commonmeta-ruby 3.4.5 → 3.5.1

data/lib/commonmeta/writers/codemeta_writer.rb

@@ -5,7 +5,9 @@ module Commonmeta
     module CodemetaWriter
       def codemeta
         return nil unless valid? || show_errors
-
+        
+        authors = Array.wrap(contributors).select { |c| c['contributorRoles'] == ['Author'] }
+        
         hsh = {
           '@context' => id.present? ? 'https://raw.githubusercontent.com/codemeta/codemeta/master/codemeta.jsonld' : nil,
           '@type' => Commonmeta::Utils::CM_TO_SO_TRANSLATIONS.fetch(type, 'SoftwareSourceCode'),
@@ -13,7 +15,7 @@ module Commonmeta
           'identifier' => to_schema_org_identifiers(alternate_identifiers),
           'codeRepository' => url,
           'name' => parse_attributes(titles, content: 'title', first: true),
-          'authors' => creators,
+          'authors' => to_schema_org(authors),
           'description' => parse_attributes(descriptions, content: 'description', first: true),
           'version' => version,
           'tags' => if subjects.present?

data/lib/commonmeta/writers/csv_writer.rb

@@ -7,7 +7,9 @@ module Commonmeta
 
       def csv
         return nil unless valid?
-
+        
+        authors = contributors.select { |c| c['contributorRoles'] == ['Author'] }
+ 
         bib = {
           doi: doi_from_url(id),
           url: url,
@@ -15,7 +17,7 @@ module Commonmeta
           state: state,
           type: Commonmeta::Utils::CM_TO_BIB_TRANSLATIONS.fetch(type, 'misc'),
           title: parse_attributes(titles, content: 'title', first: true),
-          author: authors_as_string(creators),
+          author: authors_as_string(authors),
           publisher: publisher['name']
         }.values
 

data/lib/commonmeta/writers/datacite_writer.rb

@@ -22,7 +22,7 @@ module Commonmeta
           'doi' => doi_from_url(id),
           'url' => url,
           'types' => types,
-          'creators' => Array.wrap(creators).map { |c| datacite_contributor(c) },
+          'creators' => Array.wrap(contributors).map { |c| datacite_contributor(c) },
           'titles' => titles,
           'publisher' => publisher.to_h['name'],
           'container' => container,

data/lib/commonmeta/writers/jats_writer.rb

@@ -39,10 +39,12 @@ module Commonmeta
       end
 
       def insert_authors(xml)
-        return unless creators.present?
-
+        authors = contributors.select { |c| c['contributorRoles'] == ['Author'] }
+ 
+        return unless authors.present?
+        
         xml.send(:'person-group', 'person-group-type' => 'author') do
-          Array.wrap(creators).each do |au|
+          Array.wrap(authors).each do |au|
             xml.name do
               insert_contributor(xml, au)
             end
@@ -51,10 +53,12 @@ module Commonmeta
       end
 
       def insert_editors(xml)
-        return unless contributors.present?
+        editors = contributors.select { |c| c['contributorRoles'] == ['Editor'] }
+ 
+        return unless editors.present?
 
         xml.send(:'person-group', 'person-group-type' => 'editor') do
-          Array.wrap(contributors).each do |con|
+          Array.wrap(editors).each do |con|
             xml.name do
               insert_contributor(xml, con)
             end

data/lib/commonmeta/writers/ris_writer.rb

@@ -4,13 +4,14 @@ module Commonmeta
   module Writers
     module RisWriter
       def ris
+        authors = contributors.select { |c| c['contributorRoles'] == ['Author'] }
         sn = container.to_h['identifier']
         sn = sn.downcase if sn.present? && container.to_h['identifierType'] == 'DOI'
         {
           'TY' => Commonmeta::Utils::CM_TO_RIS_TRANSLATIONS.fetch(type, 'GEN'),
           'T1' => parse_attributes(titles, content: 'title', first: true),
           'T2' => container.to_h['title'],
-          'AU' => to_ris(creators),
+          'AU' => to_ris(authors),
           'DO' => doi_from_url(id),
           'UR' => url,
           'AB' => parse_attributes(descriptions, content: 'description', first: true),

data/lib/commonmeta/writers/schema_org_writer.rb

@@ -2,8 +2,11 @@
 
 module Commonmeta
   module Writers
-    module SchemaOrgWriter
+    module SchemaOrgWriter 
       def schema_hsh
+        authors = contributors.select { |c| c['contributorRoles'] == ['Author'] }
+        editors = contributors.select { |c| c['contributorRoles'] == ['Editor'] }
+
         { '@context' => 'http://schema.org',
           '@type' => Commonmeta::Utils::CM_TO_SO_TRANSLATIONS.fetch(type, 'CreativeWork'),
           '@id' => id,
@@ -11,8 +14,8 @@ module Commonmeta
           'url' => url,
           'additionalType' => additional_type,
           'name' => parse_attributes(titles, content: 'title', first: true),
-          'author' => to_schema_org(creators),
-          'editor' => to_schema_org(contributors),
+          'author' => to_schema_org(authors),
+          'editor' => to_schema_org(editors),
           'description' => parse_attributes(descriptions, content: 'description', first: true),
           'license' => license.to_h['url'],
           'version' => version,
@@ -32,7 +35,7 @@ module Commonmeta
           'spatialCoverage' => to_schema_org_spatial_coverage(geo_locations),
           'citation' => Array.wrap(references).map { |r| to_schema_org_citation(r) },
           '@reverse' => reverse.presence,
-          'contentUrl' => Array.wrap(content_url).unwrap,
+          'contentUrl' => Array.wrap(files).map { |f| f["url"] }.unwrap,
           'schemaVersion' => schema_version,
           'periodical' => if type == 'Dataset'
                             nil

data/resources/{commonmeta_v0.9.3.json → commonmeta_v0.10.3.json}

@@ -1,7 +1,7 @@
 {
   "$schema": "http://json-schema.org/draft-07/schema#",
-  "$id": "https://commonmeta.org/commonmeta_v0.9.3.json",
-  "title": "Commonmeta v0.9.3",
+  "$id": "https://commonmeta.org/commonmeta_v0.10.3.json",
+  "title": "Commonmeta v0.10.3",
   "description": "JSON representation of the Commonmeta schema.",
   "additionalProperties": false,
   "definitions": {
@@ -24,6 +24,53 @@
       },
       "uniqueItems": true
     },
+    "contributorRole": {
+      "description": "The type of contribution made by a contributor",
+      "enum": [
+        "Author",
+        "Editor",
+        "Chair",
+        "Reviewer",
+        "ReviewAssistant",
+        "StatsReviewer",
+        "ReviewerExternal",
+        "Reader",
+        "Translator",
+        "ContactPerson",
+        "DataManager",
+        "Distributor",
+        "HostingInstitution",
+        "Producer",
+        "ProjectLeader",
+        "ProjectManager",
+        "ProjectMember",
+        "RegistrationAgency",
+        "RegistrationAuthority",
+        "RelatedPerson",
+        "ResearchGroup",
+        "RightsHolder",
+        "Researcher",
+        "Sponsor",
+        "WorkPackageLeader",
+        "Conceptualization",
+        "DataCuration",
+        "FormalAnalysis",
+        "FundingAcquisition",
+        "Investigation",
+        "Methodology",
+        "ProjectAdministration",
+        "Resources",
+        "Software",
+        "Supervision",
+        "Validation",
+        "Visualization",
+        "WritingOriginalDraft",
+        "WritingReviewEditing",
+        "Maintainer",
+        "Other"
+      ],
+      "type": "string"
+    },
     "latitude": {
       "type": "number",
       "minimum": -90,
@@ -91,36 +138,52 @@
       "type": "string",
       "format": "uri"
     },
-    "creators": {
-      "description": "The creators of the resource.",
+    "contributors": {
+      "description": "The contributors to the resource.",
       "type": "array",
       "items": {
         "type": "object",
         "properties": {
           "id": {
-            "description": "The unique identifier for the creator.",
-            "type": "string",
-            "format": "uri"
+            "description": "The unique identifier for the contributor.",
+            "type": "string"
           },
           "type": {
-            "description": "The type of the creator.",
+            "description": "The type of the contributor.",
             "type": "string",
             "enum": ["Organization", "Person"]
           },
+          "contributorRoles": {
+            "description": "List of roles assumed by the contributor when working on the resource.",
+            "items": {
+              "$ref": "#/definitions/contributorRole"
+            },
+            "type": "array",
+            "uniqueItems": true
+          },
           "name": {
-            "description": "The name of the creator.",
+            "description": "The name of the contributor.",
             "type": "string"
           },
           "givenName": {
-            "description": "The given name of the creator.",
+            "description": "The given name of the contributor.",
             "type": "string"
           },
           "familyName": {
-            "description": "The family name of the creator.",
+            "description": "The family name of the contributor.",
             "type": "string"
           },
           "affiliation": { "$ref": "#/definitions/affiliations" }
-        }
+        },
+        "anyOf": [
+          {
+            "required": ["familyName"]
+          },
+          {
+            "required": ["name"]
+          }
+        ],
+        "required": ["type", "contributorRoles"]
       },
       "minItems": 1,
       "uniqueItems": true
@@ -142,8 +205,38 @@
     },
     "date": {
       "description": "The dates for the resource.",
+      "$comment": "The date fields are not required. Dates should be formatted as ISO 8601 dates.",
       "type": "object",
-      "format": "date"
+      "properties": {
+        "created": {
+          "description": "The date the resource was created.",
+          "type": "string"
+        },
+        "submitted": {
+          "description": "The date the resource was submitted.",
+          "type": "string"
+        },
+        "accepted": {
+          "description": "The date the resource was accepted.",
+          "type": "string"
+        },
+        "published": {
+          "description": "The date the resource was published.",
+          "type": "string"
+        },
+        "updated": {
+          "description": "The date the resource was updated.",
+          "type": "string"
+        },
+        "available": {
+          "description": "The date the resource was made available.",
+          "type": "string"
+        },
+        "withdrawn": {
+          "description": "The date the resource was withdrawn.",
+          "type": "string"
+        }
+      }
     },
     "titles": {
       "description": "The titles of the resource.",
@@ -178,9 +271,11 @@
           "enum": [
             "Book",
             "BookSeries",
+            "DataCatalog",
             "Journal",
             "Periodical",
             "ProceedingsSeries",
+            "Repository",
             "Series"
           ]
         },
@@ -190,37 +285,6 @@
         }
       }
     },
-    "contributors": {
-      "description": "The contributors to the resource.",
-      "type": "array",
-      "items": {
-        "type": "object",
-        "properties": {
-          "id": {
-            "description": "The unique identifier for the contributor.",
-            "type": "string"
-          },
-          "type": {
-            "description": "The type of the contributor.",
-            "type": "string",
-            "enum": ["Organization", "Person"]
-          },
-          "name": {
-            "description": "The name of the contributor.",
-            "type": "string"
-          },
-          "givenName": {
-            "description": "The given name of the contributor.",
-            "type": "string"
-          },
-          "familyName": {
-            "description": "The family name of the contributor.",
-            "type": "string"
-          },
-          "affiliation": { "$ref": "#/definitions/affiliations" }
-        }
-      }
-    },
     "subjects": {
       "type": "array",
       "items": {
@@ -271,7 +335,7 @@
         "properties": {
           "key": { "type": "string" },
           "doi": { "type": "string" },
-          "creator": { "type": "string" },
+          "contributor": { "type": "string" },
           "title": { "type": "string" },
           "publisher": { "type": "string" },
           "publicationYear": { "type": "string" },
@@ -281,7 +345,6 @@
           "lastPage": { "type": "string" },
           "containerTitle": { "type": "string" },
           "edition": { "type": "string" },
-          "contributor": { "type": "string" },
           "unstructured": { "type": "string" }
         },
         "required": ["key"]
@@ -300,17 +363,21 @@
           "type": {
             "type": "string",
             "enum": [
-              "IsIdenticalTo",
               "IsNewVersionOf",
               "IsPreviousVersionOf",
+              "IsVersionOf",
+              "HasVersion",
               "IsPartOf",
               "HasPart",
               "IsVariantFormOf",
+              "IsOriginalFormOf",
+              "IsIdenticalTo",
               "IsTranslationOf",
-              "IsVersionOf",
-              "IsReviewOf",
+              "IsReviewedBy",
+              "Reviews",
               "IsPreprintOf",
-              "HasPreprint"
+              "HasPreprint",
+              "isSupplementTo"
             ]
           }
         },
@@ -421,12 +488,20 @@
         "required": ["alternateIdentifier", "alternateIdentifierType"]
       }
     },
-    "content_url": {
-      "description": "The URL of the content of the resource.",
+    "files": {
+      "description": "The downloadable files for the resource.",
       "type": "array",
       "items": {
-        "type": "string",
-        "format": "uri"
+        "type": "object",
+        "properties": {
+          "bucket": { "type": "string" },
+          "key": { "type": "string" },
+          "checksum": { "type": "string" },
+          "url": { "type": "string", "format": "uri" },
+          "size": { "type": "integer" },
+          "mimeType": { "type": "string" }
+        },
+        "required": ["url"]
       }
     },
     "schema_version": {
@@ -443,5 +518,13 @@
       "enum": ["findable", "not_found"]
     }
   },
-  "required": ["id", "type", "url", "creators", "titles", "publisher", "date"]
+  "required": [
+    "id",
+    "type",
+    "url",
+    "contributors",
+    "titles",
+    "publisher",
+    "date"
+  ]
 }

data/resources/csl-citation.json

@@ -0,0 +1,99 @@
+{
+  "description": "JSON schema for CSL citation objects",
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://resource.citationstyles.org/schema/latest/input/json/csl-citation.json",
+  "type": "object",
+  "properties": {
+    "schema": {
+      "type": "string",
+      "enum": [
+        "https://resource.citationstyles.org/schema/latest/input/json/csl-citation.json"
+      ]
+    },
+    "citationID": {
+      "type": ["string", "number"]
+    },
+    "citationItems": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "properties": {
+          "id": {
+            "type": ["string", "number"]
+          },
+          "itemData": {
+            "$ref": "https://resource.citationstyles.org/schema/v1.0/input/json/csl-data.json#/items"
+          },
+          "prefix": {
+            "type": "string"
+          },
+          "suffix": {
+            "type": "string"
+          },
+          "locator": {
+            "type": "string"
+          },
+          "label": {
+            "type": "string",
+            "enum": [
+              "act",
+              "appendix",
+              "article-locator",
+              "book",
+              "canon",
+              "chapter",
+              "column",
+              "elocation",
+              "equation",
+              "figure",
+              "folio",
+              "issue",
+              "line",
+              "note",
+              "opus",
+              "page",
+              "paragraph",
+              "part",
+              "rule",
+              "scene",
+              "section",
+              "sub-verbo",
+              "supplement",
+              "table",
+              "timestamp",
+              "title-locator",
+              "verse",
+              "version",
+              "volume"
+            ]
+          },
+          "suppress-author": {
+            "type": ["string", "number", "boolean"]
+          },
+          "author-only": {
+            "type": ["string", "number", "boolean"]
+          },
+          "uris": {
+            "type": "array",
+            "items": {
+              "type": "string"
+            }
+          }
+        },
+        "required": ["id"],
+        "additionalProperties": false
+      }
+    },
+    "properties": {
+      "type": "object",
+      "properties": {
+        "noteIndex": {
+          "type": "number"
+        }
+      },
+      "additionalProperties": false
+    }
+  },
+  "required": ["schema", "citationID"],
+  "additionalProperties": false
+}

data/spec/author_utils_spec.rb

@@ -87,7 +87,7 @@ describe Commonmeta::Metadata, vcr: true do
       author = { "email" => "info@ucop.edu", "name" => "University of California, Santa Barbara",
                  "role" => { "namespace" => "http://www.ngdc.noaa.gov/metadata/published/xsd/schema/resources/Codelist/gmxCodelists.xml#CI_RoleCode", "roleCode" => "copyrightHolder" }, "type" => "Organization" }
       response = subject.get_one_author(author)
-      expect(response).to eq("name" => "University of California, Santa Barbara", "type" => "Organization")
+      expect(response).to eq("name" => "University of California, Santa Barbara", "type" => "Organization", "contributorRoles" => ["Author"])
     end
 
     it "has familyName" do
@@ -97,14 +97,14 @@ describe Commonmeta::Metadata, vcr: true do
       response = subject.get_one_author(meta.dig("creators").first)
       expect(response).to eq(
         "id" => "https://orcid.org/0000-0003-1419-2405",
-        "givenName" => "Martin", "familyName" => "Fenner", "type" => "Person",
+        "givenName" => "Martin", "familyName" => "Fenner", "type" => "Person", "contributorRoles" => ["Author"],
       )
     end
 
     it "has name with title" do
       author = { "name" => "Tejas S. Sathe, MD" }
       response = subject.get_one_author(author)
-      expect(response).to eq("givenName" => "Tejas S.", "familyName" => "Sathe", "type" => "Person")
+      expect(response).to eq("givenName" => "Tejas S.", "familyName" => "Sathe", "type" => "Person", "contributorRoles" => ["Author"])
     end
 
     it "has name in display-order with ORCID" do
@@ -112,7 +112,7 @@ describe Commonmeta::Metadata, vcr: true do
       subject = described_class.new(input: input)
       meta = JSON.parse(subject.raw).dig("data", "attributes")
       response = subject.get_one_author(meta.dig("creators").first)
-      expect(response).to eq("type" => "Person",
+      expect(response).to eq("type" => "Person", "contributorRoles" => ["Author"],
                              "id" => "https://orcid.org/0000-0003-4881-1606",
                              "givenName" => "Andrea", "familyName" => "Bedini")
     end
@@ -123,43 +123,43 @@ describe Commonmeta::Metadata, vcr: true do
                  "type" => "Organization", "role" => { "namespace" => "http://www.ngdc.noaa.gov/metadata/published/xsd/schema/resources/Codelist/gmxCodelists.xml#CI_RoleCode", "roleCode" => "copyrightHolder" } }
       response = subject.get_one_author(author)
       expect(response).to eq("name" => "University of California, Santa Barbara",
-                             "type" => "Organization")
+                             "type" => "Organization", "contributorRoles" => ["Author"])
     end
 
     it "is another organization" do
       author = { "name" => "University of California, Santa Barbara",
                  "id" => "https://ror.org/02t274463" }
       response = subject.get_one_author(author)
-      expect(response).to eq("id" => "https://ror.org/02t274463", "name" => "University of California, Santa Barbara", "type" => "Organization")
+      expect(response).to eq("id" => "https://ror.org/02t274463", "name" => "University of California, Santa Barbara", "type" => "Organization", "contributorRoles" => ["Author"])
     end
 
     it "is anonymous" do
       author = { "id" => "https://ror.org/05745n787" }
       response = subject.get_one_author(author)
-      expect(response).to eq("type" => "Person", "affiliation" => [{"id"=>"https://ror.org/05745n787"}])
+      expect(response).to eq("type" => "Person", "contributorRoles" => ["Author"], "affiliation" => [{ "id" => "https://ror.org/05745n787" }])
     end
 
     it "name with affiliation crossref" do
       input = "10.7554/elife.01567"
       subject = described_class.new(input: input, from: "crossref")
-      response = subject.get_one_author(subject.creators.first)
+      response = subject.get_one_author(subject.contributors.first)
       expect(response).to eq("affiliation" => [{ "name" => "Department of Plant Molecular Biology, University of Lausanne, Lausanne, Switzerland" }], "familyName" => "Sankar",
                              "givenName" => "Martial",
-                             "type" => "Person")
+                             "type" => "Person", "contributorRoles" => ["Author"])
     end
 
     it "only familyName and givenName" do
       input = "https://doi.pangaea.de/10.1594/PANGAEA.836178"
       subject = described_class.new(input: input, from: "schema_org")
-      response = subject.get_one_author(subject.creators.first)
-      expect(response).to eq("type" => "Person", "givenName" => "Emma", "familyName" => "Johansson")
+      response = subject.get_one_author(subject.contributors.first)
+      expect(response).to eq("type" => "Person", "contributorRoles" => ["Author"], "givenName" => "Emma", "familyName" => "Johansson")
     end
 
     it "affiliation is space" do
       input = "10.1177/0042098011428175"
       subject = described_class.new(input: input)
-      response = subject.get_one_author(subject.creators.first)
-      expect(response).to eq("familyName"=>"Petrovici", "givenName"=>"Norbert", "type"=>"Person")
+      response = subject.get_one_author(subject.contributors.first)
+      expect(response).to eq("familyName" => "Petrovici", "givenName" => "Norbert", "type" => "Person", "contributorRoles" => ["Author"])
     end
   end
 
@@ -201,9 +201,9 @@ describe Commonmeta::Metadata, vcr: true do
 
   context "get_affiliations" do
     it "name" do
-      affiliation = [{ 'name' => 'University of Zurich, Zurich, Switzerland' }]
+      affiliation = [{ "name" => "University of Zurich, Zurich, Switzerland" }]
       response = subject.get_affiliations(affiliation)
-      expect(response).to eq([{ 'name' => 'University of Zurich, Zurich, Switzerland' }])
+      expect(response).to eq([{ "name" => "University of Zurich, Zurich, Switzerland" }])
     end
 
     it "name and ROR ID" do
@@ -216,7 +216,7 @@ describe Commonmeta::Metadata, vcr: true do
     it "only ROR ID" do
       affiliation = { "affiliationIdentifier" => "https://ror.org/02t274463" }
       response = subject.get_affiliations(affiliation)
-      expect(response).to eq([{"id"=>"https://ror.org/02t274463"}])
+      expect(response).to eq([{ "id" => "https://ror.org/02t274463" }])
     end
   end
 end

data/spec/cli_spec.rb

@@ -269,7 +269,7 @@ describe Commonmeta::CLI do
       it 'to datacite invalid' do
         file = fixture_path + "datacite_missing_creator.xml"
         subject.options = { to: "datacite", show_errors: "true" }
-        expect { subject.convert file }.to output(/root is missing required keys: id, type, url, creators, titles, publisher, date/).to_stderr
+        expect { subject.convert file }.to output(/root is missing required keys: id, type, url, contributors, titles, publisher, date/).to_stderr
       end
 
       it 'to datacite invalid ignore errors' do