commonmeta-ruby 3.2.5 → 3.2.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f85e675e7fd41ff19edb24241a9566eb6e0fca9f09a8adf1cc8bddb182e7d171
-  data.tar.gz: 7a6c7047decafd643d3091ed73bf2b69554794ff885b86416f73a768bcdcab3f
+  metadata.gz: 7a50ad8ff25b9ff753b60cc9a34f578ae064cedc2417a5db146898e8202f8dc3
+  data.tar.gz: cc7f32e49a6d0621a30884a8226a9dda04cd5b13ed6545aefc28fd8f19386acb
 SHA512:
-  metadata.gz: b40188daa6660a0a81b99369bde7ae3f125705411d5d470ec2de827971e26596fe45e25a397aa6db06847934e86c11d992cba5f957d38d40886465d8b553b037
-  data.tar.gz: '09934d0036707c2dd30f66c5c352374b37efb85d6e092c572736acbb9d00ab3c1532d72deed4a4f0e5ebce137ba1a877803f61e18351fb114d6675555b3f45a8'
+  metadata.gz: 16d052d4e0dda3aed49235f43914f3d88e46627741500edb987cbf6dad00091234aa2304541f8ba6d713b05c9b158717d2c5632aa024e60dc694f7126752dda1
+  data.tar.gz: d6a7044f3b3722dc71f22f8c98e12d3f5c5e79b82bdf1aff745ccb74ecbd4febd2d53926924b635b97334c30df233dd793802dabfe0ea038f387488acf0ae4ef

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    commonmeta-ruby (3.2.5)
+    commonmeta-ruby (3.2.7)
       activesupport (>= 4.2.5, < 8.0)
       addressable (~> 2.8.1, < 2.8.2)
       base32-url (>= 0.5.0, < 1)
@@ -10,6 +10,7 @@ PATH
       citeproc-ruby (~> 2.0)
       csl-styles (~> 2.0)
       edtf (~> 3.0, >= 3.0.4)
+      feedparser (~> 2.2)
       gender_detector (~> 2.0)
       http (~> 5.1, >= 5.1.1)
       json-ld-preloaded (~> 3.2, >= 3.2.2)
@@ -57,7 +58,7 @@ GEM
     docile (1.4.0)
     domain_name (0.5.20190701)
       unf (>= 0.0.5, < 1.0.0)
-    ebnf (2.3.3)
+    ebnf (2.3.4)
       htmlentities (~> 4.3)
       rdf (~> 3.2)
       scanf (~> 1.0)
@@ -65,6 +66,9 @@ GEM
       unicode-types (~> 1.8)
     edtf (3.1.1)
       activesupport (>= 3.0, < 8.0)
+    feedparser (2.2.0)
+      logutils (>= 0.6.1)
+      textutils (>= 1.0.0)
     ffi (1.15.5)
     ffi-compiler (1.0.1)
       ffi (>= 1.0.0)
@@ -86,6 +90,7 @@ GEM
     http-form_data (2.3.0)
     i18n (1.14.1)
       concurrent-ruby (~> 1.0)
+    iniparser (1.0.1)
     json (2.6.3)
     json-canonicalization (0.3.2)
     json-ld (3.2.5)
@@ -110,6 +115,7 @@ GEM
     llhttp-ffi (0.4.0)
       ffi-compiler (~> 1.0)
       rake (~> 13.0)
+    logutils (0.6.1)
     loofah (2.21.3)
       crass (~> 1.0.2)
       nokogiri (>= 1.12.0)
@@ -129,6 +135,8 @@ GEM
       addressable (>= 2.4.0)
       nokogiri (>= 1.8.0)
       public_suffix (>= 4.0.0, < 5)
+    props (1.2.0)
+      iniparser (>= 0.1.0)
     public_suffix (4.0.7)
     racc (1.7.0)
     rack (3.0.7)
@@ -205,6 +213,7 @@ GEM
       rubocop-capybara (~> 2.17)
       rubocop-factory_bot (~> 2.22)
     ruby-progressbar (1.13.0)
+    rubyzip (2.3.2)
     scanf (1.0.0)
     simplecov (0.22.0)
       docile (~> 1.1)
@@ -218,6 +227,11 @@ GEM
       matrix (~> 0.4)
       rdf (~> 3.2)
     temple (0.10.2)
+    textutils (1.4.0)
+      activesupport
+      logutils (>= 0.6.1)
+      props (>= 1.1.2)
+      rubyzip (>= 1.0.0)
     thor (1.2.2)
     tilt (2.2.0)
     tzinfo (2.0.6)

data/commonmeta.gemspec

@@ -26,6 +26,7 @@ Gem::Specification.new do |s|
   s.add_dependency 'citeproc-ruby', '~> 2.0'
   s.add_dependency 'csl-styles', '~> 2.0'
   s.add_dependency 'edtf', '~> 3.0', '>= 3.0.4'
+  s.add_dependency 'feedparser', '~> 2.2'
   s.add_dependency 'gender_detector', '~> 2.0'
   s.add_dependency 'http', '~> 5.1', '>= 5.1.1'
   s.add_dependency 'json-ld-preloaded', '~> 3.2', '>= 3.2.2'

data/lib/commonmeta/author_utils.rb

@@ -24,6 +24,8 @@ module Commonmeta
       family_name = parse_attributes(author.fetch('familyName', nil)) ||
                     parse_attributes(author.fetch('family', nil))
 
+      name = cleanup_author(name)
+
       # parse author identifier
       id = parse_attributes(author.fetch('id', nil), first: true) ||
            parse_attributes(author.fetch('identifier', nil), first: true) ||
@@ -62,8 +64,6 @@ module Commonmeta
       # parse author contributor role
       contributor_type = parse_attributes(author.fetch('contributorType', nil))
 
-      name = cleanup_author(name)
-
       # split name for type Person into given/family name if not already provided
       if type == 'Person' && given_name.blank? && family_name.blank?
         Namae.options[:include_particle_in_family] = true
@@ -98,6 +98,19 @@ module Commonmeta
         author = author.gsub(/[[:space:]]([A-Z]\.)?(-?[A-Z]\.)$/, ', \1\2')
       end
 
+      # strip suffixes, e.g. "John Smith, MD" as the named parser doesn't handle them
+      author = author.split(',').first if %w[MD PhD].include? author.split(', ').last
+
+      # remove email addresses
+      email = validate_email(author)
+      author = author.gsub(email, '') if email.present?
+
+      # strip spaces at the beginning and end of string
+      author = author.strip
+
+      # remove parentheses around names
+      author = author[1..-2] if author[0] == '(' && author[-1] == ')'
+
       # remove spaces around hyphens
       author = author.gsub(' - ', '-')
 
@@ -113,12 +126,16 @@ module Commonmeta
       # check if a name has only one word, e.g. "FamousOrganization"
       return false if name.to_s.split(' ').size == 1
 
+      # check for suffixes, e.g. "John Smith, MD"
+      return true if %w[MD PhD].include? name.split(', ').last
+
       # check of name can be parsed into given/family name
       Namae.options[:include_particle_in_family] = true
       names = Namae.parse(name)
+
       parsed_name = names.first
       return true if parsed_name && parsed_name.given
-
+      
       false
     end
 

data/lib/commonmeta/cli.rb

@@ -1,9 +1,9 @@
 # frozen_string_literal: true
 
-require 'thor'
+require "thor"
 
-require_relative 'doi_utils'
-require_relative 'utils'
+require_relative "doi_utils"
+require_relative "utils"
 
 module Commonmeta
   class CLI < Thor
@@ -18,18 +18,18 @@ module Commonmeta
     # from http://stackoverflow.com/questions/22809972/adding-a-version-option-to-a-ruby-thor-cli
     map %w[--version -v] => :__print_version
 
-    desc '--version, -v', 'print the version'
+    desc "--version, -v", "print the version"
 
     def __print_version
       puts Commonmeta::VERSION
     end
 
-    desc '', 'convert metadata'
-    method_option :from, aliases: '-f'
-    method_option :to, aliases: '-t', default: 'schema_org'
+    desc "", "convert metadata"
+    method_option :from, aliases: "-f"
+    method_option :to, aliases: "-t", default: "schema_org"
     method_option :regenerate, type: :boolean, force: false
-    method_option :style, aliases: '-s', default: 'apa'
-    method_option :locale, aliases: '-l', default: 'en-US'
+    method_option :style, aliases: "-s", default: "apa"
+    method_option :locale, aliases: "-l", default: "en-US"
     method_option :show_errors, type: :boolean, force: false
     method_option :doi
     method_option :depositor
@@ -47,7 +47,7 @@ module Commonmeta
                               depositor: options[:depositor],
                               email: options[:email],
                               registrant: options[:registrant])
-      to = options[:to] || 'schema_org'
+      to = options[:to] || "schema_org"
 
       if options[:show_errors] && !metadata.valid?
         warn metadata.errors
@@ -56,38 +56,34 @@ module Commonmeta
       end
     end
 
-    desc '', 'encode'
+    desc "", "encode"
 
     def encode(prefix)
       puts encode_doi(prefix)
     end
 
-    desc '', 'encode_id'
+    desc "", "encode_id"
 
     def encode_id
       puts encode_container_id
     end
 
-    desc '', 'decode'
+    desc "", "decode"
 
     def decode(doi)
       puts decode_doi(doi)
     end
 
-    desc '', 'decode_id'
+    desc "", "decode_id"
 
     def decode_id(id)
       puts decode_container_id(id)
     end
 
-    desc '', 'feed'
+    desc "", "json_feed"
 
-    def json_feed(id)
-      if id.present?
-        puts get_json_feed(id)
-      else
-        puts get_json_feed
-      end
+    def json_feed_unregistered
+      puts get_json_feed_unregistered
     end
 
     default_task :convert

data/lib/commonmeta/crossref_utils.rb

@@ -61,7 +61,6 @@ module Commonmeta
           insert_crossref_publication_date(xml)
           insert_crossref_abstract(xml)
           insert_crossref_issn(xml)
-          insert_crossref_alternate_identifiers(xml)
           insert_crossref_access_indicators(xml)
           insert_doi_data(xml)
           insert_citation_list(xml)
@@ -78,8 +77,8 @@ module Commonmeta
         insert_crossref_titles(xml)
         insert_posted_date(xml)
         insert_institution(xml)
+        insert_publisher_item(xml)
         insert_crossref_abstract(xml)
-        insert_crossref_alternate_identifiers(xml)
         insert_crossref_access_indicators(xml)
         insert_doi_data(xml)
         insert_citation_list(xml)
@@ -132,13 +131,14 @@ module Commonmeta
       xml.citation_list do
         Array.wrap(references).each do |ref|
           xml.citation('key' => ref['key']) do
-            xml.journal_article(ref['journal_title'])
-            xml.author(ref['author'])
-            xml.volume(ref['volume'])
-            xml.first_page(ref['first_page'])
-            xml.cYear(ref['cYear'])
-            xml.article_title(ref['article_title'])
+            xml.journal_article(ref['journal_title']) if ref['journal_title'].present?
+            xml.author(ref['author']) if ref['author'].present?
+            xml.volume(ref['volume']) if ref['volume'].present?
+            xml.first_page(ref['first_page']) if ref['first_page'].present?
+            xml.cYear(ref['cYear']) if ref['cYear'].present?
+            xml.article_title(ref['article_title']) if ref['article_title'].present?
             xml.doi(ref['doi']) if ref['doi'].present?
+            xml.unstructured_citation(ref['url']) if ref['url'].present?
           end
         end
       end
@@ -159,16 +159,6 @@ module Commonmeta
     #     'resourceTypeGeneral' => types["resourceTypeGeneral"] || Metadata::SO_TO_DC_TRANSLATIONS[types["schemaOrg"]] || "Other")
     # end
 
-    def insert_crossref_alternate_identifiers(xml)
-      alternate_identifier = Array.wrap(alternate_identifiers).reject do |r|
-        %w[DOI, URL].include?(r['alternate_identifier_type'])
-      end.first
-      return xml if alternate_identifier.blank?
-
-      xml.item_number(alternate_identifier['alternateIdentifier'],
-                      'item_number_type' => alternate_identifier['alternateIdentifierType'])
-    end
-
     def insert_crossref_access_indicators(xml)
       return xml if license.blank?
 
@@ -265,6 +255,20 @@ module Commonmeta
       end
     end
 
+    def insert_publisher_item(xml)
+      return xml if alternate_identifiers.blank?
+
+      xml.publisher_item do
+        alternate_identifiers.each do |alternate_identifier|
+          attributes = {
+            'item_number_type' => alternate_identifier['alternateIdentifierType'] ? alternate_identifier['alternateIdentifierType'].downcase : nil
+          }.compact
+
+          xml.item_number(alternate_identifier['alternateIdentifier'], attributes)
+        end
+      end
+    end
+
     def insert_doi_data(xml)
       return xml if doi_from_url(id).blank? || url.blank?
 

data/lib/commonmeta/readers/json_feed_reader.rb

@@ -1,4 +1,5 @@
 # frozen_string_literal: true
+require "uri"
 
 module Commonmeta
   module Readers
@@ -9,7 +10,7 @@ module Commonmeta
         url = normalize_id(id)
         response = HTTP.get(url)
         return { "string" => nil, "state" => "not_found" } unless response.status.success?
-  
+
         { "string" => response.body.to_s }
       end
 
@@ -58,6 +59,8 @@ module Commonmeta
 
           sum
         end
+        references = get_references(meta)
+        alternate_identifiers = [{ "alternateIdentifier" => meta["uuid"], "alternateIdentifierType" => "UUID" }]
 
         { "id" => id,
           "type" => type,
@@ -71,23 +74,40 @@ module Commonmeta
           "descriptions" => descriptions,
           "license" => license,
           "subjects" => subjects.presence,
+          "references" => references.presence,
+          "alternate_identifiers" => alternate_identifiers,
           "state" => state }.compact.merge(read_options)
       end
 
-      def get_json_feed(id = nil)
+      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
+
+          sum
+        end
+      end
+
+      def get_json_feed_unregistered
         # get JSON Feed items not registered as DOIs
-       
-        url = json_feed_url(id)
+
+        url = json_feed_unregistered_url
         response = HTTP.get(url)
         return { "string" => nil, "state" => "not_found" } unless response.status.success?
 
-        if id.present?
-          blog = JSON.parse(response.body.to_s)
-          blog["items"].select { |item| !validate_doi(item["id"]) }.map { |item| item["uuid"] }.first
-        else
-          posts = JSON.parse(response.body.to_s)
-          posts.select { |post| !validate_doi(post["id"]) }.map { |post| post["uuid"] }.first
-        end
+        posts = JSON.parse(response.body.to_s)
+        posts.map { |post| post["uuid"] }.first
+      end
+
+      def get_json_feed_updated
+        # get JSON Feed items updated since last check
+
+        url = json_feed_unregistered_url
+        response = HTTP.get(url)
+        return { "string" => nil, "state" => "not_found" } unless response.status.success?
+
+        posts = JSON.parse(response.body.to_s)
+        posts.map { |post| post["uuid"] }.first
       end
     end
   end

data/lib/commonmeta/utils.rb

@@ -557,6 +557,12 @@ module Commonmeta
       end
     end
 
+    def validate_email(str)
+      email_regex = /\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,4}\b/i
+      parsed_emails = str.scan(email_regex)
+      parsed_emails.first
+    end
+
     def parse_attributes(element, options = {})
       content = options[:content] || "__content__"
 
@@ -604,6 +610,12 @@ module Commonmeta
       # clean up URL
       uri.path = PostRank::URI.clean(uri.path)
 
+      # optionally remove query and fragment
+      if options[:remove_query_string]
+        uri.query = nil
+        uri.fragment = nil
+      end
+
       uri.to_s
     rescue Addressable::URI::InvalidURIError
       nil
@@ -1381,9 +1393,7 @@ module Commonmeta
       Base32::URL.decode(id)
     end
 
-    def json_feed_url(id = nil)
-      return "https://rogue-scholar.org/api/blogs/#{id}" if id.present?
-
+    def json_feed_unregistered_url
       "https://rogue-scholar.org/api/posts/unregistered"
     end
   end

data/lib/commonmeta/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Commonmeta
-  VERSION = '3.2.5'
+  VERSION = '3.2.7'
 end