uc3-dmp-id 0.1.25 → 0.1.26

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 69c311d2bc8bd7acee827939e1d99c4fdf233ddf5e29386f682ac60f67478a6a
-  data.tar.gz: 829d403c8ada7d01f444494d163bee0fe1c34db320a93bb34abfcc98aebeb122
+  metadata.gz: f692820577dff088fec1d1df2e4975a00a6b30952fa66979b1fa8449523d092e
+  data.tar.gz: 8bf2326f3a6fddf9454e915c596eb40231c939337e517de73f6ea998f04a9d6f
 SHA512:
-  metadata.gz: 8dc439bf6244f758ceb5c1afb5c2d825b58705342b3a146dfcae9af7dcde86352705d703a2aa0e003922f99c61ec2757aaf7c0ea26279c12efdfedbb6cec8b1b
-  data.tar.gz: fd03efac1ba2cacc9be9d40334c835259461a84bb32715e5cbc9723c202d56a52aecdcc8d67ed29a1f050ee0bac853e973955f2211ee41a9da34984a5d13b342
+  metadata.gz: fcc0689438e54715882ed7c30aa7d608e9c3abaa1e61fec7589a213c7665f257be933f93a482dc4d589756517c94a7f9b4803930f0a2a1bd51d877cd890b8a90
+  data.tar.gz: df18ed7a17053c9527e6a111150b81c2a6d285bda1263b854b0cc3fae5b703e3ef9dd3213b62230431c898c948cf80f6e821da3515089e6cfc156622223bdf7e

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

@@ -29,25 +29,28 @@ module Uc3DmpId
 
     # Compare the incoming hash with the DMP details that were gathered during initialization.
     #
-    # The Hash should contain:
-    #  {
-    #    title: "Example research project",
-    #    abstract: "Lorem ipsum psuedo abstract",
-    #    keywords: ["foo", "bar"],z
-    #    people: [
-    #      {
-    #        id: "https://orcid.org/blah",
-    #        last_name: "doe",
-    #        affiliation: { id: "https://ror.org/blah", name: "Foo" }
-    #      }
-    #    ],
-    #    fundings: [
-    #      { id: "https://doi.org/crossref123", name: "Bar", grant: ["1234", "http://foo.bar/543"] }
-    #    ],
-    #    repositories: [
-    #      { id: ["http://some.repo.org", "https://doi.org/re3data123"], name: "Repo" }
-    #    ]
-    #  }
+    # The incoming Hash should match the documents found in OpenSearch. For example:
+    # {
+    #   "people": ["john doe", "jdoe@example.com"],
+    #   "people_ids": ["https://orcid.org/0000-0000-0000-ZZZZ"],
+    #   "affiliations": ["example college"],
+    #   "affiliation_ids": ["https://ror.org/00000zzzz"],
+    #   "funder_ids": ["https://doi.org/10.13039/00000000000"],
+    #   "funders": ["example funder (example.gov)"],
+    #   "funder_opportunity_ids": ["485yt8325ty"],
+    #   "grant_ids": [],
+    #   "funding_status": "planned",
+    #   "dmp_id": "doi.org/11.22222/A1B2c3po",
+    #   "title": "example data management plan",
+    #   "visibility": "private",
+    #   "featured": 0,
+    #   "description": "the example project abstract",
+    #   "project_start": "2022-01-03",
+    #   "project_end": "2024-12-23",
+    #   "created": "2023-08-07",
+    #   "modified": "2023-08-07",
+    #   "registered": "2023-08-07"
+    # }
     # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
     def compare(hash:)
       return [] unless hash.is_a?(Hash) && !hash['title'].nil?
@@ -57,18 +60,18 @@ module Uc3DmpId
       scoring = @dmps.map do |entry|
         dmp = entry.fetch('_source', {})
         response = { dmp_id: dmp['_id'], confidence: 'None', score: 0, notes: [] }
-        response = _grants_match?(array: hash['fundings'], dmp:, response:)
+        response = _grants_match?(array: hash.fetch('grant_ids', []), dmp:, response:)
         return response if response[:confidence] != 'None'
 
-        response = _opportunities_match?(array: hash['fundings'], dmp:, response:)
-        response = _orcids_match?(array: hash['people'], dmp:, response:)
-        response = _last_name_and_affiliation_match?(array: hash['people'], dmp:, response:)
+        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
-        response = _repository_match?(array: hash['repositories'], dmp:, response:) if response[:score].positive?
+        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['abstract'], 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
 
@@ -160,28 +163,20 @@ module Uc3DmpId
     # rubocop:enable Metrics/AbcSize
 
     # Returns whether or not the inciming list of creators/contributors match those on the DMP. Expecting:
-    #   [
-    #      {
-    #        id: "https://orcid.org/blah",
-    #        last_name: "doe",
-    #        affiliation: { id: "https://ror.org/blah", name: "Foo" }
-    #      }
-    #    ]
+    #   {
+    #     people: ["john doe", "jdoe@example.com"],
+    #     affiliations: ["example college"],
+    #     affiliation_ids: ["https://ror.org/blah"]
+    #   }
     # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
-    def _last_name_and_affiliation_match?(array:, dmp:, response:)
-      return response unless array.is_a?(Array) && dmp.is_a?(Hash) && response.is_a?(Hash)
-      return response unless dmp['people'].is_a?(Array) && !dmp['people'].empty?
-
-      array = array.select { |repo| repo.is_a?(Hash) }
-      affiliations = array.map { |person| person['affiliation'] }&.flatten&.compact&.uniq
-      last_names = array.map { |person| person['last_name']&.downcase&.strip }&.flatten&.compact&.uniq
-      rors = affiliations.map { |affil| affil['id']&.downcase&.strip }&.flatten&.compact&.uniq
-      affil_names = affiliations.map { |affil| affil['name']&.downcase&.strip }&.flatten&.compact&.uniq
+    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?
 
       # Check the person last names and affiliation name and RORs
-      last_names_matched = _compare_arrays(array_a: dmp['people'], array_b: last_names)
-      rors_matched = _compare_arrays(array_a: dmp.fetch('affiliation_ids', []), array_b: rors)
-      affil_names_matched = _compare_arrays(array_a: dmp.fetch('affiliations', []), array_b: affil_names)
+      last_names_matched = _compare_arrays(array_a: dmp['people'], array_b: hash['people'])
+      rors_matched = _compare_arrays(array_a: dmp.fetch('affiliation_ids', []), array_b: hash['affiliation_ids'])
+      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
 
       response[:score] += last_names_matched + rors_matched + affil_names_matched
@@ -191,20 +186,16 @@ module Uc3DmpId
     # rubocop:enable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
 
     # Returns whether or not the incoming list of repositories match those defined in the DMP. Expecting:
-    #    [
-    #      { id: ["http://some.repo.org", "https://doi.org/re3data123"], name: "Repo" }
-    #    ]
+    #    {
+    #      repo_ids: ["http://some.repo.org", "https://doi.org/re3data123"],
+    #      repos: ["repo"]
+    #    }
     # rubocop:disable Metrics/AbcSize
-    def _repository_match?(array:, dmp:, response:)
-      return response unless array.is_a?(Array) && dmp.is_a?(Hash) && response.is_a?(Hash)
-      return response unless dmp['repositories'].is_a?(Array) && !dmp['repositories'].empty?
-
-      # We only care about repositories with ids/urls
-      ids = array.select { |repo| repo.is_a?(Hash) }
-                 .map { |repo| repo['id'].map { |id| id&.downcase&.strip } }
-                 .flatten.compact.uniq
+    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?
 
-      matched = _compare_arrays(array_a: dmp['repositories'], array_b: ids)
+      matched = _compare_arrays(array_a: dmp['repo_ids'], array_b: hash['repo_ids'])
       return response if matched <= 0
 
       response[:score] += matched

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

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Uc3DmpId
-  VERSION = '0.1.25'
+  VERSION = '0.1.26'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: uc3-dmp-id
 version: !ruby/object:Gem::Version
-  version: 0.1.25
+  version: 0.1.26
 platform: ruby
 authors:
 - Brian Riley
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-03-09 00:00:00.000000000 Z
+date: 2024-03-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: json