bolognese 1.10.0 → 2.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 532586fba70709be6c7581b595a2c0d72ad1000c9c54b951ad5e4758715e81c1
-  data.tar.gz: 3029c21ac460e41479459df642e3d501b1195e6cc8a0e26d6a4ba5bba91ce13f
+  metadata.gz: 23ccf372b59b7d9c8919ec7ad7137eef6e99512eed488977e847f938f389aa2f
+  data.tar.gz: 390e56023d57e637a52776a98dcd0bea4fa0acd8b8459ced05cbc6785f69fe0a
 SHA512:
-  metadata.gz: 3893a351e49ad8e8acb0ef5265fb3acdc149e67a99174de0d0855d594e7d9c16f09baeda4d9b1aba7235aef661135e7d1bd1ac401df81601c99217e155153395
-  data.tar.gz: f6225b13dae84dcc897fa1dda7aadb266e667f8bf6a3a2da2e3f6ad134f314f17c3c6613ef8c2dec54091d9f4b4f055023bdb00ace39baf018b1654820cf7eff
+  metadata.gz: 3a7364a75d69846ada1320bc0e8954b72695216d2807ec39ab1694ac4d120d97806addb9ceed8b9da8c1eb06ae28740cf6d2c118cb76a05e15ca993bf81ad470
+  data.tar.gz: 6281527f6b312d398a96b9f1df9be55de991f6ac939417131633b24ddaa24741c3b86c3b2e4e73408ca1350e0b17addf9439ae30dff998bc6dda07e8c4459db7

data/.github/workflows/build.yml

@@ -7,17 +7,3 @@ on:
 jobs:
   ci:
     uses: ./.github/workflows/ci.yml
-  build:
-    needs: ci
-    runs-on: ubuntu-latest
-    env:
-      SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }}
-    steps:
-    - name: Notify Slack
-      uses: adamkdean/simple-slack-notify@1.0.4
-      with:
-        channel: '#ops'
-        username: 'GitHub Actions'
-        color: 'good'
-        text: 'A new version of the bolognese gem is ready for release.'
-

data/.github/workflows/release.yml

@@ -9,8 +9,6 @@ jobs:
   build:
     needs: ci
     runs-on: ubuntu-latest
-    env:
-      SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }}
     steps:
     - uses: actions/checkout@v3
     - name: Set up Ruby 2.6
@@ -32,11 +30,3 @@ jobs:
         gem push *.gem
       env:
         GEM_HOST_API_KEY: "${{secrets.RUBYGEMS_AUTH_TOKEN}}"
-
-    - name: Notify Slack
-      uses: adamkdean/simple-slack-notify@1.0.4
-      with:
-        channel: '#ops'
-        username: 'GitHub Actions'
-        color: 'good'
-        text: 'A new version of the bolognese gem has been released.'

data/CHANGELOG.md

@@ -7,6 +7,7 @@
 **Merged pull requests:**
 
 - Add DFG-to-FOS mappings utility function [\#147](https://github.com/datacite/bolognese/pull/147) ([jrhoads](https://github.com/jrhoads))
+- Test against multiple versions of ruby \(2.6, 2.7, 3.0, and 3.1\) [\#141](https://github.com/datacite/bolognese/pull/141) ([jrhoads](https://github.com/jrhoads))
 
 ## [1.9.18](https://github.com/datacite/bolognese/tree/1.9.18) (2022-10-11)
 
@@ -64,6 +65,7 @@
 **Merged pull requests:**
 
 - update nokogiri to latest, including updating maremma [\#130](https://github.com/datacite/bolognese/pull/130) ([orangewolf](https://github.com/orangewolf))
+- mapping for DFG-to-OECD [\#125](https://github.com/datacite/bolognese/pull/125) ([kjgarza](https://github.com/kjgarza))
 
 ## [1.9.12](https://github.com/datacite/bolognese/tree/1.9.12) (2021-08-19)
 
@@ -79,7 +81,6 @@
 
 **Merged pull requests:**
 
-- Test against multiple versions of ruby \(2.6, 2.7, 3.0, and 3.1\) [\#141](https://github.com/datacite/bolognese/pull/141) ([jrhoads](https://github.com/jrhoads))
 - Allow Orcid URLs to have been issued from the sandbox [\#121](https://github.com/datacite/bolognese/pull/121) ([prdanelli](https://github.com/prdanelli))
 
 ## [1.9.10](https://github.com/datacite/bolognese/tree/1.9.10) (2021-07-20)
@@ -115,7 +116,6 @@
 
 **Merged pull requests:**
 
-- mapping for DFG-to-OECD [\#125](https://github.com/datacite/bolognese/pull/125) ([kjgarza](https://github.com/kjgarza))
 - Fix missing affiliation identifier [\#117](https://github.com/datacite/bolognese/pull/117) ([richardhallett](https://github.com/richardhallett))
 - Add support for book title and additional metadata when reading Crossref Book Chapter DOI [\#115](https://github.com/datacite/bolognese/pull/115) ([prdanelli](https://github.com/prdanelli))
 - Valid Funder DOIs are being rejected for not starting with a 5 [\#114](https://github.com/datacite/bolognese/pull/114) ([prdanelli](https://github.com/prdanelli))

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (1.10.0)
+    bolognese (2.0.0)
       activesupport (>= 4.2.5)
       benchmark_methods (~> 0.7)
       bibtex-ruby (>= 5.1.0)
@@ -11,7 +11,6 @@ PATH
       concurrent-ruby (~> 1.1, >= 1.1.5)
       csl-styles (~> 1.0, >= 1.0.1.10)
       edtf (~> 3.0, >= 3.0.4)
-      faraday (~> 0.17.3)
       gender_detector (~> 0.1.2)
       iso8601 (~> 0.9.1)
       json-ld-preloaded (~> 3.1, >= 3.1.3)
@@ -30,26 +29,26 @@ PATH
 GEM
   remote: https://rubygems.org/
   specs:
-    activesupport (6.1.7)
+    activesupport (6.1.7.3)
       concurrent-ruby (~> 1.0, >= 1.0.2)
       i18n (>= 1.6, < 2)
       minitest (>= 5.1)
       tzinfo (~> 2.0)
       zeitwerk (~> 2.3)
-    addressable (2.8.0)
-      public_suffix (>= 2.0.2, < 5.0)
-    amazing_print (1.4.0)
+    addressable (2.8.1)
+      public_suffix (>= 2.0.2, < 6.0)
     benchmark_methods (0.7)
     bibtex-ruby (6.0.0)
       latex-decode (~> 0.0)
     builder (3.2.4)
+    byebug (11.1.3)
     citeproc (1.0.10)
       namae (~> 1.0)
     citeproc-ruby (1.1.14)
       citeproc (~> 1.0, >= 1.0.9)
       csl (~> 1.6)
     colorize (0.8.1)
-    concurrent-ruby (1.1.10)
+    concurrent-ruby (1.2.2)
     crack (0.4.5)
       rexml
     crass (1.0.6)
@@ -60,22 +59,31 @@ GEM
       csl (~> 1.0)
     diff-lcs (1.5.0)
     docile (1.4.0)
-    ebnf (2.3.1)
-      amazing_print (~> 1.4)
+    ebnf (2.3.3)
       htmlentities (~> 4.3)
       rdf (~> 3.2)
       scanf (~> 1.0)
       sxp (~> 1.2)
-      unicode-types (~> 1.7)
-    edtf (3.1.0)
+      unicode-types (~> 1.8)
+    edtf (3.1.1)
       activesupport (>= 3.0, < 8.0)
     excon (0.71.1)
-    faraday (0.17.5)
-      multipart-post (>= 1.2, < 3)
+    faraday (2.7.6)
+      faraday-net_http (>= 2.0, < 3.1)
+      ruby2_keywords (>= 0.0.4)
     faraday-encoding (0.0.5)
       faraday
-    faraday_middleware (0.14.0)
-      faraday (>= 0.7.4, < 1.0)
+    faraday-excon (2.1.0)
+      excon (>= 0.27.4)
+      faraday (~> 2.0)
+    faraday-follow_redirects (0.3.0)
+      faraday (>= 1, < 3)
+    faraday-gzip (0.1.0)
+      faraday (>= 1.0)
+      zlib (~> 2.1)
+    faraday-multipart (1.0.4)
+      multipart-post (~> 2)
+    faraday-net_http (3.0.2)
     gender_detector (0.1.2)
       unicode_utils (>= 1.3.0)
     haml (5.2.2)
@@ -83,19 +91,19 @@ GEM
       tilt
     hashdiff (1.0.1)
     htmlentities (4.3.4)
-    i18n (1.12.0)
+    i18n (1.14.1)
       concurrent-ruby (~> 1.0)
     iso8601 (0.9.1)
     json (2.6.2)
-    json-canonicalization (0.3.0)
-    json-ld (3.2.3)
+    json-canonicalization (0.3.2)
+    json-ld (3.2.5)
       htmlentities (~> 4.3)
-      json-canonicalization (~> 0.3)
+      json-canonicalization (~> 0.3, >= 0.3.2)
       link_header (~> 0.0, >= 0.0.8)
       multi_json (~> 1.15)
-      rack (~> 2.2)
-      rdf (~> 3.2, >= 3.2.9)
-    json-ld-preloaded (3.2.1)
+      rack (>= 2.2, < 4)
+      rdf (~> 3.2, >= 3.2.10)
+    json-ld-preloaded (3.2.2)
       json-ld (~> 3.2)
       rdf (~> 3.2)
     jsonlint (0.3.0)
@@ -103,30 +111,33 @@ GEM
       optimist (~> 3)
     latex-decode (0.4.0)
     link_header (0.0.8)
-    loofah (2.19.0)
+    loofah (2.21.3)
       crass (~> 1.0.2)
-      nokogiri (>= 1.5.9)
-    maremma (4.9.8)
+      nokogiri (>= 1.12.0)
+    maremma (4.9.9)
       activesupport (>= 4.2.5)
       addressable (>= 2.3.6)
       builder (~> 3.2, >= 3.2.2)
       excon (~> 0.71.0)
-      faraday (~> 0.17.3)
-      faraday-encoding (~> 0.0.4)
-      faraday_middleware (~> 0.14.0)
-      nokogiri (>= 1.11.2, < 1.14.0)
+      faraday (>= 2.0)
+      faraday-encoding (~> 0.0.5)
+      faraday-excon (~> 2.1.0)
+      faraday-follow_redirects (~> 0.3.0)
+      faraday-gzip (~> 0.1.0)
+      faraday-multipart (~> 1.0.4)
+      nokogiri (>= 1.13.1, < 1.14.0)
       oj (>= 2.8.3)
       oj_mimic_json (~> 1.0, >= 1.0.1)
     matrix (0.4.2)
     mini_portile2 (2.8.0)
-    minitest (5.16.3)
+    minitest (5.18.0)
     multi_json (1.15.0)
-    multipart-post (2.2.3)
+    multipart-post (2.3.0)
     namae (1.1.1)
-    nokogiri (1.13.6)
+    nokogiri (1.13.9)
       mini_portile2 (~> 2.8.0)
       racc (~> 1.4)
-    oj (3.13.21)
+    oj (3.14.2)
     oj_mimic_json (1.0.1)
     optimist (3.0.1)
     postrank-uri (1.0.24)
@@ -135,22 +146,23 @@ GEM
       public_suffix (>= 2.0.0, < 2.1)
     public_suffix (2.0.5)
     racc (1.6.0)
-    rack (2.2.3.1)
+    rack (2.2.4)
     rack-test (0.8.3)
       rack (>= 1.0, < 3)
     rake (12.3.3)
-    rdf (3.2.9)
+    rdf (3.2.11)
       link_header (~> 0.0, >= 0.0.8)
     rdf-aggregate-repo (3.2.1)
       rdf (~> 3.2)
-    rdf-rdfa (3.2.0)
-      haml (~> 5.2)
+    rdf-rdfa (3.2.2)
+      haml (>= 5.2, < 7)
       htmlentities (~> 4.3)
       rdf (~> 3.2)
       rdf-aggregate-repo (~> 3.2)
       rdf-vocab (~> 3.2)
       rdf-xsd (~> 3.2)
-    rdf-rdfxml (3.2.0)
+    rdf-rdfxml (3.2.1)
+      haml (~> 5.2)
       htmlentities (~> 4.3)
       rdf (~> 3.2)
       rdf-rdfa (~> 3.2)
@@ -158,7 +170,7 @@ GEM
     rdf-turtle (3.2.1)
       ebnf (~> 2.3)
       rdf (~> 3.2)
-    rdf-vocab (3.2.2)
+    rdf-vocab (3.2.6)
       rdf (~> 3.2, >= 3.2.4)
     rdf-xsd (3.2.1)
       rdf (~> 3.2)
@@ -170,38 +182,40 @@ GEM
       rspec-mocks (~> 3.11.0)
     rspec-core (3.11.0)
       rspec-support (~> 3.11.0)
-    rspec-expectations (3.11.0)
+    rspec-expectations (3.11.1)
       diff-lcs (>= 1.2.0, < 2.0)
       rspec-support (~> 3.11.0)
     rspec-mocks (3.11.1)
       diff-lcs (>= 1.2.0, < 2.0)
       rspec-support (~> 3.11.0)
-    rspec-support (3.11.0)
+    rspec-support (3.11.1)
     rspec-xsd (0.1.0)
       nokogiri (~> 1.6)
       rspec (~> 3)
+    ruby2_keywords (0.0.5)
     scanf (1.0.0)
     simplecov (0.17.1)
       docile (~> 1.1)
       json (>= 1.8, < 3)
       simplecov-html (~> 0.10.0)
     simplecov-html (0.10.2)
-    sxp (1.2.2)
-      matrix
+    sxp (1.2.4)
+      matrix (~> 0.4)
       rdf (~> 3.2)
-    temple (0.8.2)
-    thor (1.2.1)
-    tilt (2.0.11)
-    tzinfo (2.0.5)
+    temple (0.10.2)
+    thor (1.2.2)
+    tilt (2.2.0)
+    tzinfo (2.0.6)
       concurrent-ruby (~> 1.0)
     unicode-types (1.8.0)
     unicode_utils (1.4.0)
     vcr (3.0.3)
-    webmock (3.14.0)
+    webmock (3.18.1)
       addressable (>= 2.8.0)
       crack (>= 0.3.2)
       hashdiff (>= 0.4.0, < 2.0.0)
-    zeitwerk (2.6.1)
+    zeitwerk (2.6.8)
+    zlib (2.1.1)
 
 PLATFORMS
   ruby
@@ -209,6 +223,7 @@ PLATFORMS
 DEPENDENCIES
   bolognese!
   bundler (>= 1.0)
+  byebug
   hashdiff (>= 1.0.0.beta1, < 2.0.0)
   rack-test (~> 0)
   rake (~> 12.0)
@@ -219,4 +234,4 @@ DEPENDENCIES
   webmock (~> 3.0, >= 3.0.1)
 
 BUNDLED WITH
-   2.3.12
+   2.4.20

data/bolognese.gemspec

@@ -17,7 +17,7 @@ Gem::Specification.new do |s|
 
   # Declare dependencies here, rather than in the Gemfile
   s.add_dependency 'maremma', '>= 4.9.4', '< 5'
-  s.add_dependency 'faraday', '~> 0.17.3'
+  #s.add_dependency 'faraday', '~> 0.17.3'
   s.add_dependency 'nokogiri', '>= 1.13.2', '< 1.14'
   s.add_dependency 'loofah', '~> 2.0', '>= 2.0.3'
   s.add_dependency 'builder', '~> 3.2', '>= 3.2.2'
@@ -49,6 +49,7 @@ Gem::Specification.new do |s|
   s.add_development_dependency 'webmock', '~> 3.0', '>= 3.0.1'
   s.add_development_dependency 'simplecov', '0.17.1'
   s.add_development_dependency 'hashdiff', ['>= 1.0.0.beta1', '< 2.0.0']
+  s.add_development_dependency 'byebug'
 
   s.require_paths = ["lib"]
   s.files       = `git ls-files`.split($/)

data/lib/bolognese/author_utils.rb

@@ -30,6 +30,7 @@ module Bolognese
       name_type = parse_attributes(author.fetch("creatorName", nil), content: "nameType", first: true) || parse_attributes(author.fetch("contributorName", nil), content: "nameType", first: true)
 
       name_identifiers = Array.wrap(author.fetch("nameIdentifier", nil)).map do |ni|
+        ni["__content__"] = ni["__content__"].strip
         if ni["nameIdentifierScheme"] == "ORCID"
           {
             "nameIdentifier" => normalize_orcid(ni["__content__"]),
@@ -144,16 +145,28 @@ module Bolognese
           affiliation_identifier_scheme = nil
           scheme_uri = nil
         else
+          scheme_uri = a["schemeURI"]
           if a["affiliationIdentifier"].present?
             affiliation_identifier = a["affiliationIdentifier"]
             if a["schemeURI"].present?
               schemeURI = a["schemeURI"].end_with?("/") ? a["schemeURI"] : a["schemeURI"] + "/"
             end
             affiliation_identifier = !affiliation_identifier.to_s.start_with?("https://") && schemeURI.present? ? normalize_id(schemeURI + affiliation_identifier) : normalize_id(affiliation_identifier)
+            # The normalize_id(affiliation_identifier) method currently discards affiliation identifiers that don't start with a URL, 
+            # for example: affiliation_identifier = "05bp8ka05".
+            # To address this issue, we are introducing the following change to handle such affiliation identifiers.
+            # when `normalize_id` method could not normalize, it returns nil, hence we have following condition
+            if affiliation_identifier.nil?
+              if a["affiliationIdentifierScheme"] == "ROR"
+                scheme_uri = "https://ror.org"
+                affiliation_identifier = normalize_ror(a["affiliationIdentifier"])
+              else
+                affiliation_identifier = a["affiliationIdentifier"]
+              end
+            end
           end
           name = a["__content__"].to_s.squish.presence
           affiliation_identifier_scheme = a["affiliationIdentifierScheme"]
-          scheme_uri = a["SchemeURI"]
         end
 
         { "name" => name,

data/lib/bolognese/datacite_utils.rb

@@ -106,9 +106,19 @@ module Bolognese
         end
       end
     end
-
+    
     def insert_publisher(xml)
-      xml.publisher(publisher || container && container["title"])
+      if publisher.is_a?(Hash)
+        attributes = {
+          'publisherIdentifier' => publisher["publisherIdentifier"],
+          'publisherIdentifierScheme' => publisher["publisherIdentifierScheme"],
+          'schemeURI' => publisher["schemeUri"],
+          "xml:lang" => publisher["lang"]
+        }.compact
+        xml.publisher(publisher["name"] || container && container["title"], attributes)
+      else
+        xml.publisher(publisher || container && container["title"])
+      end
     end
 
     def insert_publication_year(xml)
@@ -171,7 +181,7 @@ module Bolognese
             s["subject"] = subject
           end
 
-          attributes = { "subjectScheme" => s["subjectScheme"], "schemeURI" => s["schemeUri"], "valueURI" => s["valueUri"], "xml:lang" => s["lang"] }.compact
+          attributes = { "subjectScheme" => s["subjectScheme"], "schemeURI" => s["schemeUri"], "valueURI" => s["valueUri"], "classificationCode" => s["classificationCode"], "xml:lang" => s["lang"] }.compact
 
           xml.subject(s["subject"], attributes)
         end
@@ -375,13 +385,11 @@ module Bolognese
               end
             end
             if geo_location["geoLocationPolygon"]
-              geo_location["geoLocationPolygon"].each do |geo_location_polygon|
-                xml.geoLocationPolygon do
-                  geo_location_polygon.each do |polygon_point|
-                    xml.polygonPoint do
-                      xml.pointLatitude(polygon_point.dig("polygonPoint", "pointLatitude"))
-                      xml.pointLongitude(polygon_point.dig("polygonPoint", "pointLongitude"))
-                    end
+              xml.geoLocationPolygon do
+                Array.wrap(geo_location["geoLocationPolygon"]).each do |polygon_point|
+                  xml.polygonPoint do
+                    xml.pointLatitude(polygon_point.dig("polygonPoint", "pointLatitude"))
+                    xml.pointLongitude(polygon_point.dig("polygonPoint", "pointLongitude"))
                   end
                 end
               end

data/lib/bolognese/metadata.rb

@@ -217,7 +217,7 @@ module Bolognese
     end
 
     def publisher
-      @publisher ||= meta.fetch("publisher", nil)
+      @publisher ||= normalize_publisher(meta["publisher"]) if meta.fetch("publisher", nil).present?
     end
 
     def identifiers

data/lib/bolognese/metadata_utils.rb

@@ -161,7 +161,7 @@ module Bolognese
         "volume" => container.to_h["volume"],
         "issue" => container.to_h["issue"],
         "page" => page,
-        "publisher" => publisher,
+        "publisher" => publisher["name"],
         "title" => parse_attributes(titles, content: "title", first: true),
         "URL" => url,
         "copyright" => Array.wrap(rights_list).map { |l| l["rights"] }.first,

data/lib/bolognese/readers/bibtex_reader.rb

@@ -86,11 +86,11 @@ module Bolognese
           "titles" => meta.try(:title).present? ? [{ "title" => meta.try(:title).to_s }] : [],
           "creators" => creators,
           "container" => container,
-          "publisher" => meta.try(:publisher).to_s.presence,
+          "publisher" => meta.try(:publisher).present? ? { "name" => meta.publisher.to_s } : nil,
           "related_identifiers" => related_identifiers,
           "dates" => dates,
           "publication_year" => publication_year,
-          "descriptions" => meta.try(:abstract).present? ? [{ "description" => meta.try(:abstract) && sanitize(meta.abstract.to_s).presence, "descriptionType" => "Abstract" }] : [],
+          "descriptions" => meta.try(:abstract).present? ? [{ "description" => meta.try(:abstract) && sanitize(meta.abstract.to_s, new_line: true).presence, "descriptionType" => "Abstract" }] : [],
           "rights_list" => rights_list,
           "state" => state
         }.merge(read_options)

data/lib/bolognese/readers/citeproc_reader.rb

@@ -107,7 +107,7 @@ module Bolognese
           "related_identifiers" => related_identifiers,
           "dates" => dates,
           "publication_year" => publication_year,
-          "descriptions" => meta.fetch("abstract", nil).present? ? [{ "description" => sanitize(meta.fetch("abstract")), "descriptionType" => "Abstract" }] : [],
+          "descriptions" => meta.fetch("abstract", nil).present? ? [{ "description" => sanitize(meta.fetch("abstract"), new_line: true), "descriptionType" => "Abstract" }] : [],
           "rights_list" => rights_list,
           "version_info" => meta.fetch("version", nil),
           "subjects" => subjects,

data/lib/bolognese/readers/codemeta_reader.rb

@@ -43,7 +43,7 @@ module Bolognese
         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?
         publication_year = meta.fetch("datePublished")[0..3] if meta.fetch("datePublished", nil).present?
-        publisher = meta.fetch("publisher", nil)
+        publisher = { "name" => meta.fetch("publisher", nil) } if meta.fetch("publisher", nil).present?
         state = meta.present? || read_options.present? ? "findable" : "not_found"
         schema_org = meta.fetch("@type", nil)
         types = {
@@ -76,7 +76,7 @@ module Bolognese
           #{}"is_part_of" => is_part_of,
           "dates" => dates,
           "publication_year" => publication_year,
-          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description")), "descriptionType" => "Abstract" }] : nil,
+          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description"), new_line: true), "descriptionType" => "Abstract" }] : nil,
           "rights_list" => rights_list,
           "version_info" => meta.fetch("version", nil),
           "subjects" => subjects,

data/lib/bolognese/readers/crosscite_reader.rb

@@ -7,7 +7,10 @@ module Bolognese
         errors = jsonlint(string)
         return { "errors" => errors } if errors.present?
 
-        string.present? ? Maremma.from_json(string) : {}
+        crosscite = string.present? ? Maremma.from_json(string) : {}
+        crosscite["publisher"] = normalize_publisher(crosscite["publisher"]) if crosscite.fetch("publisher", nil).present?
+
+        crosscite
       end
     end
   end

data/lib/bolognese/readers/crossref_reader.rb

@@ -40,8 +40,7 @@ module Bolognese
         journal_metadata = nil
         journal_issue = {}
         journal_metadata = nil
-        publisher = query.dig("crm_item", 0)
-        publisher = nil unless publisher.is_a?(String)
+        publisher = query.dig("crm_item", 0).is_a?(String) ? { "name" => query.dig("crm_item", 0) } : nil
 
         case model
         when "book"
@@ -232,11 +231,11 @@ module Bolognese
 
       def crossref_description(bibliographic_metadata)
         abstract = Array.wrap(bibliographic_metadata.dig("abstract")).map do |r|
-          { "descriptionType" => "Abstract", "description" => sanitize(parse_attributes(r, content: 'p')) }.compact
+          { "descriptionType" => "Abstract", "description" => sanitize(parse_attributes(r, content: 'p'), new_line: true) }.compact
         end
 
         description = Array.wrap(bibliographic_metadata.dig("description")).map do |r|
-          { "descriptionType" => "Other", "description" => sanitize(parse_attributes(r)) }.compact
+          { "descriptionType" => "Other", "description" => sanitize(parse_attributes(r), new_line: true) }.compact
         end
 
         (abstract + description)

data/lib/bolognese/readers/datacite_json_reader.rb

@@ -7,7 +7,10 @@ module Bolognese
         errors = jsonlint(string)
         return { "errors" => errors } if errors.present?
 
-        string.present? ? Maremma.from_json(string).transform_keys! { |key| key.underscore } : {}
+        datacite_json = string.present? ? Maremma.from_json(string).transform_keys! { |key| key.underscore } : {}
+        datacite_json["publisher"] = normalize_publisher(datacite_json["publisher"]) if datacite_json.fetch("publisher", nil).present?
+
+        datacite_json
       end
     end
   end

data/lib/bolognese/readers/datacite_reader.rb

@@ -94,13 +94,29 @@ module Bolognese
 
         titles = get_titles(meta)
 
+        publisher = Array.wrap(meta.dig("publisher")).map do |r|
+          if r.blank?
+            nil
+          elsif r.is_a?(String)
+            { "name" => r.strip }
+          elsif r.is_a?(Hash)
+            {
+              "name" => r["__content__"].strip,
+              "publisherIdentifier" => r["publisherIdentifierScheme"] == "ROR" ? normalize_ror(r["publisherIdentifier"]) : r["publisherIdentifier"],
+              "publisherIdentifierScheme" => r["publisherIdentifierScheme"],
+              "schemeUri" => r["schemeURI"],
+              "lang" => r["lang"],
+            }.compact
+          end
+        end.compact.first
+
         descriptions = Array.wrap(meta.dig("descriptions", "description")).map do |r|
           if r.blank?
             nil
           elsif r.is_a?(String)
-            { "description" => sanitize(r), "descriptionType" => "Abstract" }
+            { "description" => sanitize(r, new_line: true), "descriptionType" => "Abstract" }
           elsif r.is_a?(Hash)
-            { "description" => sanitize(r["__content__"]), "descriptionType" => r["descriptionType"], "lang" => r["lang"] }.compact
+            { "description" => sanitize(r["__content__"], new_line: true), "descriptionType" => r["descriptionType"], "lang" => r["lang"] }.compact
           end
         end.compact
         rights_list = Array.wrap(meta.dig("rightsList", "rights")).map do |r|
@@ -230,14 +246,14 @@ module Bolognese
             "creators" => get_authors(Array.wrap(ri.dig("creators", "creator"))),
             "titles" => get_titles(ri),
             "publicationYear" => ri["publicationYear"],
-            "volume" => ri["volume"],
-            "issue" => ri["issue"],
+            "volume" => parse_attributes(ri["volume"]),
+            "issue" => parse_attributes(ri["issue"]),
             "number" => number,
             "numberType" => numberType,
-            "firstPage" => ri["firstPage"],
-            "lastPage" => ri["lastPage"],
-            "publisher" => ri["publisher"],
-            "edition" => ri["edition"],
+            "firstPage" => parse_attributes(ri["firstPage"]),
+            "lastPage" => parse_attributes(ri["lastPage"]),
+            "publisher" => parse_attributes(ri["publisher"]),
+            "edition" => parse_attributes(ri["edition"]),
             "contributors" => get_authors(Array.wrap(ri.dig("contributors", "contributor"))),
           }.compact
         end
@@ -287,7 +303,7 @@ module Bolognese
           "creators" => get_authors(Array.wrap(meta.dig("creators", "creator"))),
           "contributors" => get_authors(Array.wrap(meta.dig("contributors", "contributor"))),
           "container" => set_container(meta),
-          "publisher" => parse_attributes(meta.fetch("publisher", nil), first: true).to_s.strip.presence,
+          "publisher" => publisher,
           "agency" => "datacite",
           "funding_references" => funding_references,
           "dates" => dates,

data/lib/bolognese/readers/npm_reader.rb

@@ -103,7 +103,7 @@ module Bolognese
           #"related_identifiers" => related_identifiers,
           #"dates" => dates,
           #"publication_year" => publication_year,
-          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description")), "descriptionType" => "Abstract" }] : [],
+          "descriptions" => meta.fetch("description", nil).present? ? [{ "description" => sanitize(meta.fetch("description"), new_line: true), "descriptionType" => "Abstract" }] : [],
           "rights_list" => rights_list,
           "version_info" => meta.fetch("version", nil),
           "subjects" => subjects