commonmeta-ruby 3.2.14 → 3.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b549fc46c70c5962ee3c971968771a6e5f13124eb89b03aab36228afc94c01fe
-  data.tar.gz: 1545d0de4e821265cc19bb744456d025f6bc14ea52c19cbae2585a1d501501a4
+  metadata.gz: dde90e2b65c0eabd771cda34b2b5054fbe6bf80b573444fe0bcdde14c8392429
+  data.tar.gz: 487ffa495c4ffae32184c0a6e11ef68a5f9e82ea18bfe4183d851cb4558d5f81
 SHA512:
-  metadata.gz: 3d7a785d1fdbf3bd226a64f489ef12e37cb474fe0334c894e98c0a86b90fba23d02d937eed3bc4f4e710791209556ab2ece950e50491d48874e128e95f62c44e
-  data.tar.gz: cad955e4ba4066a42ff821f03437cfe09fb926c283785050d34682cc95e676d89d608f03ca63c18574d012e5fcad168c5f289cd7d465f525846c877e91368635
+  metadata.gz: 468b8f5ffc67f878cf97a39b0733e01ee9a56c1a2fd82874feba18fb9de6876b2d835ab11b2dd40021139dff988152d66fe25af0adde1e5884884017435dc711
+  data.tar.gz: 1126af3e7ae9b0cbb25d9d2f4497e894ae33e2a34bb9cc975924233fcf48f8cd0527b67f7983aeee375d4e98f7d58005a913e5301a848aca2da92ce051b93732

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    commonmeta-ruby (3.2.14)
+    commonmeta-ruby (3.3)
       activesupport (>= 4.2.5, < 8.0)
       addressable (~> 2.8.1, < 2.8.2)
       base32-url (>= 0.7.0, < 1)
@@ -121,7 +121,7 @@ GEM
       crass (~> 1.0.2)
       nokogiri (>= 1.12.0)
     matrix (0.4.2)
-    minitest (5.18.0)
+    minitest (5.18.1)
     multi_json (1.15.0)
     namae (1.1.1)
     nokogiri (1.15.2-arm64-darwin)

data/bin/commonmeta

@@ -2,7 +2,7 @@
 
 require File.expand_path("../../lib/commonmeta", __FILE__)
 
-if (ARGV & %w(--version -v help --help encode decode encode_id decode_id json_feed_not_indexed json_feed_unregistered json_feed_by_blog)).empty?
+if (ARGV & %w(--version -v help --help encode decode encode_id decode_id encode_by_blog encode_by_uuid json_feed_not_indexed json_feed_unregistered json_feed_by_blog)).empty?
   Commonmeta::CLI.start(ARGV.dup.unshift("convert"))
 else
   Commonmeta::CLI.start

data/lib/commonmeta/author_utils.rb

@@ -41,7 +41,7 @@ module Commonmeta
       elsif id.nil? && author['ORCID'].present?
         id = author.fetch('ORCID')
       end
-      id = normalize_orcid(id)
+      id = normalize_orcid(id) || normalize_ror(id)
 
       # parse author type, i.e. "Person", "Organization" or not specified
       type = author.fetch('type', nil)

data/lib/commonmeta/cli.rb

@@ -68,6 +68,20 @@ module Commonmeta
       puts encode_container_id
     end
 
+    desc "", "encode_by_blog"
+
+    def encode_by_blog(blog_id)
+      prefix = get_doi_prefix_by_blog_id(blog_id)
+      puts encode_doi(prefix)
+    end
+
+    desc "", "encode_by_uuid"
+
+    def encode_by_uuid(uuid)
+      prefix = get_doi_prefix_by_json_feed_item_uuid(uuid)
+      puts encode_doi(prefix)
+    end
+
     desc "", "decode"
 
     def decode(doi)

data/lib/commonmeta/crossref_utils.rb

@@ -94,25 +94,67 @@ module Commonmeta
 
     def insert_crossref_creators(xml)
       xml.contributors do
-        Array.wrap(creators).each_with_index do |person, index|
-          xml.person_name("contributor_role" => "author",
-                          "sequence" => index.zero? ? "first" : "additional") do
-            insert_crossref_person(xml, person)
+        Array.wrap(creators).each_with_index do |creator, index|
+          if creator["type"] == "Organization"
+            xml.organization("contributor_role" => "author",
+                             "sequence" => index.zero? ? "first" : "additional") do
+              insert_crossref_organization(xml, creator)
+            end
+          elsif creator["givenName"].present? || creator["familyName"].present?
+            xml.person_name("contributor_role" => "author",
+                            "sequence" => index.zero? ? "first" : "additional") do
+              insert_crossref_person(xml, creator)
+            end
+          else
+            xml.unknown("contributor_role" => "author",
+                        "sequence" => index.zero? ? "first" : "additional") do
+              insert_crossref_anonymous(xml, creator)
+            end
+          end
+        end
+      end
+    end
+
+    def insert_crossref_person(xml, creator)
+      xml.given_name(creator["givenName"]) if creator["givenName"].present?
+      xml.surname(creator["familyName"]) if creator["familyName"].present?
+      if creator.dig("id") && URI.parse(creator.dig("id")).host == "orcid.org"
+        xml.ORCID(creator.dig("id"))
+      end
+      if creator["affiliation"].present?
+        xml.affiliations do
+          xml.institution do
+            xml.institution_name(creator.dig("affiliation", 0, "name")) if creator.dig("affiliation", 0, "name").present?
+            xml.institution_id(creator.dig("affiliation", 0, "affiliationIdentifier"), "type" => creator.dig("affiliation", 0, "affiliationIdentifierScheme")) if creator.dig("affiliation", 0, "affiliationIdentifier").present?
           end
         end
       end
     end
 
-    def insert_crossref_person(xml, person)
-      xml.given_name(person["givenName"]) if person["givenName"].present?
-      xml.surname(person["familyName"]) if person["familyName"].present?
-      if person.dig("id") && URI.parse(person.dig("id")).host == "orcid.org"
-        xml.ORCID(person.dig("id"))
+    def insert_crossref_organization(xml, creator)
+      xml.name(creator["name"]) if creator["name"].present?
+      if creator["affiliation"].present?
+        xml.affiliations do
+          xml.institution do
+            xml.institution_name(creator.dig("affiliation", 0, "name")) if creator.dig("affiliation", 0, "name").present?
+            xml.institution_id(creator.dig("affiliation", 0, "affiliationIdentifier"), "type" => creator.dig("affiliation", 0, "affiliationIdentifierScheme")) if creator.dig("affiliation", 0, "affiliationIdentifier").present?
+          end
+        end
       end
-      Array.wrap(person["affiliation"]).each do |affiliation|
-        attributes = { "affiliationIdentifier" => affiliation["affiliationIdentifier"],
-                       "affiliationIdentifierScheme" => affiliation["affiliationIdentifierScheme"], "schemeURI" => affiliation["schemeUri"] }.compact
-        xml.affiliation(affiliation["name"], attributes)
+    end
+
+    def insert_crossref_anonymous(xml, creator)
+      if person["affiliation"].present?
+        xml.anonymous do
+          xml.affiliations do
+            xml.institution do
+              xml.institution_name(creator.dig("affiliation", 0, "name")) if creator.dig("affiliation", 0, "name").present?
+              xml.institution_id(creator.dig("affiliation", 0, "affiliationIdentifier"), "type" => creator.dig("affiliation", 0, "affiliationIdentifierScheme")) if creator.dig("affiliation", 0, "affiliationIdentifier").present?
+            end
+          end
+        end
+      else
+        xml.anonymous
       end
     end
 
@@ -265,7 +307,7 @@ module Commonmeta
         }.compact
 
         # strip hyphen from UUIDs, as item_number can only be 32 characters long (UUIDv4 is 36 characters long)
-        alternate_identifier["alternateIdentifier"] = alternate_identifier["alternateIdentifier"].gsub('-','') if alternate_identifier["alternateIdentifierType"] == "UUID"
+        alternate_identifier["alternateIdentifier"] = alternate_identifier["alternateIdentifier"].gsub("-", "") if alternate_identifier["alternateIdentifierType"] == "UUID"
 
         xml.item_number(alternate_identifier["alternateIdentifier"], attributes)
       end

data/lib/commonmeta/readers/json_feed_reader.rb

@@ -20,8 +20,10 @@ module Commonmeta
 
         meta = string.present? ? JSON.parse(string) : {}
 
-        id = options[:doi] ? normalize_doi(options[:doi]) : normalize_id(meta.fetch("id", nil))
         url = normalize_url(meta.fetch("url", nil))
+        id = options[:doi] ? normalize_doi(options[:doi]) : normalize_id(meta.fetch("id", nil))
+        id = url if id.blank? && url.present?
+
         type = "Article"
         creators = if meta.fetch("authors", nil).present?
             get_authors(from_json_feed(Array.wrap(meta.fetch("authors"))))
@@ -82,7 +84,11 @@ module Commonmeta
       def get_references(meta)
         # check that references resolve
         Array.wrap(meta["references"]).reduce([]) do |sum, reference|
-          sum << reference if [200, 301, 302].include? HTTP.head(reference["doi"] || reference["url"]).status
+          if reference["doi"] && validate_doi(reference["doi"])
+            sum << reference if [200, 301, 302].include? HTTP.head(reference["doi"]).status
+          elsif reference["url"] && validate_url(reference["url"]) == "URL"
+            sum << reference if [200, 301, 302].include? HTTP.head(reference["url"]).status
+          end
 
           sum
         end
@@ -120,6 +126,28 @@ module Commonmeta
         blog = JSON.parse(response.body.to_s)
         blog["items"].map { |item| item["uuid"] }.first
       end
+
+      def get_doi_prefix_by_blog_id(blog_id)
+        # for generating a random DOI.
+
+        url = json_feed_by_blog_url(blog_id)
+        response = HTTP.get(url)
+        return { "string" => nil, "state" => "not_found" } unless response.status.success?
+
+        post = JSON.parse(response.body.to_s)
+        post.dig('prefix')
+      end
+
+      def get_doi_prefix_by_json_feed_item_uuid(uuid)
+        # for generating a random DOI. Prefix is based on the blog id.
+
+        url = json_feed_item_by_uuid_url(uuid)
+        response = HTTP.get(url)
+        return { "string" => nil, "state" => "not_found" } unless response.status.success?
+
+        post = JSON.parse(response.body.to_s)
+        post.dig('blog', 'prefix')
+      end
     end
   end
 end

data/lib/commonmeta/utils.rb

@@ -543,6 +543,11 @@ module Commonmeta
       orcid.gsub(/[[:space:]]/, "-") if orcid.present?
     end
 
+    def validate_ror(ror)
+      ror = Array(%r{\A(?:(?:http|https)://ror\.org/)?([0-9a-z]{7}\d{2})\z}.match(ror)).last
+      ror.gsub(/[[:space:]]/, "-") if ror.present?
+    end
+
     def validate_orcid_scheme(orcid_scheme)
       Array(%r{\A(http|https)://(www\.)?(orcid\.org)}.match(orcid_scheme)).last
     end
@@ -634,6 +639,14 @@ module Commonmeta
       "https://orcid.org/" + Addressable::URI.encode(orcid)
     end
 
+    def normalize_ror(ror)
+      ror = validate_ror(ror)
+      return nil unless ror.present?
+
+      # turn ROR ID into URL
+      "https://ror.org/" + Addressable::URI.encode(ror)
+    end
+
     # pick electronic issn if there are multiple
     # format issn as xxxx-xxxx
     def normalize_issn(input, options = {})
@@ -1384,6 +1397,23 @@ module Commonmeta
       "https://doi.org/#{prefix}/#{str}"
     end
 
+    def encode_doi_for_uuid(uuid, options = {})
+      # look up prefix for rogue scholar blog associated with uuid
+      # returns nil if unknown uuid or doi registration is not enabled for blog
+      json_feed_by_uuid(uuid)
+      # DOI suffix is a generated from a random number, encoded in base32
+      # suffix has 8 digits plus two checksum digits. With base32 there are
+      # 32 possible digits, so 8 digits gives 32^8 possible combinations
+      if options[:uuid]
+        str = Base32::URL.encode_uuid(options[:uuid], split: 7, checksum: true)
+      else
+        random_int = SecureRandom.random_number(32 ** 7..(32 ** 8) - 1)
+        suffix = Base32::URL.encode(random_int, checksum: true)
+        str = "#{suffix[0, 5]}-#{suffix[5, 10]}"
+      end
+      "https://doi.org/#{prefix}/#{str}"
+    end
+
     def decode_doi(doi, options = {})
       suffix = doi.split("/", 5).last
       if options[:uuid]
@@ -1415,5 +1445,9 @@ module Commonmeta
     def json_feed_by_blog_url(blog_id)
       "https://rogue-scholar.org/api/blogs/#{blog_id}"
     end
+
+    def json_feed_item_by_uuid_url(uuid)
+      "https://rogue-scholar.org/api/posts/#{uuid}"
+    end
   end
 end

data/lib/commonmeta/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Commonmeta
-  VERSION = '3.2.14'
+  VERSION = '3.3'
 end

data/spec/cli_spec.rb

@@ -311,12 +311,21 @@ describe Commonmeta::CLI do
     # end
   end
 
-  describe "encode" do
-    let(:input) { "10.53731" }
-
+  describe "encode", vcr: true do
     it "blog prefix" do
+      input = "10.53731"
       expect { subject.encode input }.to output(/https:\/\/doi.org\/10.53731/).to_stdout
     end
+
+    it "by_blog" do
+      input = "tyfqw20"
+      expect { subject.encode_by_blog input }.to output(/https:\/\/doi.org\/10.59350/).to_stdout
+    end
+
+    it "by_uuid" do
+      input = "2b22bbba-bcba-4072-94cc-3f88442fff88"
+      expect { subject.encode_by_uuid input }.to output(/https:\/\/doi.org\/10.54900/).to_stdout
+    end
   end
 
   describe "decode" do