bolognese 1.3.11 → 1.3.13

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4f48b979736d6c650c605afb1ef0593cba49c8f3e0f719f571dd8c0a46035f91
-  data.tar.gz: 87be2f1d2d8e744d4b2a7095a78d767270dd374670df6e4c02790e482bce0962
+  metadata.gz: 17907b9f240b77cd0be24a0bc4ad4787a0849e64264cbbaa26e0583183b7e8f3
+  data.tar.gz: 6d8f7836ec34a526a2478ef58cadf3e040475772eeb6bc0b0336490aaf525b48
 SHA512:
-  metadata.gz: bc6fe47b28d56eef7bd8d301fd0824608b122f502a15154240a5b9554e49d3e2fba077f7547face75f5c10340c1be4464af2afe52c5c8f90a5e345bf32812ac2
-  data.tar.gz: 2bc6010a67fddd5208fcce970296f57ca399f0b18dea24d7e4314376baebc59b8ec08e23c84e50bf450df9a1fda177279c614a9834957b76c08baa8c85760bd3
+  metadata.gz: f4dd37336d18737867915f56d93d5b27e17b81f7770c56466c6b08c8e1c695752fed591dc51fd9e45d9313e47be0e2f8b50fb7a564f95e1e1b39e4866e4a77c9
+  data.tar.gz: 1b2f2cd3262ba69e9326c5da28dc8c7b771b12cbf2568c8c9b88068216e6c610638f1977fe0ed29ded47c431a79724ec35370585976f53b2d5188102f5b44a3b

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (1.3.11)
+    bolognese (1.3.13)
       activesupport (>= 4.2.5, < 6)
       benchmark_methods (~> 0.7)
       bibtex-ruby (~> 4.1)

data/lib/bolognese.rb

@@ -18,6 +18,7 @@ require 'gender_detector'
 require 'citeproc/ruby'
 require 'citeproc'
 require 'csl/styles'
+require 'edtf'
 
 require "bolognese/version"
 require "bolognese/metadata"

data/lib/bolognese/readers/bibtex_reader.rb

@@ -50,37 +50,31 @@ module Bolognese
         end
 
         related_identifiers = if meta.try(:journal).present? && meta.try(:issn).to_s.presence
-          [{ "type" => "Periodical",
-             "relationType" => "IsPartOf",
-             "relatedIdentifierType" => "ISSN",
-             "title" => meta.journal.to_s,
-             "relatedIdentifier" => meta.try(:issn).to_s.presence }.compact]
-        else
-          nil
-        end
+                                [{ "type" => "Periodical",
+                                  "relationType" => "IsPartOf",
+                                  "relatedIdentifierType" => "ISSN",
+                                  "title" => meta.journal.to_s,
+                                  "relatedIdentifier" => meta.try(:issn).to_s.presence }.compact]
+                              end
 
         container = if meta.try(:journal).present?
-          first_page = meta.try(:pages).present? ? meta.try(:pages).split("-").map(&:strip)[0] : nil
-          last_page = meta.try(:pages).present? ? meta.try(:pages).split("-").map(&:strip)[1] : nil
-          
-          { "type" => "Journal",
-            "title" => meta.journal.to_s,
-            "identifier" => meta.try(:issn).to_s.presence,
-            "identifierType" => meta.try(:issn).present? ? "ISSN" : nil,
-            "volume" => meta.try(:volume).to_s.presence,
-            "firstPage" => first_page,
-            "lastPage" => last_page }.compact
-        else
-          nil
-        end
+                      first_page = meta.try(:pages).present? ? meta.try(:pages).split("-").map(&:strip)[0] : nil
+                      last_page = meta.try(:pages).present? ? meta.try(:pages).split("-").map(&:strip)[1] : nil
+                      
+                      { "type" => "Journal",
+                        "title" => meta.journal.to_s,
+                        "identifier" => meta.try(:issn).to_s.presence,
+                        "identifierType" => meta.try(:issn).present? ? "ISSN" : nil,
+                        "volume" => meta.try(:volume).to_s.presence,
+                        "firstPage" => first_page,
+                        "lastPage" => last_page }.compact
+                    end
 
         state = meta.try(:doi).to_s.present? || read_options.present? ? "findable" : "not_found"
-        dates = if meta.try(:date).present?
-          [{ "date" => meta.date.to_s,
-             "dateType" => "Issued" }]
-        else
-          nil
-        end
+        dates = if meta.try(:date).present? && Date.edtf(meta.date.to_s).present?
+                  [{ "date" => meta.date.to_s,
+                    "dateType" => "Issued" }]
+                end
         publication_year =  meta.try(:date).present? ? meta.date.to_s[0..3] : nil
 
         { "id" => normalize_doi(doi),

data/lib/bolognese/readers/citeproc_reader.rb

@@ -52,27 +52,23 @@ module Bolognese
           [{ "nameType" => "Organizational", "name" => ":(unav)" }]
         end
         contributors = get_authors(from_citeproc(Array.wrap(meta.fetch("editor", nil))))
-        dates = if meta.fetch("issued", nil).present?
-          [{ "date" => get_date_from_date_parts(meta.fetch("issued", nil)),
-             "dateType" => "Issued" }]
-        else
-          nil
-        end
+        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("copyright", nil)
-          [{ "rightsUri" => normalize_url(meta.fetch("copyright")) }.compact]
-        else
-          nil
-        end
+                        [{ "rightsUri" => normalize_url(meta.fetch("copyright")) }.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]
-        else
-          nil
-        end
+                                [{ "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

data/lib/bolognese/readers/crossref_reader.rb

@@ -106,38 +106,46 @@ module Bolognese
                    [{ "title" => ":{unav)" }]
                  end
 
-        date_created = Array.wrap(query.to_h["crm_item"]).find { |cr| cr["name"] == "created" }
+        date_published = crossref_date_published(bibliographic_metadata)
+        if date_published.present?
+          date_published = { "date" => date_published, "dateType" => "Issued" }
+        else
+          date_published = Array.wrap(query.to_h["crm_item"]).find { |cr| cr["name"] == "created" }
+          date_published = { "date" => date_published.fetch("__content__", "")[0..9], "dateType" => "Issued" } if date_published.present?
+        end
         date_updated = Array.wrap(query.to_h["crm_item"]).find { |cr| cr["name"] == "last-update" }
         date_updated = { "date" => date_updated.fetch("__content__", nil), "dateType" => "Updated" } if date_updated.present?
-        dates = [
-          { "date" => crossref_date_published(bibliographic_metadata) || date_created.to_h.fetch("__content__", "")[0..9].presence, "dateType" => "Issued" }, date_updated
-        ].compact
-        publication_year = crossref_date_published(bibliographic_metadata).present? ? crossref_date_published(bibliographic_metadata)[0..3] : date_created.to_h.fetch("__content__", "")[0..3].presence
+        
+        # check that date is valid iso8601 date
+        date_published = nil unless Date.edtf(date_published.to_h["date"]).present?
+        date_updated = nil unless Date.edtf(date_updated.to_h["date"]).present?
+        
+        dates = [date_published, date_updated].compact
+        publication_year = date_published.to_h.fetch("date", "")[0..3].presence
+        
         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))
         container = if journal_metadata.present? || book_metadata.present?
-          issn = normalize_issn(journal_metadata.to_h.fetch("issn", nil))
+                      issn = normalize_issn(journal_metadata.to_h.fetch("issn", nil))
 
-          { "type" => "Journal",
-            "identifier" => issn,
-            "identifierType" => issn.present? ? "ISSN" : nil,
-            "title" => parse_attributes(journal_metadata.to_h["full_title"]),
-            "volume" => parse_attributes(journal_issue.dig("journal_volume", "volume")),
-            "issue" => parse_attributes(journal_issue.dig("issue")),
-            "firstPage" => bibliographic_metadata.dig("pages", "first_page"),
-            "lastPage" => bibliographic_metadata.dig("pages", "last_page") }.compact
-        elsif book_series_metadata.to_h.fetch("series_metadata", nil).present?
-          issn = normalize_issn(book_series_metadata.dig("series_metadata", "issn"))
+                      { "type" => "Journal",
+                        "identifier" => issn,
+                        "identifierType" => issn.present? ? "ISSN" : nil,
+                        "title" => parse_attributes(journal_metadata.to_h["full_title"]),
+                        "volume" => parse_attributes(journal_issue.dig("journal_volume", "volume")),
+                        "issue" => parse_attributes(journal_issue.dig("issue")),
+                        "firstPage" => bibliographic_metadata.dig("pages", "first_page"),
+                        "lastPage" => bibliographic_metadata.dig("pages", "last_page") }.compact
+                    elsif book_series_metadata.to_h.fetch("series_metadata", nil).present?
+                      issn = normalize_issn(book_series_metadata.dig("series_metadata", "issn"))
 
-          { "type" => "Book Series",
-            "identifier" => issn,
-            "identifierType" => issn.present? ? "ISSN" : nil,
-            "title" => book_series_metadata.dig("series_metadata", "titles", "title"),
-            "volume" => bibliographic_metadata.fetch("volume", nil) }.compact
-        else
-          nil
-        end
+                      { "type" => "Book Series",
+                        "identifier" => issn,
+                        "identifierType" => issn.present? ? "ISSN" : nil,
+                        "title" => book_series_metadata.dig("series_metadata", "titles", "title"),
+                        "volume" => bibliographic_metadata.fetch("volume", nil) }.compact
+                    end
 
         identifiers = [{ "identifierType" => "DOI", "identifier" => normalize_doi(options[:doi] || options[:id] || bibliographic_metadata.dig("doi_data", "doi")) }, crossref_alternate_identifiers(bibliographic_metadata)].compact
 

data/lib/bolognese/readers/datacite_reader.rb

@@ -127,20 +127,40 @@ module Bolognese
             nil
           elsif k.is_a?(String)
             { "subject" => sanitize(k) }
-          else
+          elsif k.is_a?(Hash)
             { "subject" => sanitize(k["__content__"]), "subjectScheme" => k["subjectScheme"], "schemeUri" => k["schemeURI"], "valueUri" => k["valueURI"], "lang" => k["lang"] }.compact
           end
         end.compact
-        dates = Array.wrap(meta.dig("dates", "date")).map do |d|
-          {
-            "date" => parse_attributes(d),
-            "dateType" => parse_attributes(d, content: "dateType"),
-            "dateInformation" => parse_attributes(d, content: "dateInformation")
-          }.compact
+        dates = Array.wrap(meta.dig("dates", "date")).map do |r|
+          if r.is_a?(Hash) && date = sanitize(r["__content__"]).presence
+            if Date.edtf(date).present?
+              { "date" => date, 
+                "dateType" => parse_attributes(r, content: "dateType"),
+                "dateInformation" => parse_attributes(r, content: "dateInformation")
+              }.compact
+            end
+          end
         end.compact
         dates << { "date" => meta.fetch("publicationYear", nil), "dateType" => "Issued" } if meta.fetch("publicationYear", nil).present? && get_date(dates, "Issued").blank?
-        sizes = Array.wrap(meta.dig("sizes", "size"))
-        formats = Array.wrap(parse_attributes(meta.dig("formats", "format")))
+        sizes = Array.wrap(meta.dig("sizes", "size")).map do |k|
+          if k.blank?
+            nil
+          elsif k.is_a?(String)
+            sanitize(k).presence
+          elsif k.is_a?(Hash)
+            sanitize(k["__content__"]).presence
+          end
+        end.compact
+        formats = Array.wrap(meta.dig("formats", "format")).map do |k|
+          if k.blank?
+            nil
+          elsif k.is_a?(String)
+            sanitize(k).presence
+          elsif k.is_a?(Hash)
+            sanitize(k["__content__"]).presence
+          end
+        end.compact
+        .map { |s| s.to_s.squish.presence }.compact
         funding_references = Array.wrap(meta.dig("fundingReferences", "fundingReference")).compact.map do |fr|
           scheme_uri = parse_attributes(fr["funderIdentifier"], content: "schemeURI")
           funder_identifier = parse_attributes(fr["funderIdentifier"])

data/lib/bolognese/readers/schema_org_reader.rb

@@ -124,9 +124,9 @@ module Bolognese
           end
         end
         dates = []
-        dates << { "date" => meta.fetch("datePublished"), "dateType" => "Issued" } if meta.fetch("datePublished", nil).present?
-        dates << { "date" => meta.fetch("dateCreated"), "dateType" => "Created" } if meta.fetch("dateCreated", nil).present?
-        dates << { "date" => meta.fetch("dateModified"), "dateType" => "Updated" } if meta.fetch("dateModified", nil).present?
+        dates << { "date" => meta.fetch("datePublished"), "dateType" => "Issued" } if Date.edtf(meta.fetch("datePublished", nil)).present?
+        dates << { "date" => meta.fetch("dateCreated"), "dateType" => "Created" } if Date.edtf(meta.fetch("dateCreated", nil)).present?
+        dates << { "date" => meta.fetch("dateModified"), "dateType" => "Updated" } if Date.edtf(meta.fetch("dateModified", nil)).present?
         publication_year = meta.fetch("datePublished")[0..3] if meta.fetch("datePublished", nil).present?
 
         state = meta.present? || read_options.present? ? "findable" : "not_found"

data/lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "1.3.11"
+  VERSION = "1.3.13"
 end

data/spec/fixtures/datacite-empty-sizes.xml

@@ -0,0 +1,57 @@
+<?xml version="1.0"?>
+<resource xmlns="http://datacite.org/schema/kernel-2.2" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://datacite.org/schema/kernel-2.2 http://schema.datacite.org/meta/kernel-2.2/metadata.xsd">
+  <identifier identifierType="DOI">10.13116/c6b6060e-0648-4f88-b108-95ea00928171</identifier>
+  <creators>
+    <creator>
+      <creatorName>EvK2 CNR Committee</creatorName>
+    </creator>
+  </creators>
+  <titles>
+    <title>SHARE (Stations at High Altitude for Research on the Environment) Network</title>
+    <title titleType="Subtitle">Urdukas (Baltoro Glacier, Baltistan - Pakistan)</title>
+  </titles>
+  <publisher>EvK2 CNR Committee</publisher>
+  <publicationYear>2011</publicationYear>
+  <subjects>
+    <subject>Environmental research</subject>
+  </subjects>
+  <contributors>
+    <contributor contributorType="Funder">
+      <contributorName>EvK2 CNR Committee</contributorName>
+    </contributor>
+  </contributors>
+  <dates>
+    <date dateType="Valid">
+		</date>
+  </dates>
+  <language>en</language>
+  <resourceType resourceTypeGeneral="Dataset"/>
+  <relatedIdentifiers>
+    <relatedIdentifier relatedIdentifierType="URL" relationType="IsDocumentedBy">http://geonetwork.evk2cnr.org:8080/geonetwork/srv/en/main.home</relatedIdentifier>
+  </relatedIdentifiers>
+  <sizes>
+    <size>
+		</size>
+  </sizes>
+  <formats>
+    <format>text</format>
+  </formats>
+  <version>1.0</version>
+  <rights>
+	</rights>
+  <descriptions>
+    <description descriptionType="Abstract">The Baltoro Glacier, 62 km long, is one of the longest glaciers outside of the Polar Regions. It is located in Baltistan, in the Northern Areas of Pakistan, and runs through part of the Karakoram mountain range. In the middle of the Baltoro glacier there is the campsite of Urdukas located on the trek way to Concordia. 
+This automatic weather station was installed, in June 17, 2004 on the Baltoro Glacier, more precisely on a moraine ridge close to the left glacier margin at 3926 m asl.
+
+Geographical coordinates:
+Latitude: 35deg 43' 41" N 
+Longitude: 76deg 17' 10" EThe Urdukas AWS provides data (with a 60 minute time resolution" of the following parameters:
+- air temperature (degC)
+- atmopherich pressure (hPa)
+- relative humidity (%)
+- total precipitation (mm)
+- global solar radiation (W/m2)
+- wind speed (m/s)
+- wind direction (degree)</description>
+  </descriptions>
+</resource>

data/spec/readers/datacite_reader_spec.rb

@@ -151,6 +151,24 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     end
 
+    it "empty sizes and dates attributes" do
+      input = fixture_path + 'datacite-empty-sizes.xml'
+      subject = Bolognese::Metadata.new(input: input)
+      expect(subject.valid?).to be true
+      expect(subject.types["schemaOrg"]).to eq("Dataset")
+      expect(subject.types["resourceType"]).to be_nil
+      expect(subject.types["resourceTypeGeneral"]).to eq("Dataset")
+      expect(subject.creators.length).to eq(1)
+      expect(subject.creators.first).to eq("affiliation"=>[], "name"=>"EvK2 CNR Committee", "nameIdentifiers"=>[])
+      expect(subject.titles).to eq([{"title"=>"SHARE (Stations at High Altitude for Research on the Environment) Network"}, {"title"=>"Urdukas (Baltoro Glacier, Baltistan - Pakistan)", "titleType"=>"Subtitle"}])
+      expect(subject.dates).to eq([{"date"=>"2011", "dateType"=>"Issued"}])
+      expect(subject.publication_year).to eq("2011")
+      expect(subject.sizes).to eq([])
+      expect(subject.publisher).to eq("EvK2 CNR Committee")
+      expect(subject.agency).to eq("DataCite")
+      expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-2.2")
+    end
+
     it "multiple licenses" do
       input = "https://doi.org/10.5281/ZENODO.48440"
       subject = Bolognese::Metadata.new(input: input)
@@ -799,7 +817,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.types["resourceTypeGeneral"]).to eq("Dataset")
       expect(subject.creators).to eq([{"name"=>"Anonymous", "nameIdentifiers" => [], "affiliation" => []}])
       expect(subject.titles).to eq([{"title"=>"Messung der Bildunschaerfe in H.264-codierten Bildern und Videosequenzen"}])
-      expect(subject.dates).to eq([{"date"=>"07.04.2017", "dateType"=>"Available"}, {"date"=>"2017", "dateType"=>"Issued"}])
+      expect(subject.dates).to eq([{"date"=>"2017", "dateType"=>"Issued"}])
       expect(subject.publication_year).to eq("2017")
       expect(subject.publisher).to eq("Siemens AG")
       expect(subject.agency).to eq("DataCite")
@@ -836,7 +854,7 @@ describe Bolognese::Metadata, vcr: true do
                                      "affiliation"=>[{"name"=>"Royal Netherlands Meteorological Institute (KNMI)"}]}])
       expect(subject.titles).to eq([{"title"=>"Multi-Sensor Reanalysis (MSR) of total ozone, version 2"}])
       expect(subject.version_info).to eq("2")
-      expect(subject.dates).to eq([{"date"=>"2014-04-25", "dateType"=>"Available"}, {"date"=>"1970-04-01 / (:tba)", "dateType"=>"Collected"}, {"date"=>"2015", "dateType"=>"Issued"}])
+      expect(subject.dates).to eq([{"date"=>"2014-04-25", "dateType"=>"Available"}, {"date"=>"2015", "dateType"=>"Issued"}])
       expect(subject.publication_year).to eq("2015")
       expect(subject.publisher).to eq("Royal Netherlands Meteorological Institute (KNMI)")
       expect(subject.agency).to eq("DataCite")

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: bolognese
 version: !ruby/object:Gem::Version
-  version: 1.3.11
+  version: 1.3.13
 platform: ruby
 authors:
 - Martin Fenner
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-08-04 00:00:00.000000000 Z
+date: 2019-08-05 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: maremma
@@ -717,6 +717,7 @@ files:
 - spec/fixtures/crossref.bib
 - spec/fixtures/crossref.ris
 - spec/fixtures/crossref.xml
+- spec/fixtures/datacite-empty-sizes.xml
 - spec/fixtures/datacite-example-affiliation.xml
 - spec/fixtures/datacite-example-complicated-v3.0.xml
 - spec/fixtures/datacite-example-complicated-v4.0.xml