uc3-dmp-id 0.1.34 → 0.1.36

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e8ed82601dab8f040bdef2ca093b8a2c90f81160c5a990533947ff3750a5d97c
-  data.tar.gz: 563c25f02e5c85a41960bad34687bdb4376228cce26e00a8c92024ab4f289055
+  metadata.gz: 60af308240750c987c1bf26d88daedb3bef5a65faaed4e71a508626c44539873
+  data.tar.gz: bc7c71413c2faf9d1d9c838bbc6069a094e2e48a19754de3998b2056295f4f17
 SHA512:
-  metadata.gz: fdd55f05e2c95cf243e8398070e9d690e4ecf7d2d35ba639f937956298e2dcb6c28e9d5600af8a846b7bc8bc0c507b1456f55a56f82f80c456a885311e64e84b
-  data.tar.gz: 0ae5115a0d8856b72f10f86afe311e835e6bf52a950eaf5dd87a3b4af5acae3ad72958c3198485f53ff56184df2034f6c9347d69f745ad71b4d7e9d35e670fd1
+  metadata.gz: 0ff7acaf1a66c9fcc54b76860a86e1544e1ff03c0de11258610e6c1ba93698099d47b81f4cb96bdafc38098505c28f14150f3455e0c9d226b29aa83776d6a158
+  data.tar.gz: c576d006c4167ba347047c6693965539d54f3093108c8288882070527e67ba32869b3816340e928632f3dbab8a3b82d40460f35d4e985b39ed343f3816ee1aae

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

@@ -24,6 +24,8 @@ module Uc3DmpId
       @details_hash = {}
 
       @dmps = args.fetch(:dmps, [])
+
+      @logger&.debug(message: 'Comparator DMPs', details: @dmps)
       raise ComparatorError, MSG_MISSING_DMPS if @dmps.empty?
     end
 
@@ -56,14 +58,13 @@ module Uc3DmpId
       scoring = []
       return scoring unless hash.is_a?(Hash) && !hash['title'].nil?
 
-      @dmps.each do |entry|
-        dmp = entry.fetch('_source', {})
-
+      @dmps.each do |dmp|
+        @logger&.debug(message: 'Incoming external work', details: hash)
         # 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:)
-        scoring << respoonse if response[:confidence] != 'None'
+        scoring << response if response[:confidence] != 'None'
         next if response[:confidence] != 'None'
 
         # Compare the people involved, their affiliations and any funding opportunity numbers
@@ -79,7 +80,7 @@ module Uc3DmpId
         next if response[:score] <= 2
 
         # Set the confidence level based on the score
-        response[:dmp_id] = entry['_id']
+        response[:dmp_id] = "DMP##{dmp['dmp_id']}"
         response[:confidence] = if response[:score] > 10
                                   'High'
                                 else
@@ -112,6 +113,7 @@ module Uc3DmpId
       matched = _compare_arrays(array_a: dmp['grant_ids'], array_b: ids)
       return response if matched <= 0
 
+      @logger&.debug(message: 'Grant ID match!', details: { dmp: dmp['grant_ids'], work: ids })
       response[:confidence] = 'Absolute'
       response[:score] = 100
       response[:notes] << 'the grant ID matched'
@@ -135,6 +137,7 @@ module Uc3DmpId
       matched = _compare_arrays(array_a: dmp['funder_opportunity_ids'], array_b: ids)
       return response if matched <= 0
 
+      @logger&.debug(message: 'Opportunities match!', details: { dmp: dmp['funder_opportunity_ids'], work: ids })
       response[:score] += 5
       response[:notes] << 'the funding opportunity number matched'
       response
@@ -161,6 +164,7 @@ module Uc3DmpId
       matched = _compare_arrays(array_a: dmp['people_ids'], array_b: ids)
       return response if matched <= 0
 
+      @logger&.debug(message: 'ORCID match!', details: { dmp: dmp['people_ids'], work: ids })
       response[:score] += (matched * 2)
       response[:notes] << 'contributor ORCIDs matched'
       response
@@ -184,6 +188,14 @@ module Uc3DmpId
       affil_names_matched = _compare_arrays(array_a: dmp.fetch('affiliations', []), array_b: hash['affiliations'])
       return response if last_names_matched <= 0 && rors_matched <= 0 && affil_names_matched <= 0
 
+      @logger&.debug(
+        message: 'Contributor name match',
+        details: {
+          people: { dmp: dmp['people'], work: hash['people'] },
+          rors: { dmp: dmp['affiliation_ids'], work: hash['affiliation_ids'] },
+          places: { dmp: dmp['affiliations'], work: hash['affiliations'] }
+        }
+      )
       response[:score] += last_names_matched + rors_matched + affil_names_matched
       response[:notes] << 'contributor names and affiliations matched'
       response
@@ -203,6 +215,7 @@ module Uc3DmpId
       matched = _compare_arrays(array_a: dmp['repo_ids'], array_b: hash['repo_ids'])
       return response if matched <= 0
 
+      @logger&.debug(message: 'Repos match!', details: { dmp: dmp['repo_ids'], work: hash['repo_ids'] })
       response[:score] += matched
       response[:notes] << 'repositories matched'
       response
@@ -226,6 +239,8 @@ module Uc3DmpId
       # @logger&.debug(message: 'Text::WhiteSimilarity score', details:)
       return response if details[:nlp_score] < 0.5
 
+      @logger&.debug(message: 'Titles match', details: { dmp: dmp['title'], work: text }) if type == 'title'
+      @logger&.debug(message: 'Abstracts match', details: { dmp: dmp['description'], work: text }) unless type == 'title'
       response[:score] += details[:nlp_score] >= 0.75 ? 5 : 2
       response[:notes] << "#{type}s are similar"
       response

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

@@ -246,6 +246,7 @@ module Uc3DmpId
 
           # Change the name of the `logic` array to `notes`
           rec['notes'] = rec['logic']
+          rec['score'] = rec['score'].to_s
           # For `work-type` that equal `outputmanagementplan`, change it to `output_management_plan`
           rec['work_type'] = 'output_management_plan' if rec['work_type'] == 'outputmanagementplan'
 

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

@@ -204,6 +204,7 @@ module Uc3DmpId
             mods['related_works'][related.identifier]['status'] = related['status']
           end
         end
+
         client.put_item(json: mods, logger:)
         json.delete('dmphub_modifications')
         json

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

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Uc3DmpId
-  VERSION = '0.1.34'
+  VERSION = '0.1.36'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: uc3-dmp-id
 version: !ruby/object:Gem::Version
-  version: 0.1.34
+  version: 0.1.36
 platform: ruby
 authors:
 - Brian Riley
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-04-24 00:00:00.000000000 Z
+date: 2024-05-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: json