Merge pull request #106 from tulibraries/qa

Merge qa and master

lib/cob_index/macros/custom.rb

@@ -11,7 +11,7 @@ module Macros
     module Custom
       ARCHIVE_IT_LINKS = "archive-it.org/collections/"
       NOT_FULL_TEXT = /book review|publisher description|sample text|View cover art|Image|cover image|table of contents/i
-      GENRE_STOP_WORDS = /CD-ROM|CD-ROMs|Compact discs|Computer network resources|Databases|Electronic book|Electronic books|Electronic government information|Electronic journal|Electronic journals|Electronic newspapers|Electronic reference sources|Electronic resource|Full text|Internet resource|Internet resources|Internet videos|Online databases|Online resources|Periodical|Periodicals|Sound recordings|Streaming audio|Streaming video|Video recording|Videorecording|Web site|Web sites|Périodiques|Congrès|Ressource Internet|Périodqiue électronique/i
+      GENRE_STOP_WORDS = /CD-ROM|CD-ROMs|Compact discs|Computer network resources|Databases|Electronic book|Electronic books|Electronic government information|Electronic journal|Electronic journals|Electronic newspapers|Electronic reference sources|Electronic resource|Full text|Internet resource|Internet resources|Internet videos|Online databases|Online resources|Periodical|Periodicals|Sound recordings|Streaming audio|Streaming video|Video recording|Videorecording|Web site|Web sites|Périodiques|Congrès|Ressource Internet|Périodqiue électronique|Online-Publikation|Elektronische Publikation|Diccionarios|Libros electrónicos|Périodique électronique \(Descripteur de forme\)/i
       SEPARATOR = " — "
       A_TO_Z = ("a".."z").to_a.join("")
 
@@ -416,10 +416,9 @@ def normalize_issn
 
       def normalize_lccn
         Proc.new do |rec, acc|
-          orig = acc.dup
-          acc.map! { |x| StdNum::LCCN.normalize(x) }
-          acc << orig
-          acc.flatten!
+          acc.map! { |x|
+            formatted_x = x.gsub("#", " ")
+            StdNum::LCCN.normalize(formatted_x) }
           acc.uniq!
         end
       end

spec/cob_index/macros/custom_spec.rb

@@ -13,78 +13,74 @@
 
 RSpec.describe "custom methods" do
 
-  describe "#four_digit_year(field):" do
-    describe "#four_digit_year(field)" do
-      context "when field is nil" do
-        it "returns nil" do
-          expect(four_digit_year nil).to eq(nil)
-        end
+  describe "#four_digit_year(field)" do
+    context "when field is nil" do
+      it "returns nil" do
+        expect(four_digit_year nil).to eq(nil)
       end
+    end
 
-      context "when given an empty string" do
-        it "returns nil" do
-          expect(four_digit_year "").to eq(nil)
-          expect(four_digit_year "\n").to eq(nil)
-          expect(four_digit_year "\n\n").to eq(nil)
-          expect(four_digit_year "      ").to eq(nil)
-        end
+    context "when given an empty string" do
+      it "returns nil" do
+        expect(four_digit_year("")).to eq(nil)
+        expect(four_digit_year("\n")).to eq(nil)
+        expect(four_digit_year("\n\n")).to eq(nil)
+        expect(four_digit_year("      ")).to eq(nil)
       end
+    end
 
-      context "when contains Roman Numerals" do
-        it "returns nil" do
-          expect(four_digit_year "MCCXLV").to eq(nil)
-        end
+    context "when contains Roman Numerals" do
+      it "returns nil" do
+        expect(four_digit_year "MCCXLV").to eq(nil)
       end
+    end
 
-      it "returns nil for [n.d.],''" do
-        expect(four_digit_year '[n.d.],""').to eq(nil)
-      end
+    it 'returns nil for [n.d.],""' do
+      expect(four_digit_year '[n.d.],""').to eq(nil)
+    end
 
-      it "extracts year from MCCXLV [1745],1745" do
-        expect(four_digit_year "MCCXLV [1745],1745").to eq("1745")
-      end
+    it "extracts year from MCCXLV [1745],1745" do
+      expect(four_digit_year "MCCXLV [1745],1745").to eq("1745")
+    end
 
-      it "extracts the first possible 4 digit numeral" do
-        expect(four_digit_year "1918-1966.,1918   ").to eq("1918")
-      end
+    it "extracts the first possible 4 digit numeral" do
+      expect(four_digit_year "1918-1966.,1918   ").to eq("1918")
+    end
 
-      it "extracts the first possible 4 digit numeral" do
-        expect(four_digit_year "'18-1966.,1918   ").to eq("1966")
-        expect(four_digit_year "c1993.,1993").to eq("1993")
-        expect(four_digit_year "©2012,2012").to eq("2012")
-      end
+    it "extracts the first possible 4 digit numeral" do
+      expect(four_digit_year "18-1966.,1918   ").to eq("1966")
+      expect(four_digit_year "c1993.,1993").to eq("1993")
+      expect(four_digit_year "©2012,2012").to eq("2012")
     end
   end
 
-  describe "#to_marc_normalized" do
-    describe "#flank(field)" do
-      let(:input) {}
-      subject { Traject::Macros::Custom.flank input }
-      context "nil" do
-        it "returns an empty string" do
-          expect(subject).to be_nil
-        end
+  describe "#flank(field)" do
+    let(:input) {}
+    subject { Traject::Macros::Custom.flank input }
+    context "nil" do
+      it "returns an empty string" do
+        expect(subject).to be_nil
       end
+    end
 
-      context "empty string" do
-        let(:input) { "" }
-        it "returns an empty string" do
-          expect(subject).to eq("")
-        end
+    context "empty string" do
+      let(:input) { "" }
+      it "returns an empty string" do
+        expect(subject).to eq("")
       end
+    end
 
-      context "non empty string" do
-        let(:input) { "foo" }
-        it "returns a flanked string" do
-          expect(subject).to eq("matchbeginswith foo matchendswith")
-        end
+    context "non empty string" do
+      let(:input) { "foo" }
+      it "returns a flanked string" do
+        expect(subject).to eq("matchbeginswith foo matchendswith")
       end
+    end
 
-      context "a string that is flanked" do
-        let(:input) { "matchbeginswith foo matchendswith" }
-        it "does not reflank a string" do
-          expect(subject).to eq(input)
-        end
+    context "a string that is flanked" do
+      let(:input) { "matchbeginswith foo matchendswith" }
+      it "does not reflank a string" do
+        expect(subject).to eq(input)
       end
     end
   end

spec/cob_index/macros/normalize_lccn_spec.rb

@@ -0,0 +1,81 @@
+# frozen_string_literal: true
+
+require "rspec"
+require "cob_index/macros/marc_format_classifier"
+require "cob_index/macros/custom"
+require "traject/macros/marc21_semantics"
+
+require "traject/indexer"
+require "marc/record"
+
+include Traject::Macros::MarcFormats
+include Traject::Macros::Custom
+
+RSpec.describe Traject::Macros::Custom do
+  let(:test_class) do
+    Class.new(Traject::Indexer)
+  end
+
+  let(:records) { Traject::MarcReader.new(file, subject.settings).to_a }
+
+  let(:file) { File.new("spec/fixtures/marc_files/#{path}") }
+
+  let(:record) { MARC::XMLReader.new(StringIO.new(record_text)).first }
+
+  subject { test_class.new }
+
+  describe "#normalize_lccn" do
+    before do
+      stub_const("ENV", ENV.to_hash.merge("SOLR_DISABLE_UPDATE_DATE_CHECK" => "false"))
+      subject.instance_eval do
+        to_field("lccn_display", Traject::Macros::Marc21.extract_marc("010ab", separator: nil), &normalize_lccn)
+        settings do
+          provide "marc_source.type", "xml"
+        end
+      end
+    end
+
+    context "The simple case" do
+      let(:record_text) { '
+        <record>
+        <datafield ind1=" " ind2=" " tag="010">
+        <subfield code="a">87014950</subfield>
+        </datafield>
+        </record>
+      ' }
+
+      it "keeps the number as is" do
+        expect(subject.map_record(record)).to eq("lccn_display" => [ "87014950" ])
+      end
+    end
+
+    context "LCCN includes a # symbol" do
+      let(:record_text) { '
+        <record>
+        <datafield ind1=" " ind2=" " tag="010">
+        <subfield code="a">sn#00061556</subfield>
+        </datafield>
+        </record>
+      ' }
+
+      it "removes the # symbol and empty spaces" do
+        expect(subject.map_record(record)).to eq("lccn_display" => [ "sn00061556" ])
+      end
+    end
+
+    context "LCCN includes a / symbol" do
+      let(:record_text) { '
+        <record>
+        <datafield ind1=" " ind2=" " tag="010">
+        <subfield code="a">25004346#//r822</subfield>
+        </datafield>
+        </record>
+      ' }
+
+      it "removes the / and all text to the right of it" do
+        expect(subject.map_record(record)).to eq("lccn_display" => [ "25004346" ])
+      end
+    end
+
+  end
+end

spec/fixtures/marc_files/genre_facet_examples.xml

@@ -14,7 +14,7 @@
     <datafield ind1=' ' ind2='0' tag='650'>
       <subfield code='a'>Rock musicians</subfield>
       <subfield code='z'>England</subfield>
-      <subfield code='v'>Electronic book</subfield>
+      <subfield code='v'>Périodique électronique (Descripteur de forme)</subfield>
     </datafield>
   </record>
 </collection>