blacklight-marc 5.10.0 → 6.0.0

data/lib/railties/solr_marc.rake

@@ -4,7 +4,24 @@
 # config.properties defaults to config/demo_config.properties (in the plugin, not the rails app)
 
 require 'fileutils'
-
+require 'traject'
+# monkey patch
+class Traject::SolrJsonWriter
+  def record_id_from_context(context)
+    marc_id = context.source_record && context.source_record['001'] && context.source_record['001'].value
+    output_id = context.output_hash["id"]
+
+    return [marc_id, output_id].compact.join("/")
+  end
+end
+module SolrMarc
+  def self.indexer
+    @indexer ||= ::MarcIndexer.new
+  end
+  def self.indexer=(ix)
+    @indexer = ix
+  end
+end
 namespace :solr do
   namespace :marc do
     
@@ -24,17 +41,17 @@ namespace :solr do
     task :index => "index:work"
     namespace :index do
 
-      task :work do
+      task :work => :environment do
         solrmarc_arguments = compute_arguments        
 
         # If no marc records given, display :info task
         if  (ENV["NOOP"] || (!solrmarc_arguments["MARC_FILE"]))                    
           Rake::Task[ "solr:marc:index:info" ].execute
-        else        
-          commandStr = solrmarc_command_line( solrmarc_arguments )
-          puts commandStr
-          puts
-          `#{commandStr}`
+        else
+          require './app/models/marc_indexer' unless defined?(MarcIndexer)
+          open(solrmarc_arguments['MARC_FILE']) do |io|
+            SolrMarc.indexer.process(io)
+          end
         end
         
       end # work
@@ -59,14 +76,6 @@ namespace :solr do
 
      Note that SolrMarc search path includes directory of config_path,
      so translation_maps and index_scripts dirs will be found there. 
-  
-  SOLRMARC_JAR_PATH: #{solrmarc_arguments[:solrmarc_jar_path]}
-  
-  SOLRMARC_MEM_ARGS: #{solrmarc_arguments[:solrmarc_mem_arg]}
-  
-  SolrMarc command that will be run:
-  
-  #{solrmarc_command_line(solrmarc_arguments)}
   EOS
       end
     end # index
@@ -134,18 +143,6 @@ def compute_arguments
   return arguments
 end
 
-def solrmarc_command_line(arguments)
-  cmd = "java #{arguments[:solrmarc_mem_arg]} "
-  cmd += " -Dsolr.hosturl=#{arguments[:solr_url]} " unless arguments[:solr_url].blank?
-  
-  cmd += " -Dsolrmarc.solr.war.path=#{arguments[:solr_war_path]}" unless arguments[:solr_war_path].blank?
-  cmd += " -Dsolr.path=#{arguments[:solr_path]}" unless arguments[:solr_path].blank?
-  
-  cmd += " -jar #{arguments[:solrmarc_jar_path]} #{arguments[:config_properties_path]} #{arguments["MARC_FILE"]}"  
-  return cmd  
-end
-
-
 def locate_path(*subpath_fragments)
   local_root = File.expand_path File.join(File.dirname(__FILE__), '..', '..')  
   subpath = subpath_fragments.join('/')

data/spec/features/bookmarks_spec.rb

@@ -0,0 +1,17 @@
+require 'spec_helper'
+
+describe "Bookmark tools" do
+  it "should have an endnote export" do
+    visit solr_document_path('2007020969')
+    click_button 'Bookmark'
+    visit "/bookmarks.endnote?q="
+    expect(page).to have_content " %@ 9780743297790"
+  end
+
+  it "should have a refworks export" do
+    visit solr_document_path('2007020969')
+    click_button 'Bookmark'
+    visit "/bookmarks.refworks_marc_txt?q="
+    expect(page).to have_content "LEADER 01490cam a2200361 a 4500001      2007020969"
+  end
+end

data/spec/features/librarian_view_spec.rb

@@ -2,7 +2,7 @@ require 'spec_helper'
 
 describe "Librarian view" do
   it "should show marc fields" do
-    visit catalog_path('2009373513')
+    visit solr_document_path('2009373513')
     click_link "Librarian View"
     expect(page).to have_content "Librarian View"
     expect(page).to have_content "LEADER 01213nam a22003614a 4500"

data/spec/features/record_view_spec.rb

@@ -0,0 +1,13 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+
+describe "Record View" do
+  describe "Tools" do
+    it "should have a refworks link" do
+      visit solr_document_path('2008305903')
+      expect(page).to have_link "Export to Refworks"
+      link = find_link("Export to Refworks")
+      expect(link['href']).to include CGI.escape(solr_document_path('2008305903', format: :refworks_marc_txt))
+    end
+  end
+end

data/spec/features/search_formats_spec.rb

@@ -0,0 +1,31 @@
+# -*- encoding : utf-8 -*-
+
+require 'spec_helper'
+
+describe "Search Formats" do
+  before do
+    # Get all the fields from Solr, so the #index documents have the MARC field (to trigger appropriate
+    # export formats)
+    CatalogController.blacklight_config.default_solr_params[:fl] = '*'
+  end
+
+  it "should return new-line separated RIS records" do
+    visit "/catalog.endnote?q=" 
+    expect(page.status_code).to eq 200
+    rmt_regex = /^%0/
+    expect(page).to have_content "%A Bstan-ʼdzin-rgya-mtsho,"
+    expect(page).to have_content "%@ 9789573908678"
+    expect(page.body).to match rmt_regex
+    expect(page.body.scan(rmt_regex).length).to eq 10
+  end
+
+  it "should return new-line separated MARC records" do
+    visit "/catalog.refworks_marc_txt?q=" 
+    expect(page.status_code).to eq 200
+    rmt_regex = /LEADER .+\n(\d\d\d .. .+\n)+/
+    expect(page).to have_content "LEADER 01221cam a22002534a 4500001"
+    expect(page).to have_content "LEADER 01127cam a22002895a 4500001"
+    expect(page.body).to match rmt_regex
+    expect(page.body.scan(rmt_regex).length).to eq 10 
+  end
+end

data/spec/helpers/blacklight_marc_helper_spec.rb

@@ -23,4 +23,24 @@ describe BlacklightMarcHelper do
       expect(helper.render_endnote_texts([one, two])).to eq "one\ntwo\n"
     end
   end
+  require 'spec_helper'
+
+  describe '#document_action_path' do
+    before do
+      allow(helper).to receive_messages(controller_name: 'catalog')
+    end
+
+    let(:document_action_config) { Blacklight::Configuration::ToolConfig.new(tool_config) }
+    let(:document) { SolrDocument.new(id: '123') }
+
+    subject { helper.document_action_path(document_action_config, id: document) }
+
+    context "for endnote" do
+      let(:tool_config) { { if: :render_refworks_action?, partial: "document_action",
+        name: :endnote, key: :endnote, path: :single_endnote_catalog_path } }
+
+      it { is_expected.to eq '/catalog/123.endnote' }
+    end
+  end
+
 end

data/spec/lib/indexer/dewey_spec.rb

@@ -0,0 +1,86 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+
+describe Blacklight::Marc::Indexer::Dewey do
+  let(:test_class) do
+    c = Class.new(Blacklight::Marc::Indexer)
+    c.send :include, Blacklight::Marc::Indexer::Dewey
+    c
+  end
+  subject {test_class.new}
+  before do
+    $:.unshift './test_support/config'
+    subject.instance_eval do
+      to_field 'dewey100', dewey_facets(base: :hundreds, translation_map: 'test_dewey')
+      to_field 'dewey010', dewey_facets(base: :tens, translation_map: 'test_dewey')
+      to_field 'dewey_text', dewey_text
+      settings do
+        store "writer_class_name", "Traject::JsonWriter"
+      end
+    end
+  end
+  let(:dewey) do
+    MARC::DataField.new('082',' ',' ',['a','343/13'])
+  end
+  let(:huey) do
+    MARC::DataField.new('082',' ',' ',['a','lol/wut'])
+  end
+  describe '#dewey_facets' do
+    it do
+      expect(subject).to respond_to(:dewey_facets)
+    end
+    it 'should map valid dewey' do
+      record = double('Record')
+      # just return bytes for control field, subfield array for data
+      expect(record).to receive(:fields).with(['082','880']).exactly(2).times.and_return([dewey]) 
+      val = []
+      subject.dewey_facets(base: :hundreds, translation_map: 'test_dewey').call(record,val,subject)
+      expect(val).to eql(['300s - Social Sciences'])
+      val = []
+      subject.dewey_facets(base: :tens, translation_map: 'test_dewey').call(record,val,subject)
+      expect(val).to eql(['340s - Law'])
+    end
+    it 'should ignore invalid dewey' do
+      record = double('Record')
+      # just return bytes for control field, subfield array for data
+      expect(record).to receive(:fields).with(['082','880']).exactly(2).times.and_return([huey]) 
+      val = []
+      subject.dewey_facets(base: :hundreds, translation_map: 'test_dewey').call(record,val,subject)
+      expect(val).to eql([])
+      val = []
+      subject.dewey_facets(base: :tens, translation_map: 'test_dewey').call(record,val,subject)
+      expect(val).to eql([])
+    end
+  end
+  describe '#dewey_text' do
+    it do
+      expect(subject).to respond_to(:dewey_text)
+    end
+    it 'should generate formatted, unmapped text' do
+      record = double('Record')
+      # just return bytes for control field, subfield array for data
+      expect(record).to receive(:fields).with(['082','880']).and_return([dewey]) 
+      val = []
+      subject.dewey_text.call(record,val,subject)
+      expect(val).to eql(['343 13'])
+    end
+    it 'should ignore invalid dewey' do
+      record = double('Record')
+      # just return bytes for control field, subfield array for data
+      expect(record).to receive(:fields).with(['082','880']).times.and_return([huey]) 
+      val = []
+      subject.dewey_text.call(record,val,subject)
+      expect(val).to eql([])
+    end
+  end
+  describe '#valid_dewey?' do
+    it do
+      expect(subject).to respond_to(:valid_dewey?)
+    end
+    it 'shoud audit values' do
+      expect(subject.valid_dewey?'111.4').to be_truthy
+      expect(subject.valid_dewey?'111.a').to be_truthy
+      expect(subject.valid_dewey?'a11.4').to be_falsy
+    end
+  end
+end

data/spec/lib/indexer/formats_spec.rb

@@ -0,0 +1,65 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+#require 'marc'
+describe Blacklight::Marc::Indexer::Formats do
+  let(:test_class) do
+    c = Class.new(Blacklight::Marc::Indexer)
+    c.send :include, Blacklight::Marc::Indexer::Formats
+    c
+  end
+
+  subject {test_class.new}
+  before do
+    subject.instance_eval do
+      to_field 'format', get_format
+      settings do
+        provide "writer_class_name", "Traject::JsonWriter"
+      end
+    end
+  end
+
+  describe '#get_format' do
+    let(:electronic) do
+      MARC::DataField.new('245',' ',' ',['h','electronic resource'])
+    end
+    it 'should respond' do
+      expect(subject).to respond_to(:get_format)
+    end
+    it 'should map for 245h fields' do
+      record = double('Record')
+      # just return bytes for control field, subfield array for data
+      expect(record).to receive(:fields).with(['245','880']).and_return([electronic]) 
+      val = []
+      subject.get_format.call(record,val,subject)
+      expect(val).to eql(['Electronic'])
+    end
+    it 'should map for 007 fields' do
+      record = double('Record')
+      allow(record).to receive(:fields).with(["245", "880"]).and_return([])
+      expect(record).to receive(:fields).with(["007", "880"]).and_return([MARC::ControlField.new('007','CA')])
+      val = []
+      subject.get_format.call(record,val,subject)
+      expect(val).to eql(['TapeCartridge'])
+    end
+    it 'should map for leader' do
+      record = double('Record')
+      allow(record).to receive(:fields).with(["245", "880"]).and_return([])
+      allow(record).to receive(:fields).with(["007", "880"]).and_return([])
+      allow(record).to receive(:[]).with('001').and_return(MARC::ControlField.new('001',''))
+      expect(record).to receive(:leader).and_return('012345am')
+      val = []
+      subject.get_format.call(record,val,subject)
+      expect(val).to eql(['Book'])
+    end
+    it 'should default' do
+      record = double('Record')
+      allow(record).to receive(:fields).with(["245", "880"]).and_return([])
+      allow(record).to receive(:fields).with(["007", "880"]).and_return([])
+      allow(record).to receive(:[]).with('001').and_return(MARC::ControlField.new('001',''))
+      allow(record).to receive(:leader).and_return('012345##')
+      val = []
+      subject.get_format.call(record,val,subject)
+      expect(val).to eql(['Unknown'])
+    end
+  end
+end

data/spec/lib/tasks/solr_marc_task_spec.rb

@@ -23,36 +23,32 @@ describe "solr:marc:*" do
       Rake.application.rake_require "../lib/railties/solr_marc"
       Rake::Task.define_task(:environment)
     end
-        
+
     describe 'solr:marc:index_test_data' do        
       it 'should print out usage using NOOP=true' do
         root = Rails.root
         ENV['NOOP'] = "true"
-        o = capture_stdout do      
-          @rake['solr:marc:index_test_data'].invoke      
+        o = capture_stdout do
+          @rake['solr:marc:index_test_data'].invoke
         end
-        
-        expect(o).to match(Regexp.escape("SolrMarc command that will be run:"))
+
+        expect(o).to match(Regexp.escape("Possible environment variables, with settings as invoked"))
       end    
     end
     
     describe "solr:marc:index" do
-      it "should produce proper java command" do
-        # can NOT figure out how to actually run solr:marc:index and trap
-        # it's backtick system call. So we'll run solr:marc:index:info and
-        # just test it's dry run output
+      after do
+        SolrMarc.indexer= nil
+      end
+      it "should index the test data configured" do
+        indexer = double(MarcIndexer)
+        allow(indexer).to receive(:process).with("dummy.mrc")
+        SolrMarc.indexer= indexer
+        #expect(indexer).to receive(:process).with("dummy.mrc")
         ENV["MARC_FILE"] = "dummy.mrc"
         output = capture_stdout do
-          @rake['solr:marc:index:info'].invoke
+          @rake['solr:marc:index:work'].invoke
         end
-        output =~ /SolrMarc command that will be run:\n\s*\n\s*(.*)\n/
-        java_cmd = $1
-        
-        expect(java_cmd).not_to be_nil
-        expect(java_cmd).to match "java -Xmx512m"
-        expect(java_cmd).to match /-jar .*\/SolrMarc\.jar/
-        expect(java_cmd).to match "#{Rails.root}/config/SolrMarc/config-test.properties dummy.mrc"
-        expect(java_cmd).to match "-Dsolr.hosturl=http://127.0.0.1:[0-9]{2,5}/solr"
       end
       
     end  

data/spec/lib/traject_indexer_spec.rb

@@ -0,0 +1,132 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+def music_record
+"<collection xmlns=\"http://www.loc.gov/MARC21/slim\">
+<record>
+  <leader>01828cjm a2200409 a 4500</leader>
+  <controlfield tag=\"001\">a4768316</controlfield>
+  <controlfield tag=\"003\">SIRSI</controlfield>
+  <controlfield tag=\"007\">sd fungnnmmned</controlfield>
+  <controlfield tag=\"008\">020117p20011990xxuzz    h              d</controlfield>
+
+  <datafield tag=\"245\" ind1=\"0\" ind2=\"0\">
+    <subfield code=\"a\">Music for horn</subfield>
+    <subfield code=\"h\">[sound recording] /</subfield>
+    <subfield code=\"c\">Brahms, Beethoven, von Krufft.</subfield>
+  </datafield>
+
+  <datafield tag=\"260\" ind1=\" \" ind2=\" \">
+    <subfield code=\"a\">[United States] :</subfield>
+    <subfield code=\"b\">Harmonia Mundi USA,</subfield>
+    <subfield code=\"c\">p2001.</subfield>
+  </datafield>
+
+  <datafield tag=\"700\" ind1=\"1\" ind2=\" \">
+    <subfield code=\"a\">Greer, Lowell.</subfield>
+  </datafield>
+
+  <datafield tag=\"700\" ind1=\"1\" ind2=\" \">
+    <subfield code=\"a\">Lubin, Steven.</subfield>
+  </datafield>
+
+  <datafield tag=\"700\" ind1=\"1\" ind2=\" \">
+    <subfield code=\"a\">Chase, Stephanie,</subfield>
+    <subfield code=\"d\">1957-</subfield>
+  </datafield>
+
+  <datafield tag=\"700\" ind1=\"1\" ind2=\"2\">
+    <subfield code=\"a\">Brahms, Johannes,</subfield>
+    <subfield code=\"d\">1833-1897.</subfield>
+    <subfield code=\"t\">Trios,</subfield>
+    <subfield code=\"m\">piano, violin, horn,</subfield>
+    <subfield code=\"n\">op. 40,</subfield>
+    <subfield code=\"r\">E? major.</subfield>
+  </datafield>
+
+  <datafield tag=\"700\" ind1=\"1\" ind2=\"2\">
+    <subfield code=\"a\">Beethoven, Ludwig van,</subfield>
+    <subfield code=\"d\">1770-1827.</subfield>
+    <subfield code=\"t\">Sonatas,</subfield>
+    <subfield code=\"m\">horn, piano,</subfield>
+    <subfield code=\"n\">op. 17,</subfield>
+    <subfield code=\"r\">F major.</subfield>
+  </datafield>
+
+  <datafield tag=\"700\" ind1=\"1\" ind2=\"2\">
+    <subfield code=\"a\">Krufft, Nikolaus von,</subfield>
+    <subfield code=\"d\">1779-1818.</subfield>
+    <subfield code=\"t\">Sonata,</subfield>
+    <subfield code=\"m\">horn, piano,</subfield>
+    <subfield code=\"r\">F major.</subfield>
+  </datafield>
+</record>
+</collection>"
+end
+module TestIndexer
+  def initialize
+    super
+    settings do
+      # have to override the defaults!
+      store "marc_source.type", "xml"
+      store "writer_class_name", "TestIndexer::Writer"
+    end
+    to_field "id", trim(extract_marc("001"), :first => true)
+    to_field "format", extract_marc('007[0-1]') do |rec, acc|
+      legacy_formats = Traject::TranslationMap.new("test_formats")
+      acc << rec.leader[6..7]
+      acc << rec.leader[7]
+      acc.map! {|x| legacy_formats[x]}
+      acc.compact! # get rid of nils
+      acc << 'Unknown' # the default
+      acc.replace Array(acc[0]) # just take the first
+    end
+    to_field "mapped", map_value(literal('k'),translation_map:"test_formats")
+  end
+  class Writer
+    def self.accumulator=(acc)
+      @acc = acc
+    end
+    def self.accumulator
+      @acc
+    end
+  attr_reader :settings
+
+    def initialize(argSettings)
+    end
+
+
+    def serialize(context);end
+    def close;end
+    def put(context)
+      puts context.output_hash.inspect
+      TestIndexer::Writer.accumulator << context.output_hash
+    end
+  end
+end
+
+describe Blacklight::Marc::Indexer do
+  before do
+    $:.unshift './test_support/config'
+  end
+  subject { Blacklight::Marc::Indexer.new }
+  it do
+    expect(subject).to respond_to(:process)
+  end
+  context "XML indexing" do
+    let(:test_class) do
+      c = Class.new(Blacklight::Marc::Indexer)
+      c.send :include, TestIndexer
+    end
+    subject {test_class.new}
+    it do
+      io = StringIO.new(music_record)
+      vals = []
+      TestIndexer::Writer.accumulator= vals
+      subject.writer_class = TestIndexer::Writer
+      subject.process(io)
+      json = vals.first
+      expect(json['format']).to eql(['Sound Disc'])
+      expect(json['mapped']).to eql(['Image'])
+    end
+  end
+end