bolognese 1.9.12 → 2.0.0

data/spec/readers/datacite_reader_spec.rb

@@ -42,7 +42,7 @@ describe Bolognese::Metadata, vcr: true do
         {"subject"=>"taxonomy"},
         {"subject"=>"mitochondrial genome"},
         {"subject"=>"Parasites"}])
-      expect(subject.publisher).to eq("Dryad")
+      expect(subject.publisher).to eq({"name"=>"Dryad"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     end
@@ -65,7 +65,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.publication_year).to eq("2016")
       expect(subject.related_identifiers.length).to eq(3)
       expect(subject.related_identifiers.last).to eq("relatedIdentifier"=>"10.5438/0000-00ss", "relatedIdentifierType"=>"DOI", "relationType"=>"IsPartOf")
-      expect(subject.publisher).to eq("DataCite")
+      expect(subject.publisher).to eq({"name"=>"DataCite"})
       expect(subject.subjects).to eq([{"subject"=>"datacite"},
         {"subject"=>"doi"},
         {"subject"=>"metadata"},
@@ -96,7 +96,7 @@ describe Bolognese::Metadata, vcr: true do
        {"subject"=>"000 Computer science, knowledge, general works",
         "subjectScheme"=>"DDC"}])
       expect(subject.publication_year).to eq("2013")
-      expect(subject.publisher).to eq("Schloss Dagstuhl - Leibniz-Zentrum fuer Informatik GmbH, Wadern/Saarbruecken, Germany")
+      expect(subject.publisher).to eq({"name"=>"Schloss Dagstuhl - Leibniz-Zentrum fuer Informatik GmbH, Wadern/Saarbruecken, Germany"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-2.1")
     end
@@ -121,7 +121,7 @@ describe Bolognese::Metadata, vcr: true do
         "name"=>"Brown University"},
        {"affiliationIdentifier"=>"https://grid.ac/institutes/grid.268117.b",
         "affiliationIdentifierScheme"=>"GRID",
-        "name"=>"Wesleyan University"}])
+        "name"=>"Wesleyan University", "schemeUri"=>"https://grid.ac/institutes/"}])
       expect(subject.creators.dig(2, "affiliation")).to eq([{"affiliationIdentifier"=>"https://ror.org/05gq02987",
         "affiliationIdentifierScheme"=>"ROR", "name"=>"Brown University"}])
       expect(subject.titles).to eq([{"lang"=>"en-US", "title"=>"Full DataCite XML Example"}, {"lang"=>"en-US", "title"=>"Demonstration of DataCite Properties.", "titleType"=>"Subtitle"}])
@@ -141,6 +141,7 @@ describe Bolognese::Metadata, vcr: true do
         "awardTitle"=>"Full DataCite XML Example",
         "funderIdentifier"=>"https://doi.org/10.13039/100000001",
         "funderIdentifierType"=>"Crossref Funder ID",
+        "schemeUri"=>"https://doi.org/",
         "funderName"=>"National Science Foundation"}])
       expect(subject.related_identifiers.length).to eq(2)
       expect(subject.related_identifiers.first).to eq("relatedIdentifier"=>"https://data.datacite.org/application/citeproc+json/10.5072/example-full", "relatedIdentifierType"=>"URL", "relationType"=>"HasMetadata", "relatedMetadataScheme"=>"citeproc+json", "schemeUri"=>"https://github.com/citation-style-language/schema/raw/master/csl-data.json")
@@ -148,7 +149,12 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.language).to eq("en-US")
       expect(subject.sizes).to eq(["4 kB"])
       expect(subject.formats).to eq(["application/xml"])
-      expect(subject.publisher).to eq("DataCite")
+      expect(subject.publisher).to eq(
+        {
+          "name"=>"DataCite",
+          "lang"=>"en",
+        }
+      )
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     end
@@ -165,7 +171,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.titles).to eq([{"lang"=>"en", "title"=>"Data underpinning - The Sun as a planet-host star: Proxies from SDO images for HARPS radial-velocity variations"}])
       expect(subject.dates).to eq([{"date"=>"2016-01-20", "dateType"=>"Available"}, {"date"=>"2016", "dateType"=>"Issued"}])
       expect(subject.publication_year).to eq("2016")
-      expect(subject.publisher).to eq("University of St Andrews")
+      expect(subject.publisher).to eq({"name"=>"University of St Andrews"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     end
@@ -184,7 +190,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.publication_year).to eq("2011")
       expect(subject.sizes).to eq([])
       expect(subject.subjects).to eq([{"subject"=>"Environmental research"}])
-      expect(subject.publisher).to eq("EvK2 CNR Committee")
+      expect(subject.publisher).to eq({"name"=>"EvK2 CNR Committee"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-2.2")
     end
@@ -201,7 +207,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.types["citeproc"]).to eq("article")
       expect(subject.creators).to eq([{"nameType"=>"Personal", "name"=>"Garza, Kristian", "givenName"=>"Kristian", "familyName"=>"Garza", "nameIdentifiers" => [], "affiliation" => []}])
       expect(subject.titles).to eq([{"title"=>"Analysis Tools For Crossover Experiment Of Ui Using Choice Architecture"}])
-      expect(subject.descriptions.first["description"]).to start_with("This tools are used to analyse the data produced by the Crosssover Experiment")
+      expect(subject.descriptions.first["description"]).to start_with(" \n\nThis tools are used to analyse the data produced by the Crosssover Experiment I designed to test Choice Architecture techniques as UI interventions in a SEEk4Science data catalogue. It contains:\n\n- Data structures for the experimental data.<br>\n- Visualisation functions<br>\n- Analysis functions\n\n## Installation\n\n- R<br>\n- python<br>\n- ipython 4\n\nClone and use.\n\n## Usage\n\n<br>\n```python<br>\nsource('parallel_plot.r')<br>\nwith(z, parallelset(trt,response, freq=count, alpha=0.2))<br>\n```\n\n<br>\n## Contributing\n\n1. Fork it!<br>\n2. Create your feature branch: `git checkout -b my-new-feature`<br>\n3. Commit your changes: `git commit -am 'Add some feature'`<br>\n4. Push to the branch: `git push origin my-new-feature`<br>\n5. Submit a pull request :D\n\n<br>\n## License\n\nThis work supports my PhD Thesis at University of Manchester.")
       expect(subject.rights_list).to eq([{"rights"=>
         "Creative Commons Attribution Non Commercial Share Alike 4.0 International",
         "rightsIdentifier"=>"cc-by-nc-sa-4.0",
@@ -216,7 +222,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.subjects).to eq([{"subject"=>"choice architecture"},
         {"subject"=>"crossover experiment"},
         {"subject"=>"hci"}])
-      expect(subject.publisher).to eq("Zenodo")
+      expect(subject.publisher).to eq({"name"=>"Zenodo"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     end
@@ -242,7 +248,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.publication_year).to eq("2016")
       expect(subject.related_identifiers.length).to eq(1)
       expect(subject.related_identifiers.last).to eq("relatedIdentifier"=>"10.6084/m9.figshare.4234751", "relatedIdentifierType"=>"DOI", "relationType"=>"IsIdenticalTo")
-      expect(subject.publisher).to eq("figshare")
+      expect(subject.publisher).to eq({"name"=>"figshare"})
       expect(subject.subjects).to eq([
         {"schemeUri"=>
           "http://www.abs.gov.au/ausstats/abs@.nsf/0/6BB427AB9696C225CA2574180004463E",
@@ -285,7 +291,7 @@ describe Bolognese::Metadata, vcr: true do
         "schemeUri"=>"https://spdx.org/licenses/"}])
       expect(subject.dates).to eq([{"date"=>"2015-06-14", "dateType"=>"Created"}, {"date"=>"2020-06-02", "dateType"=>"Updated"}, {"date"=>"2020", "dateType"=>"Issued"}])
       expect(subject.publication_year).to eq("2020")
-      expect(subject.publisher).to eq("figshare")
+      expect(subject.publisher).to eq({"name"=>"figshare"})
       expect(subject.subjects).to eq([{"subject"=>"Evolutionary Biology"},
         {"subject"=>"FOS: Biological sciences", "subjectScheme"=>"Fields of Science and Technology (FOS)", "schemeUri"=>"http://www.oecd.org/science/inno/38235147.pdf"},
         {"subject"=>"60412 Quantitative Genetics (incl. Disease and Trait Mapping Genetics)", "subjectScheme"=>"FOR", "schemeUri"=>"http://www.abs.gov.au/ausstats/abs@.nsf/0/6BB427AB9696C225CA2574180004463E"}])
@@ -337,7 +343,7 @@ describe Bolognese::Metadata, vcr: true do
       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.publisher).to eq({"name"=>"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/",
@@ -381,7 +387,7 @@ describe Bolognese::Metadata, vcr: true do
       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.publisher).to eq({"name"=>"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/",
@@ -408,7 +414,7 @@ describe Bolognese::Metadata, vcr: true do
       subject = Bolognese::Metadata.new(input: input)
       expect(subject.types["schemaOrg"]).to eq("Collection")
       expect(subject.language).to eq("de")
-      expect(subject.publisher).to eq("Universitätsbibliothek Tübingen")
+      expect(subject.publisher).to eq({"name"=>"Universitätsbibliothek Tübingen"})
       expect(subject.publication_year).to eq("2015")
       expect(subject.valid?).to be false
       expect(subject.errors).to eq("13:0: ERROR: Element '{http://datacite.org/schema/kernel-2.2}publisher': This element is not expected. Expected is ( {http://datacite.org/schema/kernel-2.2}publicationYear ).")
@@ -420,10 +426,12 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.funding_references.first).to eq({
         "funderIdentifier"=>"http://www.isni.org/isni/0000000119587073",
         "funderIdentifierType"=>"ISNI",
+        "schemeUri"=>"http://www.isni.org/isni/",
         "funderName"=>"National Science Foundation (NSF)"})
-      expect(subject.funding_references.last).to eq({
+      expect(subject.funding_references[2]).to eq({
         "funderIdentifier"=>"https://doi.org/10.13039/501100000780",
         "funderIdentifierType"=>"Crossref Funder ID",
+        "schemeUri"=>"https://doi.org/",
         "funderName"=>"European Commission"})
       expect(subject.funding_references[1]).to eq({
           "funderIdentifier"=>"1234",
@@ -431,6 +439,35 @@ describe Bolognese::Metadata, vcr: true do
           "funderName"=>"Acme Inc"})
     end
 
+    it "funder identifier with normalized identifiers and non-normalized identifiers" do
+      input = fixture_path + 'datacite-funderIdentifier.xml'
+      subject = Bolognese::Metadata.new(input: input)
+      expect(subject.funding_references[3]).to eq({
+        "funderIdentifier"=>"0000 0001 2298 6657",
+        "funderIdentifierType"=>"ISNI",
+        "schemeUri"=>"http://www.isni.org/isni/",
+        "funderName"=>"University of Washington"})
+      expect(subject.funding_references[4]).to eq({
+        "funderIdentifier"=>"127467345",
+        "funderIdentifierType"=>"Other",
+        "schemeUri"=>"http://viaf.org/viaf/",
+        "funderName"=>"National Library of Ireland."})
+      expect(subject.funding_references[5]).to eq({
+          "funderIdentifier"=>"https://ror.org/04wxnsj81",
+          "funderIdentifierType"=>"ROR",
+          "schemeUri"=>"https://ror.org",
+          "funderName"=>"DataCite"})
+      expect(subject.funding_references[6]).to eq({
+          "funderIdentifier"=>"https://ror.org/038wwg650",
+          "funderIdentifierType"=>"ROR",
+          "schemeUri"=>"https://ror.org",
+          "funderName"=>"Department of Agriculture"})
+      expect(subject.funding_references[7]).to eq({
+        "funderIdentifier"=>"https://doi.org/10.13039/501100005695",
+        "funderIdentifierType"=>"Crossref Funder ID",
+        "funderName"=>"Tottori University"})
+    end
+
     it "geo_location empty" do
       input = fixture_path + 'datacite-geolocation-empty.xml'
       subject = Bolognese::Metadata.new(input: input)
@@ -500,7 +537,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.creators.length).to eq(1)
       expect(subject.creators.first).to eq("familyName"=>"Fenner", "givenName"=>"Martin", "nameIdentifiers" => [{"nameIdentifier"=>"https://orcid.org/0000-0003-1419-2405", "nameIdentifierScheme"=>"ORCID", "schemeUri"=>"https://orcid.org"}], "name"=>"Fenner, Martin")
       expect(subject.titles).to eq([{"title"=>"Eating your own Dog Food"}])
-      expect(subject.publisher).to eq("DataCite")
+      expect(subject.publisher).to eq({"name"=>"DataCite"})
       expect(subject.publication_year).to eq("2016")
       expect(subject.related_identifiers).to eq([{"relatedIdentifier"=>"10.5438/0012",
         "relatedIdentifierType"=>"DOI",
@@ -515,7 +552,7 @@ describe Bolognese::Metadata, vcr: true do
         "descriptionType"=>"SeriesInformation",
         "lang"=>"en"},
        {"description"=>
-        "Eating your own dog food is a slang term to describe that an organization should itself use the products and services it provides. For DataCite this means that we should use DOIs with appropriate metadata and strategies for long-term preservation for...",
+        "Eating your own dog food is a slang term to \n describe that an organization should itself use the products and services it provides. \n For DataCite this means that we should use DOIs with appropriate metadata and strategies for long-term preservation for...\n • Unicode Bullet Point: This is an example of a bullet point.\\u2605\n ■ Unicode Black Square: This is an example of a black square.",
         "descriptionType"=>"Abstract"}])
       expect(subject.container).to eq("firstPage"=>"3", "identifier"=>"10.5438/0000-00SS", "identifierType"=>"DOI", "issue"=>"9", "lastPage"=>"4", "title"=>"DataCite Blog", "type"=>"Series", "volume"=>"2")
     end
@@ -530,7 +567,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.creators.length).to eq(3)
       expect(subject.creators.first).to eq("familyName"=>"Schumann", "givenName"=>"Kai", "name"=>"Schumann, Kai", "nameType"=>"Personal", "nameIdentifiers" => [], "affiliation" => [])
       expect(subject.titles).to eq([{"title"=>"Gridded results of swath bathymetric mapping of Disko Bay, Western Greenland, 2007-2008"}])
-      expect(subject.publisher).to eq("PANGAEA - Data Publisher for Earth & Environmental Science")
+      expect(subject.publisher).to eq({"name"=>"PANGAEA - Data Publisher for Earth & Environmental Science"})
       expect(subject.publication_year).to eq("2011")
       expect(subject.related_identifiers).to eq([{"relatedIdentifier"=>"10.5072/timeseries", "relatedIdentifierType"=>"DOI", "relationType"=>"Continues"}])
       expect(subject.geo_locations).to eq([{"geoLocationPlace"=>"Disko Bay", "geoLocationPoint"=>{"pointLatitude"=>"69.000000", "pointLongitude"=>"-52.000000"}}])
@@ -547,7 +584,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.creators.length).to eq(6)
       expect(subject.creators.first).to eq("familyName"=>"Bales", "givenName"=>"Roger", "name"=>"Bales, Roger", "nameType"=>"Personal", "affiliation" => [{"name"=>"UC Merced"}, {"name"=>"NSF"}], "nameIdentifiers" => [])
       expect(subject.titles).to eq([{"title"=>"Southern Sierra Critical Zone Observatory (SSCZO), Providence Creek meteorological data, soil moisture and temperature, snow depth and air temperature"}])
-      expect(subject.publisher).to eq("UC Merced")
+      expect(subject.publisher).to eq({"name"=>"UC Merced"})
       expect(subject.dates).to eq([{"date"=>"2014-10-17", "dateType"=>"Updated"}, {"date"=>"2016-03-14T17:02:02Z", "dateType"=>"Available"}, {"date"=>"2013", "dateType"=>"Issued"}])
       expect(subject.publication_year).to eq("2013")
       expect(subject.subjects).to eq([{"subject"=>"Earth sciences"},
@@ -592,7 +629,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.id).to eq("https://doi.org/10.18429/jacow-ipac2016-tupmy003")
       expect(subject.types["schemaOrg"]).to eq("ScholarlyArticle")
       expect(subject.creators.length).to eq(12)
-      expect(subject.creators.first).to eq("nameType"=>"Personal", "nameIdentifiers" => [{"nameIdentifier"=>"http://jacow.org/JACoW-00077389", "nameIdentifierScheme"=>"JACoW-ID", "schemeUri"=>"http://jacow.org/"}], "name"=>"Otani, Masashi", "givenName"=>"Masashi", "familyName"=>"Otani", "affiliation" => [{"name"=>"KEK, Tsukuba, Japan"}])
+      expect(subject.creators.first).to eq("nameType"=>"Personal", "nameIdentifiers" => [{"nameIdentifier"=>"JACoW-00077389", "nameIdentifierScheme"=>"JACoW-ID", "schemeUri"=>"http://jacow.org/"}], "name"=>"Otani, Masashi", "givenName"=>"Masashi", "familyName"=>"Otani", "affiliation" => [{"name"=>"KEK, Tsukuba, Japan"}])
     end
 
     it "author with wrong orcid scheme" do
@@ -627,7 +664,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.titles).to eq([{"title"=>"Dataset for \"Direct Evidence for Solid-Like Hydrogen in a Nanoporous Carbon Hydrogen Storage Material at Supercritical Temperatures\""}])
       expect(subject.descriptions.first["description"]).to start_with("Dataset for Direct Evidence for Solid-Like Hydrogen")
       expect(subject.publication_year).to eq("2015")
-      expect(subject.publisher).to eq("University of Bath")
+      expect(subject.publisher).to eq({"name"=>"University of Bath"})
       expect(subject.funding_references.length).to eq(5)
       expect(subject.funding_references.first).to eq("awardNumber" => "EP/J016454/1",
         "awardTitle" => "SUPERGEN Hub Funding",
@@ -661,7 +698,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.titles).to eq([{"title"=>"Technical and Human Infrastructure for Open Research (THOR)"}])
       expect(subject.descriptions.first["description"]).to start_with("Five years ago, a global infrastructure")
       expect(subject.publication_year).to eq("2015")
-      expect(subject.publisher).to eq("DataCite")
+      expect(subject.publisher).to eq({"name"=>"DataCite"})
       expect(subject.funding_references).to eq([{"awardNumber"=>"654039",
         "awardTitle"=>"THOR – Technical and Human Infrastructure for Open Research",
         "awardUri"=>"http://cordis.europa.eu/project/rcn/194927_en.html",
@@ -685,7 +722,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.titles).to eq([{"title"=>"МОДЕЛЬ СИСТЕМНОЙ ДИНАМИКИ ПРОЦЕССА ОБУЧЕНИЯ"}, {"title"=>"MODEL OF SYSTEM DYNAMICS OF PROCESS OF TRAINING", "titleType"=>"TranslatedTitle"}])
       expect(subject.descriptions.first["description"]).to start_with("Актуальность данной работы обусловлена важностью учета в учебном процессе личностных качеств обучаем")
       expect(subject.publication_year).to eq("2019")
-      expect(subject.publisher).to eq("МОДЕЛИРОВАНИЕ, ОПТИМИЗАЦИЯ И ИНФОРМАЦИОННЫЕ ТЕХНОЛОГИИ")
+      expect(subject.publisher).to eq({"name"=>"МОДЕЛИРОВАНИЕ, ОПТИМИЗАЦИЯ И ИНФОРМАЦИОННЫЕ ТЕХНОЛОГИИ"})
       expect(subject.funding_references.length).to eq(1)
       expect(subject.funding_references.first).to eq("awardNumber"=>"проект № 170100728", "funderName"=>"РФФИ")
       expect(subject.agency).to eq("datacite")
@@ -722,7 +759,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.types["ris"]).to eq("BOOK")
       expect(subject.types["citeproc"]).to eq("book")
       expect(subject.creators).to eq([{"nameType"=>"Personal", "name"=>"Smith, John", "givenName"=>"John", "familyName"=>"Smith", "nameIdentifiers" => [], "affiliation" => []}, {"name"=>"つまらないものですが","nameIdentifiers"=>
-        [{"nameIdentifier"=>"http://isni.org/isni/0000000134596520",
+        [{"nameIdentifier"=>"0000000134596520",
           "nameIdentifierScheme"=>"ISNI",
           "schemeUri"=>"http://isni.org/isni/"}],
           "affiliation" => []}])
@@ -738,7 +775,7 @@ describe Bolognese::Metadata, vcr: true do
         "rightsIdentifierScheme"=>"SPDX",
         "rightsUri"=>"https://creativecommons.org/licenses/by-nd/2.0/legalcode",
         "schemeUri"=>"https://spdx.org/licenses/"}])
-      expect(subject.publisher).to eq("Springer")
+      expect(subject.publisher).to eq({"name"=>"Springer"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     end
@@ -753,7 +790,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.types["ris"]).to eq("BOOK")
       expect(subject.types["citeproc"]).to eq("book")
       expect(subject.creators).to eq([{"nameType"=>"Personal", "name"=>"Smith, John", "givenName"=>"John", "familyName"=>"Smith", "nameIdentifiers" => [], "affiliation" => []}, {"name"=>"つまらないものですが","nameIdentifiers"=>
-        [{"nameIdentifier"=>"http://isni.org/isni/0000000134596520",
+        [{"nameIdentifier"=>"0000000134596520",
           "nameIdentifierScheme"=>"ISNI",
           "schemeUri"=>"http://isni.org/isni/"}],
           "affiliation" => []}])
@@ -768,7 +805,7 @@ describe Bolognese::Metadata, vcr: true do
         "rightsIdentifierScheme"=>"SPDX",
         "rightsUri"=>"https://creativecommons.org/licenses/by-nd/2.0/legalcode",
         "schemeUri"=>"https://spdx.org/licenses/"}])
-      expect(subject.publisher).to eq("Springer")
+      expect(subject.publisher).to eq({"name"=>"Springer"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4.0")
     end
@@ -797,7 +834,7 @@ describe Bolognese::Metadata, vcr: true do
         "rightsIdentifierScheme"=>"SPDX",
         "rightsUri"=>"https://creativecommons.org/publicdomain/zero/1.0/legalcode",
         "schemeUri"=>"https://spdx.org/licenses/"}])
-      expect(subject.publisher).to eq("Dryad Digital Repository")
+      expect(subject.publisher).to eq({"name"=>"Dryad Digital Repository"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-3")
     end
@@ -812,7 +849,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.types["ris"]).to eq("BOOK")
       expect(subject.types["citeproc"]).to eq("book")
       expect(subject.creators).to eq([{"nameType"=>"Personal", "name"=>"Smith, John", "givenName"=>"John", "familyName"=>"Smith", "nameIdentifiers" => [], "affiliation" => []}, {"name"=>"つまらないものですが","nameIdentifiers"=>
-        [{"nameIdentifier"=>"http://isni.org/isni/0000000134596520",
+        [{"nameIdentifier"=>"0000000134596520",
           "nameIdentifierScheme"=>"ISNI",
           "schemeUri"=>"http://isni.org/isni/"}],
           "affiliation" => []}])
@@ -827,7 +864,7 @@ describe Bolognese::Metadata, vcr: true do
         "rightsIdentifierScheme"=>"SPDX",
         "rightsUri"=>"https://creativecommons.org/licenses/by-nd/2.0/legalcode",
         "schemeUri"=>"https://spdx.org/licenses/"}])
-      expect(subject.publisher).to eq("Springer")
+      expect(subject.publisher).to eq({"name"=>"Springer"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-3.0")
     end
@@ -851,7 +888,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.publication_year).to eq("2010")
       expect(subject.related_identifiers.length).to eq(1)
       expect(subject.related_identifiers.last).to eq("relatedIdentifier"=>"10.5272/oldertestpub", "relatedIdentifierType"=>"DOI", "relationType"=>"IsPartOf")
-      expect(subject.publisher).to eq("Springer")
+      expect(subject.publisher).to eq({"name"=>"Springer"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-2.2")
     end
@@ -864,7 +901,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.types["resourceType"]).to eq("Monograph")
       expect(subject.types["resourceTypeGeneral"]).to eq("Text")
       expect(subject.creators).to eq([{"nameType"=>"Personal", "name"=>"Smith, John", "givenName"=>"John", "familyName"=>"Smith", "nameIdentifiers" => [], "affiliation" => []}, {"name"=>"つまらないものですが","nameIdentifiers"=>
-        [{"nameIdentifier"=>"http://isni.org/isni/0000000134596520",
+        [{"nameIdentifier"=>"0000000134596520",
           "nameIdentifierScheme"=>"ISNI",
           "schemeUri"=>"http://isni.org/isni/"}],
           "affiliation" => []}])
@@ -885,7 +922,7 @@ describe Bolognese::Metadata, vcr: true do
         "rightsIdentifierScheme"=>"SPDX",
         "rightsUri"=>"https://creativecommons.org/licenses/by-nd/2.0/legalcode",
         "schemeUri"=>"https://spdx.org/licenses/"}])
-      expect(subject.publisher).to eq("Springer")
+      expect(subject.publisher).to eq({"name"=>"Springer"})
       expect(subject.agency).to eq("datacite")
     end
 
@@ -902,7 +939,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.titles).to eq([{"title"=>"LAMMPS Data-File Generator"}])
       expect(subject.dates).to eq([{"date"=>"2018-07-18", "dateType"=>"Valid"}, {"date"=>"2018-07-18", "dateType"=>"Accepted"}, {"date"=>"2018", "dateType"=>"Issued"}])
       expect(subject.publication_year).to eq("2018")
-      expect(subject.publisher).to eq("nanoHUB")
+      expect(subject.publisher).to eq({"name"=>"nanoHUB"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-2.2")
     end
@@ -918,7 +955,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.types["ris"]).to eq("DATA")
       expect(subject.types["bibtex"]).to eq("misc")
       expect(subject.creators.length).to eq(1)
-      expect(subject.creators.first).to eq("affiliation" => [{"affiliationIdentifier"=>"https://ror.org/04zt3wx35", "affiliationIdentifierScheme"=>"ROR", "name"=>"Canada Mortgage and Housing Corporation"}],
+      expect(subject.creators.first).to eq("affiliation" => [{"affiliationIdentifier"=>"https://ror.org/04zt3wx35", "affiliationIdentifierScheme"=>"ROR", "name"=>"Canada Mortgage and Housing Corporation", "schemeUri"=>"https://ror.org"}],
         "name" => "Statistique Canada",
          "nameIdentifiers" => [],
         "nameType" => "Organizational")
@@ -934,7 +971,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.creators).to eq([{"familyName"=>"Wong", "givenName"=>"Takmeng", "name"=>"Wong, Takmeng"}])
       expect(subject.titles).to eq([{"title"=>"CERES Level 3 Cloud Type Historgram Terra+Aqua HDF file - Edition4"}])
       expect(subject.publication_year).to eq("2016")
-      expect(subject.publisher).to eq("NASA Langley Atmospheric Science Data Center DAAC")
+      expect(subject.publisher).to eq({"name"=>"NASA Langley Atmospheric Science Data Center DAAC"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     end
@@ -975,7 +1012,7 @@ describe Bolognese::Metadata, vcr: true do
         "subjectScheme"=>"CESSDA Topic Classification"}])
       expect(subject.dates).to eq([{"date"=>"1995-12", "dateType"=>"Collected"}, {"date"=>"1996", "dateType"=>"Issued"}])
       expect(subject.publication_year).to eq("1996")
-      expect(subject.publisher).to eq("GESIS Data Archive")
+      expect(subject.publisher).to eq({"name"=>"GESIS Data Archive"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     end
@@ -997,7 +1034,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.container).to eq("firstPage"=>"Spain; 3353", "lastPage"=>"3356", "title"=>"23rd European Photovoltaic Solar Energy Conference and Exhibition", "type"=>"Series", "volume"=>"1-5 September 2008")
       expect(subject.descriptions[1]["description"]).to start_with("Aim of this paper is the presentation")
       expect(subject.subjects).to eq([{"subject"=>"PV Systems"}, {"subject"=>"Off-grid Applications"}])
-      expect(subject.publisher).to eq("WIP-Munich")
+      expect(subject.publisher).to eq({"name"=>"WIP-Munich"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-2.2")
     end
@@ -1028,7 +1065,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.titles).to eq([{"title"=>"Messung der Bildunschaerfe in H.264-codierten Bildern und Videosequenzen"}])
       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.publisher).to eq({"name"=>"Siemens AG"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-3")
     end
@@ -1068,7 +1105,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.version_info).to eq("2")
       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.publisher).to eq({"name"=>"Royal Netherlands Meteorological Institute (KNMI)"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     end
@@ -1094,7 +1131,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.titles).to eq([{"title"=>"Test license"}])
       expect(subject.dates).to eq([{"date"=>"2018-01-12", "dateType"=>"Issued"}])
       expect(subject.publication_year).to eq("2018")
-      expect(subject.publisher).to eq("CaltechDATA")
+      expect(subject.publisher).to eq({"name"=>"CaltechDATA"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
       expect(subject.state).to eq("findable")
@@ -1111,7 +1148,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.titles).to eq([{"title"=>"Referee report. For: FL Regression Wellcome [version 1; referees: retracted]"}])
       expect(subject.dates).to eq([{"date"=>"2018", "dateType"=>"Issued"}])
       expect(subject.publication_year).to eq("2018")
-      expect(subject.publisher).to eq("F1000 Research Limited")
+      expect(subject.publisher).to eq({"name"=>"F1000 Research Limited"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-3")
       expect(subject.state).to eq("findable")
@@ -1130,7 +1167,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.creators.first).to eq("nameType"=>"Personal", "name"=>"Patel, Lina", "givenName"=>"Lina", "familyName"=>"Patel", "nameIdentifiers" => [], "affiliation" => [])
       expect(subject.titles).to eq([{"title"=>"Referee report. For: Gates - add article keywords to the metatags [version 2; referees: 1 approved]"}])
       expect(subject.publication_year).to eq("2018")
-      expect(subject.publisher).to eq("Gates Open Research")
+      expect(subject.publisher).to eq({"name"=>"Gates Open Research"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-3")
     end
@@ -1223,7 +1260,7 @@ describe Bolognese::Metadata, vcr: true do
         {"date"=>"2014", "dateType"=>"Copyrighted"},
         {"date"=>"2009-10-01/2014-01-23", "dateType"=>"Created"}])
       expect(subject.publication_year).to eq("2014")
-      expect(subject.publisher).to eq("Deutsches Elektronen-Synchrotron, DESY, Hamburg")
+      expect(subject.publisher).to eq({"name"=>"Deutsches Elektronen-Synchrotron, DESY, Hamburg"})
       expect(subject.agency).to eq("datacite")
       expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-3")
     end
@@ -1248,7 +1285,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.titles.last).to eq("title"=>"MODEL OF SYSTEM DYNAMICS OF PROCESS OF TRAINING", "titleType"=>"TranslatedTitle")
       expect(subject.dates).to eq([{"date"=>"2019-02-09", "dateType"=>"Issued"}])
       expect(subject.publication_year).to eq("2019")
-      expect(subject.publisher).to eq("МОДЕЛИРОВАНИЕ, ОПТИМИЗАЦИЯ И ИНФОРМАЦИОННЫЕ ТЕХНОЛОГИИ")
+      expect(subject.publisher).to eq({"name"=>"МОДЕЛИРОВАНИЕ, ОПТИМИЗАЦИЯ И ИНФОРМАЦИОННЫЕ ТЕХНОЛОГИИ"})
       expect(subject.funding_references.count).to eq(1)
       expect(subject.funding_references.first).to eq("awardNumber"=>"проект № 170100728", "funderName"=>"РФФИ")
       expect(subject.agency).to eq("datacite")
@@ -1303,7 +1340,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.doi).to eq("10.5061/dryad.8515")
       expect(subject.creators).to eq([{"nameIdentifiers"=> [{"nameIdentifier"=>"https://orcid.org/0000-0003-1419-2405", "nameIdentifierScheme"=>"ORCID", "schemeUri"=>"https://orcid.org"}], "name"=>"Fenner, Martin", "givenName"=>"Martin", "familyName"=>"Fenner"}])
       expect(subject.titles).to eq([{"title"=>"Eating your own Dog Food"}])
-      expect(subject.publisher).to eq("DataCite")
+      expect(subject.publisher).to eq({"name"=>"DataCite"})
       expect(subject.publication_year).to eq("2016")
     end
   end
@@ -1330,7 +1367,7 @@ describe Bolognese::Metadata, vcr: true do
     expect(subject.formats).to eq(["application/tar"])
     expect(subject.sizes).to eq(["15.7M"])
     expect(subject.container).to eq("identifier"=>"https://www.ebi.ac.uk/miriam/main/datatypes/MIR:00000663", "identifierType"=>"URL", "title"=>"GTEx", "type"=>"DataRepository")
-    expect(subject.publisher).to eq("GTEx")
+    expect(subject.publisher).to eq({"name"=>"GTEx"})
     expect(subject.funding_references.count).to eq(7)
     expect(subject.funding_references.first).to eq("funderIdentifier"=>"https://doi.org/10.13039/100000052", "funderIdentifierType"=>"Crossref Funder ID", "funderName"=>"Common Fund of the Office of the Director of the NIH")
   end
@@ -1347,7 +1384,7 @@ describe Bolognese::Metadata, vcr: true do
     expect(subject.creators.first).to eq("familyName"=>"Den Heijer", "givenName"=>"C", "name"=>"Den Heijer, C", "nameType"=>"Personal", "nameIdentifiers" => [], "affiliation" => [])
     expect(subject.titles).to eq([{"lang"=>"en", "title"=>"Meteo measurements at the Sand Motor"}])
     expect(subject.publication_year).to eq("2017")
-    expect(subject.publisher).to eq("4TU.Centre for Research Data")
+    expect(subject.publisher).to eq({"name"=>"4TU.Centre for Research Data"})
     expect(subject.agency).to eq("datacite")
     expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     expect(subject.geo_locations.first["geoLocationPlace"]).to eq("Zandmotor, sand suppletion area on the Dutch coast.")
@@ -1413,7 +1450,12 @@ describe Bolognese::Metadata, vcr: true do
       "schemeUri"=>"https://spdx.org/licenses/"
       }
     ])
-    expect(subject.publisher).to eq("DataCite")
+    expect(subject.publisher).to eq(
+      {
+        "name"=>"DataCite",
+        "lang"=>"en",
+        }
+      )
     expect(subject.agency).to eq("datacite")
     expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
     expect(subject.related_items.last).to eq(
@@ -1523,6 +1565,49 @@ describe Bolognese::Metadata, vcr: true do
     )
   end
 
+  it "Schema 4.4 related items from string with attributes" do
+    input = fixture_path + "datacite-example-relateditems-with-attributes.xml"
+    subject = Bolognese::Metadata.new(input: input)
+    expect(subject.valid?).to be true
+
+    expect(subject.related_items.last).to eq(
+      {
+        "relatedItemType"=>"Journal",
+        "relationType"=>"IsPublishedIn",
+        "relatedItemIdentifier"=>
+        {
+          "relatedItemIdentifier"=>"10.5072/john-smiths-1234",
+          "relatedItemIdentifierType"=>"DOI",
+          "relatedMetadataScheme"=>"citeproc+json",
+          "schemeURI"=>"https://github.com/citation-style-language/schema/raw/master/csl-data.json",
+          "schemeType"=>"URL"
+        },
+        "creators" =>
+        [
+          {"nameType"=>"Personal", "name"=>"Smith, John", "givenName"=>"John", "familyName"=>"Smith"}
+        ],
+        "titles"=>
+        [
+          {"title"=>"Understanding the fictional John Smith"},
+          {"title"=>"A detailed look", "titleType"=>"Subtitle"}
+        ],
+        "volume"=>"776",
+        "issue"=>"1",
+        "number"=>"1",
+        "numberType"=>"Chapter",
+        "firstPage"=>"50",
+        "lastPage"=>"60",
+        "publisher"=>"Example Inc",
+        "publicationYear"=>"1776",
+        "edition"=>"1",
+        "contributors"=>
+        [
+          {"name"=>"Hallett, Richard", "givenName"=>"Richard", "familyName"=>"Hallett", "contributorType"=>"ProjectLeader"}
+        ]
+      }
+    )
+  end
+
   it "Schema 4.4 dissertation from string" do
     input = fixture_path + "datacite-example-dissertation-v4.4.xml"
     subject = Bolognese::Metadata.new(input: input)
@@ -1608,7 +1693,12 @@ describe Bolognese::Metadata, vcr: true do
       "schemeUri"=>"https://spdx.org/licenses/"
       }
     ])
-    expect(subject.publisher).to eq("GigaScience Database")
+    expect(subject.publisher).to eq(
+      {
+        "name"=>"GigaScience Database",
+        "lang"=>"en",
+      }
+    )
     expect(subject.agency).to eq("datacite")
     expect(subject.schema_version).to eq("http://datacite.org/schema/kernel-4")
   end
@@ -1653,4 +1743,20 @@ describe Bolognese::Metadata, vcr: true do
     )
   end
 
+  describe "Schema 4.5" do
+    it "publisher with attributes" do
+      input = fixture_path + "datacite-example-full-v4.5.xml"
+      subject = Bolognese::Metadata.new(input: input)
+      expect(subject.valid?).to be true
+      expect(subject.publisher).to eq( 
+        {
+          "name" => "Example Publisher",
+          "publisherIdentifier" => "https://ror.org/04z8jg394",
+          "publisherIdentifierScheme" => "ROR",
+          "schemeUri" => "https://ror.org/",
+          "lang" => "en",
+        }
+      )
+    end
+  end
 end

data/spec/readers/npm_reader_spec.rb

@@ -38,6 +38,7 @@ describe Bolognese::Metadata, vcr: true do
       #expect(subject.url).to eq("https://blog.datacite.org/eating-your-own-dog-food")
       expect(subject.types).to eq("bibtex"=>"misc", "citeproc"=>"article", "reourceType"=>"NPM Package", "resourceTypeGeneral"=>"Software", "ris"=>"GEN", "schemaOrg"=>"SoftwareSourceCode")
       expect(subject.creators).to eq([{"name"=>":(unav)", "nameType"=>"Organizational"}])
+      expect(subject.publisher).to eq(nil)
       expect(subject.titles).to eq([{"title"=>"CITapp"}])
       expect(subject.descriptions).to eq([{"description"=>"Concealed Information Test app", "descriptionType"=>"Abstract"}])
       expect(subject.version_info).to eq("1.1.0")
@@ -54,6 +55,7 @@ describe Bolognese::Metadata, vcr: true do
       #expect(subject.url).to eq("https://blog.datacite.org/eating-your-own-dog-food")
       expect(subject.types).to eq("bibtex"=>"misc", "citeproc"=>"article", "reourceType"=>"NPM Package", "resourceTypeGeneral"=>"Software", "ris"=>"GEN", "schemaOrg"=>"SoftwareSourceCode")
       expect(subject.creators).to eq([{"affiliation"=>[], "familyName"=>"Brancotte", "givenName"=>"Bryan", "name"=>"Brancotte, Bryan", "nameIdentifiers"=>[], "nameType"=>"Personal"}])
+      expect(subject.publisher).to eq(nil)
       expect(subject.titles).to eq([{"title"=>"edam-browser"}])
       expect(subject.descriptions).to eq([{"description"=>
         +   "The EDAM Browser is a client-side web-based visualization javascript widget. Its goals are to help describing bio-related resources and service with EDAM, and to facilitate and foster community contributions to EDAM.",

data/spec/readers/ris_reader_spec.rb

@@ -38,6 +38,7 @@ describe Bolognese::Metadata, vcr: true do
                                          "familyName"=>"Sankar",
                                          "nameIdentifiers" => [],
                                          "affiliation" => [])
+      expect(subject.publisher).to eq({"name"=>"(:unav)"})
       expect(subject.titles).to eq([{"title"=>"Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth"}])
       expect(subject.descriptions.first["description"]).to start_with("Among various advantages, their small size makes model organisms preferred subjects of investigation.")
       expect(subject.dates).to eq([{"date"=>"2014", "dateType"=>"Issued"}])
@@ -55,6 +56,7 @@ describe Bolognese::Metadata, vcr: true do
       expect(subject.id).to eq("https://doi.org/10.7554/elife.01567")
       expect(subject.types).to eq("citeproc"=>"misc", "resourceTypeGeneral"=>"Dissertation", "ris"=>"THES", "schemaOrg"=>"Thesis")
       expect(subject.creators).to eq([{"nameType"=>"Personal", "name"=>"Toparlar, Y.", "givenName"=>"Y.", "familyName"=>"Toparlar", "nameIdentifiers" => [], "affiliation" => []}])
+      expect(subject.publisher).to eq({"name"=>"Technische Universiteit Eindhoven"})
       expect(subject.titles).to eq([{"title"=>"A multiscale analysis of the urban heat island effect"}])
       expect(subject.descriptions.first["description"]).to start_with("Designing the climates of cities")
       expect(subject.dates).to eq([{"date"=>"2018-04-25", "dateType"=>"Issued"}, {"date"=>"2018-04-25", "dateType"=>"Created"}])
@@ -68,6 +70,7 @@ describe Bolognese::Metadata, vcr: true do
       subject = Bolognese::Metadata.new(input: input, doi: doi)
       expect(subject.valid?).to be true
       expect(subject.id).to eq("https://doi.org/10.7554/elife.01567")
+      expect(subject.publisher).to eq({"name"=>"Drexel University"})
       expect(subject.titles).to eq([{"title"=>"Validation of an Image-based Subject-Specific Dynamic Model of the Ankle Joint Complex and its Applications to the Study of the Effect of Articular Surface Morphology on Ankle Joint Mechanics"}])
       expect(subject.descriptions.first["description"]).to eq(abs)
     end