
+        "funderName"=>"European Commission"}])
+    end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bolognese
 version: !ruby/object:Gem::Version
-  version: 1.0.19
+  version: 1.0.20
 platform: ruby
 authors:
 - Martin Fenner