uc3-dmp-id 0.1.26 → 0.1.28

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f692820577dff088fec1d1df2e4975a00a6b30952fa66979b1fa8449523d092e
-  data.tar.gz: 8bf2326f3a6fddf9454e915c596eb40231c939337e517de73f6ea998f04a9d6f
+  metadata.gz: f984b52821821510b5be7f91887b366b67f2dfd9e85f29bfaa2a988b34dff2f3
+  data.tar.gz: 5e6d1ced0429757d65c61b5007e325e6ebe9ad8cd94b5fe67d4366289ed12c11
 SHA512:
-  metadata.gz: fcc0689438e54715882ed7c30aa7d608e9c3abaa1e61fec7589a213c7665f257be933f93a482dc4d589756517c94a7f9b4803930f0a2a1bd51d877cd890b8a90
-  data.tar.gz: df18ed7a17053c9527e6a111150b81c2a6d285bda1263b854b0cc3fae5b703e3ef9dd3213b62230431c898c948cf80f6e821da3515089e6cfc156622223bdf7e
+  metadata.gz: d23256d8ef17d0644351ddb1087ce918f6e6868fbcb0a81e13e8322d9fe39451854f19924040cb9f43fcc2163d7dc1efad53d538b0fdbbf871f1a728e625043b
+  data.tar.gz: c054f89e7c9347a61e5ea6a8d7246c6739410995a10f4d433fd0837c0d64097b9b70e51710d275fb4f0c0ef051e9cb60e579c40ee258bfdd45235c3889e5410e

data/lib/uc3-dmp-id/comparator.rb

@@ -53,35 +53,39 @@ module Uc3DmpId
     # }
     # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
     def compare(hash:)
-      return [] unless hash.is_a?(Hash) && !hash['title'].nil?
+      scoring = []
+      return scoring unless hash.is_a?(Hash) && !hash['title'].nil?
 
-      # Compare the grant ids. If we have a match return the response immediately since that is
-      # a very positive match!
-      scoring = @dmps.map do |entry|
+      @dmps.each do |entry|
         dmp = entry.fetch('_source', {})
-        response = { dmp_id: dmp['_id'], confidence: 'None', score: 0, notes: [] }
+
+        # Compare the grant ids. If we have a match return the response immediately since that is
+        # a very positive match!
+        response = { confidence: 'None', score: 0, notes: [] }
         response = _grants_match?(array: hash.fetch('grant_ids', []), dmp:, response:)
-        return response if response[:confidence] != 'None'
+        scoring << respoonse if response[:confidence] != 'None'
+        next if response[:confidence] != 'None'
 
+        # Compare the people involved, their affiliations and any funding opportunity numbers
         response = _opportunities_match?(array: hash.fetch('funder_opportunity_ids', []), dmp:, response:)
         response = _orcids_match?(array: hash.fetch('people_ids', []), dmp:, response:)
         response = _last_name_and_affiliation_match?(hash:, dmp:, response:)
 
-        # Only process the following if we had some matching contributors, affiliations or opportuniy nbrs
+        # Only process the following if we had some matching people, affiliations or opportunity nbrs
         response = _repository_match?(hash:, dmp:, response:) if response[:score].positive?
-        # response = _keyword_match?(array: hash['keywords'], response:) if response[:score].positive?
         response = _text_match?(type: 'title', text: hash['title'], dmp:, response:) if response[:score].positive?
         response = _text_match?(type: 'abstract', text: hash['description'], dmp:, response:) if response[:score].positive?
         # If the score is less than 3 then we have no confidence that it is a match
-        return nil if response[:score] <= 2
+        next if response[:score] <= 2
 
         # Set the confidence level based on the score
+        response[:dmp_id] = dmp['_id']
         response[:confidence] = if response[:score] > 10
                                   'High'
                                 else
                                   (response[:score] > 5 ? 'Medium' : 'Low')
                                 end
-        response
+        scoring << response
       end
 
       # TODO: introduce a tie-breaker here (maybe the closes to the project_end date)
@@ -171,7 +175,7 @@ module Uc3DmpId
     # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
     def _last_name_and_affiliation_match?(hash:, dmp:, response:)
       return response unless hash.is_a?(Hash) && dmp.is_a?(Hash) && response.is_a?(Hash)
-      return response unless hash['people'].is_a?(Array) && !dmp['people'].empty?
+      return response unless hash['people'].is_a?(Array)
 
       # Check the person last names and affiliation name and RORs
       last_names_matched = _compare_arrays(array_a: dmp['people'], array_b: hash['people'])
@@ -193,7 +197,7 @@ module Uc3DmpId
     # rubocop:disable Metrics/AbcSize
     def _repository_match?(hash:, dmp:, response:)
       return response unless hash.is_a?(Hash) && dmp.is_a?(Hash) && response.is_a?(Hash)
-      return response unless hash['repo_ids'].is_a?(Array) && !dmp['repo_ids'].empty?
+      return response unless hash['repo_ids'].is_a?(Array)
 
       matched = _compare_arrays(array_a: dmp['repo_ids'], array_b: hash['repo_ids'])
       return response if matched <= 0

data/lib/uc3-dmp-id/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Uc3DmpId
-  VERSION = '0.1.26'
+  VERSION = '0.1.28'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: uc3-dmp-id
 version: !ruby/object:Gem::Version
-  version: 0.1.26
+  version: 0.1.28
 platform: ruby
 authors:
 - Brian Riley
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-03-10 00:00:00.000000000 Z
+date: 2024-03-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: json