fa8e4699-854f-4a43-adc1-7e92603d9eea

Build # fa8e4699-854f-4a43-adc1-7e92603d9eea

Build Type

circleci

Committed by Hector Correa

Commit Message

Rubocop nitpicking

Pull Request Pull Request #441: Indexing to a new collection

Run Details

70 of 70 new or added lines in 2 files covered. (100.0%)

2073 of 2216 relevant lines covered (93.55%)

170.04 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

74.26

/spec/lib/describe_indexer_spec.rb

# frozen_string_literal: true

# rubocop:disable Metrics/BlockLength
RSpec.describe DescribeIndexer do
  describe 'indexing a single record' do
    let(:single_item) { file_fixture("bitklavier_binaural.json").read }
    let(:indexer) do
      described_class.new(rss_url: "file://whatever.rss")
    end
    let(:indexed_record) do
      Blacklight.default_index.connection.delete_by_query("*:*")
      Blacklight.default_index.connection.commit
      indexer.index_one(single_item)
      response = Blacklight.default_index.connection.get 'select', params: { q: '*:*' }
      response["response"]["docs"].first
    end

    context "basic fields" do
      ##
      # The id is based on the DOI
      # A doi of 10.34770/r75s-9j74 will become doi-10-34770-r75s-9j74
      it "id" do
        expect(indexed_record["id"]).to eq "doi-10-34770-r75s-9j74"
      end

      it "stores a copy of the full JSON in CDATA" do
        stored_json = indexed_record["pdc_describe_json_ss"]
        parsed_json = JSON.parse(stored_json)
        expect(parsed_json["resource"]["titles"][0]["title"]).to eq "bitKlavier Grand Sample Library—Binaural Mic Image"
      end

      it "author" do
        expect(indexed_record["author_tesim"]).to eq ['Trueman, Daniel']
      end

      it "description" do
        description = "The bitKlavier Grand consists"
        expect(indexed_record["description_tsim"].first.include?(description)).to be true
      end

      it "contributors" do
        expect(indexed_record["contributor_tsim"].include?("Villalta, Andrés")).to eq true
        expect(indexed_record["contributor_tsim"].include?("Chou, Katie")).to eq true
        expect(indexed_record["contributor_tsim"].include?("Ayres, Christien")).to eq true
        expect(indexed_record["contributor_tsim"].include?("Wang, Matthew")).to eq true
      end

      it "title" do
        # title includes all titles
        main_title = "bitKlavier Grand Sample Library—Binaural Mic Image"
        alt_title = "alter title for bitKlavier"
        expect(indexed_record["title_tesim"].include?(main_title)).to eq true
        expect(indexed_record["title_tesim"].include?(alt_title)).to eq true
        # alt title does not include the main title
        expect(indexed_record["alternative_title_tesim"].include?(main_title)).to eq false
        expect(indexed_record["alternative_title_tesim"].include?(alt_title)).to eq true
      end

      it "rights" do
        expect(indexed_record["rights_name_ssi"]).to eq "GNU General Public License"
        expect(indexed_record["rights_uri_ssi"]).to eq "https://www.gnu.org/licenses/gpl-3.0.en.html"
      end

      it "keywords" do
        expect(indexed_record["subject_all_ssim"].include?("keyword1")).to eq true
        expect(indexed_record["subject_all_ssim"].include?("keyword2")).to eq true
        expect(indexed_record["subject_all_ssim"].include?("keyword3")).to eq true
      end

      it "collection tag" do
        expect(indexed_record["collection_tag_ssim"].include?("Humanities")).to eq true
        expect(indexed_record["collection_tag_ssim"].include?("Something else")).to eq true
      end

      it "community" do
        expect(indexed_record["community_name_ssi"]).to eq "Research Data"
      end

      it "genre_ssim" do
        expect(indexed_record["genre_ssim"].first).to eq "Dataset"
      end

      it "issue_date_ssim" do
        expect(indexed_record["issue_date_ssim"].first).to eq "2021"
      end

      it "publisher_ssim" do
        expect(indexed_record["publisher_ssim"].first).to eq "Princeton University"
      end

      xit 'referenced_by' do
        expect(result['referenced_by_ssim'].first).to eq 'https://arxiv.org/abs/1903.06605'
      end
    end

    context "uris" do
      it "stores full URL for ARK and DOI" do
        expect(indexed_record["uri_ssim"].include?("http://arks.princeton.edu/ark:/88435/dsp015999n653h")).to eq true
        expect(indexed_record["uri_ssim"].include?("https://doi.org/10.34770/r75s-9j74")).to eq true
      end
    end

    context "files" do
      it "stores file detailed information" do
        files = JSON.parse(indexed_record['files_ss'])
        file1 = files.find { |file| file["name"] == "file1.jpg" }
        file2 = files.find { |file| file["name"] == "file2.txt" }
        expect(file1["size"]).to eq "316781"
        expect(file1["url"]).to eq "https://g-5beea4.90d4e.bd7c.data.globus.org/pdc-describe-staging-postcuration/10.80021/3m1k-6036/122/file1.jpg"
        expect(file2["size"]).to eq "396003"
      end
    end

    context "all text catch all field" do
      it "indexes the file name in the all text catch all field" do
        files = JSON.parse(indexed_record['files_ss'])
        file_name = File.basename(files.first["name"])
        response = Blacklight.default_index.connection.get 'select', params: { q: file_name }
        expect(response["response"]["numFound"]).to eq 1

        response = Blacklight.default_index.connection.get 'select', params: { q: "non-existing-value" }
        expect(response["response"]["numFound"]).to eq 0
      end
    end
  end

  describe 'indexing an RSS feed from PDC Describe' do
    let(:rss_feed) { file_fixture("works.rss").read }
    let(:resource1) { file_fixture("bitklavier_binaural.json").read }
    let(:resource2) { file_fixture("sowing_the_seeds.json").read }
    let(:rss_url_string) { "https://pdc-describe-prod.princeton.edu/describe/works.rss" }
    let(:indexer) { described_class.new(rss_url: rss_url_string) }

    it "has a traject indexer" do
      expect(indexer.traject_indexer).to be_instance_of Traject::Indexer::NokogiriIndexer
    end

    context 'indexing to solr' do
      before do
        Blacklight.default_index.connection.delete_by_query("*:*")
        Blacklight.default_index.connection.commit
        stub_request(:get, "https://pdc-describe-prod.princeton.edu/describe/works.rss")
          .to_return(status: 200, body: rss_feed, headers: {})
        stub_request(:get, "https://pdc-describe-prod.princeton.edu/describe/works/6.json")
          .to_return(status: 200, body: resource1, headers: {})
        stub_request(:get, "https://pdc-describe-prod.princeton.edu/describe/works/20.json")
          .to_return(status: 200, body: resource2, headers: {})
      end

      it "sends items to solr" do
        response = Blacklight.default_index.connection.get 'select', params: { q: '*:*' }
        expect(response["response"]["numFound"]).to eq 0

        # If index_pdc_describe == false, do not index PDC Describe.
        # This is a safety measure so we don't index in production until we're ready
        # See config/pdc_discovery.yml to change this setting for real
        Rails.configuration.pdc_discovery.index_pdc_describe = false
        indexer.index
        response = Blacklight.default_index.connection.get 'select', params: { q: '*:*' }
        expect(response["response"]["numFound"]).to eq 0

        # If index_pdc_describe == true, DO index PDC Describe.
        Rails.configuration.pdc_discovery.index_pdc_describe = true
        indexer.index
        response = Blacklight.default_index.connection.get 'select', params: { q: '*:*' }
        expect(response["response"]["numFound"]).to eq 2
      end
    end
  end
end
# rubocop:enable Metrics/BlockLength

1	# frozen_string_literal: true
2
3	# rubocop:disable Metrics/BlockLength
4	RSpec.describe DescribeIndexer do	1✔
5	describe 'indexing a single record' do	1✔
6	let(:single_item) { file_fixture("bitklavier_binaural.json").read }	17✔
7	let(:indexer) do	1✔
8	described_class.new(rss_url: "file://whatever.rss")	16✔
9	end
10	let(:indexed_record) do	1✔
11	Blacklight.default_index.connection.delete_by_query(":")	16✔
12	Blacklight.default_index.connection.commit	16✔
13	indexer.index_one(single_item)	16✔
14	response = Blacklight.default_index.connection.get 'select', params: { q: ':' }	×
15	response["response"]["docs"].first	×
16	end
17
18	context "basic fields" do	1✔
19	##
20	# The id is based on the DOI
21	# A doi of 10.34770/r75s-9j74 will become doi-10-34770-r75s-9j74
22	it "id" do	1✔
23	expect(indexed_record["id"]).to eq "doi-10-34770-r75s-9j74"	1✔
24	end
25
26	it "stores a copy of the full JSON in CDATA" do	1✔
27	stored_json = indexed_record["pdc_describe_json_ss"]	1✔
28	parsed_json = JSON.parse(stored_json)	×
29	expect(parsed_json["resource"]["titles"][0]["title"]).to eq "bitKlavier Grand Sample Library—Binaural Mic Image"	×
30	end
31
32	it "author" do	1✔
33	expect(indexed_record["author_tesim"]).to eq ['Trueman, Daniel']	1✔
34	end
35
36	it "description" do	1✔
37	description = "The bitKlavier Grand consists"	1✔
38	expect(indexed_record["description_tsim"].first.include?(description)).to be true	1✔
39	end
40
41	it "contributors" do	1✔
42	expect(indexed_record["contributor_tsim"].include?("Villalta, Andrés")).to eq true	1✔
43	expect(indexed_record["contributor_tsim"].include?("Chou, Katie")).to eq true	×
44	expect(indexed_record["contributor_tsim"].include?("Ayres, Christien")).to eq true	×
45	expect(indexed_record["contributor_tsim"].include?("Wang, Matthew")).to eq true	×
46	end
47
48	it "title" do	1✔
49	# title includes all titles
50	main_title = "bitKlavier Grand Sample Library—Binaural Mic Image"	1✔
51	alt_title = "alter title for bitKlavier"	1✔
52	expect(indexed_record["title_tesim"].include?(main_title)).to eq true	1✔
53	expect(indexed_record["title_tesim"].include?(alt_title)).to eq true	×
54	# alt title does not include the main title
55	expect(indexed_record["alternative_title_tesim"].include?(main_title)).to eq false	×
56	expect(indexed_record["alternative_title_tesim"].include?(alt_title)).to eq true	×
57	end
58
59	it "rights" do	1✔
60	expect(indexed_record["rights_name_ssi"]).to eq "GNU General Public License"	1✔
61	expect(indexed_record["rights_uri_ssi"]).to eq "https://www.gnu.org/licenses/gpl-3.0.en.html"	×
62	end
63
64	it "keywords" do	1✔
65	expect(indexed_record["subject_all_ssim"].include?("keyword1")).to eq true	1✔
66	expect(indexed_record["subject_all_ssim"].include?("keyword2")).to eq true	×
67	expect(indexed_record["subject_all_ssim"].include?("keyword3")).to eq true	×
68	end
69
70	it "collection tag" do	1✔
71	expect(indexed_record["collection_tag_ssim"].include?("Humanities")).to eq true	1✔
72	expect(indexed_record["collection_tag_ssim"].include?("Something else")).to eq true	×
73	end
74
75	it "community" do	1✔
76	expect(indexed_record["community_name_ssi"]).to eq "Research Data"	1✔
77	end
78
79	it "genre_ssim" do	1✔
80	expect(indexed_record["genre_ssim"].first).to eq "Dataset"	1✔
81	end
82
83	it "issue_date_ssim" do	1✔
84	expect(indexed_record["issue_date_ssim"].first).to eq "2021"	1✔
85	end
86
87	it "publisher_ssim" do	1✔
88	expect(indexed_record["publisher_ssim"].first).to eq "Princeton University"	1✔
89	end
90
91	xit 'referenced_by' do	1✔
92	expect(result['referenced_by_ssim'].first).to eq 'https://arxiv.org/abs/1903.06605'	×
93	end
94	end
95
96	context "uris" do	1✔
97	it "stores full URL for ARK and DOI" do	1✔
98	expect(indexed_record["uri_ssim"].include?("http://arks.princeton.edu/ark:/88435/dsp015999n653h")).to eq true	1✔
99	expect(indexed_record["uri_ssim"].include?("https://doi.org/10.34770/r75s-9j74")).to eq true	×
100	end
101	end
102
103	context "files" do	1✔
104	it "stores file detailed information" do	1✔
105	files = JSON.parse(indexed_record['files_ss'])	1✔
106	file1 = files.find { \|file\| file["name"] == "file1.jpg" }	×
107	file2 = files.find { \|file\| file["name"] == "file2.txt" }	×
108	expect(file1["size"]).to eq "316781"	×
109	expect(file1["url"]).to eq "https://g-5beea4.90d4e.bd7c.data.globus.org/pdc-describe-staging-postcuration/10.80021/3m1k-6036/122/file1.jpg"	×
110	expect(file2["size"]).to eq "396003"	×
111	end
112	end
113
114	context "all text catch all field" do	1✔
115	it "indexes the file name in the all text catch all field" do	1✔
116	files = JSON.parse(indexed_record['files_ss'])	1✔
117	file_name = File.basename(files.first["name"])	×
118	response = Blacklight.default_index.connection.get 'select', params: { q: file_name }	×
119	expect(response["response"]["numFound"]).to eq 1	×
120
121	response = Blacklight.default_index.connection.get 'select', params: { q: "non-existing-value" }	×
122	expect(response["response"]["numFound"]).to eq 0	×
123	end
124	end
125	end
126
127	describe 'indexing an RSS feed from PDC Describe' do	1✔
128	let(:rss_feed) { file_fixture("works.rss").read }	2✔
129	let(:resource1) { file_fixture("bitklavier_binaural.json").read }	2✔
130	let(:resource2) { file_fixture("sowing_the_seeds.json").read }	2✔
131	let(:rss_url_string) { "https://pdc-describe-prod.princeton.edu/describe/works.rss" }	3✔
132	let(:indexer) { described_class.new(rss_url: rss_url_string) }	3✔
133
134	it "has a traject indexer" do	1✔
135	expect(indexer.traject_indexer).to be_instance_of Traject::Indexer::NokogiriIndexer	1✔
136	end
137
138	context 'indexing to solr' do	1✔
139	before do	1✔
140	Blacklight.default_index.connection.delete_by_query(":")	1✔
141	Blacklight.default_index.connection.commit	1✔
142	stub_request(:get, "https://pdc-describe-prod.princeton.edu/describe/works.rss")	1✔
143	.to_return(status: 200, body: rss_feed, headers: {})
144	stub_request(:get, "https://pdc-describe-prod.princeton.edu/describe/works/6.json")	1✔
145	.to_return(status: 200, body: resource1, headers: {})
146	stub_request(:get, "https://pdc-describe-prod.princeton.edu/describe/works/20.json")	1✔
147	.to_return(status: 200, body: resource2, headers: {})
148	end
149
150	it "sends items to solr" do	1✔
151	response = Blacklight.default_index.connection.get 'select', params: { q: ':' }	1✔
152	expect(response["response"]["numFound"]).to eq 0	1✔
153
154	# If index_pdc_describe == false, do not index PDC Describe.
155	# This is a safety measure so we don't index in production until we're ready
156	# See config/pdc_discovery.yml to change this setting for real
157	Rails.configuration.pdc_discovery.index_pdc_describe = false	1✔
158	indexer.index	1✔
159	response = Blacklight.default_index.connection.get 'select', params: { q: ':' }	1✔
160	expect(response["response"]["numFound"]).to eq 0	1✔
161
162	# If index_pdc_describe == true, DO index PDC Describe.
163	Rails.configuration.pdc_discovery.index_pdc_describe = true	1✔
164	indexer.index	1✔
165	response = Blacklight.default_index.connection.get 'select', params: { q: ':' }	1✔
166	expect(response["response"]["numFound"]).to eq 2	1✔
167	end
168	end
169	end
170	end
171	# rubocop:enable Metrics/BlockLength

pulibrary / pdc_discovery / fa8e4699-854f-4a43-adc1-7e92603d9eea

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous