adds specs for solr_docs_from_terms*

bin/subjects/solr_docs_from_terms_and_dump_files.rb

@@ -3,55 +3,65 @@
 require "time"
 require "authority_browse"
 
-dumpfile = ARGV.shift
-termsfile = ARGV.shift
-outfile = ARGV.shift
+module SubjectToSolrDocsWrapper
+  def self.run(dumpfile, termsfile, outfile)
+    warn "Loading the dumpfile. 500k entries, each dot is 100k"
+    s = Time.now
+    subjects = AuthorityBrowse::LocSKOSRDF::Subject::Subjects.load(dumpfile)
 
-$stderr.sync = true
+    t = Time.now
+    warn "\nDumpfile loaded in #{(t - s) / 60} minutes"
 
-unless dumpfile && termsfile && outfile
-  warn "\n\nUsage:"
-  warn "    #{$0} <dumpfile> <termsfile> <outfile>"
-  warn "\n\n where:"
-  warn "      _dumpfile_ is produced by the skos_to_dumpfile script"
-  warn "      _termsfiles_ is a tab-delimited set of term-count pairs"
-  warn "      _outfile_ is where you want the resulting solr docs to be"
-  warn ""
-  warn "The whole process balloons up to about 8GB, so allocate accordingly"
-  warn "\n\n"
-  exit 1
-end
+    warn "Load terms-with-counts file. 5.5M-ish terms, each dot is 100k."
+    subjects.load_terms(termsfile)
 
-unless Pathname.new(dumpfile).exist?
-  warn "Dumpfile '#{dumpfile}' can't be found"
-end
+    x = Time.now
+    warn "\nTerms file loaded in #{(x - t) / 60} minutes"
 
-unless Pathname.new(termsfile).exist?
-  warn "Terms file '#{termsfile}' can't be found"
-end
+    warn "Determine counts for the cross-references"
+    subjects.add_xref_counts!
 
-warn "Loading the dumpfile. 500k entries, each dot is 100k"
-s = Time.now
-subjects = AuthorityBrowse::LocSKOSRDF::Subject::Subjects.load(dumpfile)
+    d = Time.now
+    warn "Cross-refs set up in #{d - x} seconds"
 
-t = Time.now
-warn "\nDumpfile loaded in #{(t - s) / 60} minutes"
+    warn "Dump solr docs to '#{outfile}'"
+    Zinzout.zout(outfile) do |out|
+      subjects.each { |s| out.puts s.to_solr_doc.to_json }
+    end
+    o = Time.now
+    warn "Solr documents dumped in #{(o - d) / 60} minutes"
+  end
+end
+
+dumpfile = ARGV.shift
+termsfile = ARGV.shift
+outfile = ARGV.shift
 
-warn "Load terms-with-counts file. 5.5M-ish terms, each dot is 100k."
-subjects.load_terms(termsfile)
+$stderr.sync = true
 
-x = Time.now
-warn "\nTerms file loaded in #{(x - t) / 60} minutes"
+# :nocov:
+if ENV["APP_ENV"] != "test"
+  unless dumpfile && termsfile && outfile
+    warn "\n\nUsage:"
+    warn "    #{$0} <dumpfile> <termsfile> <outfile>"
+    warn "\n\n where:"
+    warn "      _dumpfile_ is produced by the skos_to_dumpfile script"
+    warn "      _termsfiles_ is a tab-delimited set of term-count pairs"
+    warn "      _outfile_ is where you want the resulting solr docs to be"
+    warn ""
+    warn "The whole process balloons up to about 8GB, so allocate accordingly"
+    warn "\n\n"
+    exit 1
+  end
 
-warn "Determine counts for the cross-references"
-subjects.add_xref_counts!
+  unless Pathname.new(dumpfile).exist?
+    warn "Dumpfile '#{dumpfile}' can't be found"
+  end
 
-d = Time.now
-warn "Cross-refs set up in #{d - x} seconds"
+  unless Pathname.new(termsfile).exist?
+    warn "Terms file '#{termsfile}' can't be found"
+  end
 
-warn "Dump solr docs to '#{outfile}'"
-Zinzout.zout(outfile) do |out|
-  subjects.each { |s| out.puts s.to_solr_doc.to_json }
+  SubjectToSolrDocsWrapper.run(dumpfile, termsfile, outfile)
 end
-o = Time.now
-warn "Solr documents dumped in #{(o - d) / 60} minutes"
+# :nocov:

spec/integrations/solr_docs_from_terms_and_dump_files_spec.rb

@@ -0,0 +1,16 @@
+require_relative "../../bin/subjects/solr_docs_from_terms_and_dump_files.rb"
+RSpec.describe SubjectToSolrDocsWrapper do
+  before(:each) do
+    @dumpfile = "spec/fixtures/civil_war_dumpfile.jsonl.gz"
+    @termsfile = "spec/fixtures/civil_war_terms.tsv.gz"
+    @outfile = "tmp/outfile.json"
+  end
+  it "runs something" do
+    expect(File.exist?(@outfile)).to eq(false)
+    described_class.run(@dumpfile, @termsfile, @outfile)
+    expect(File.exist?(@outfile)).to eq(true)
+  end
+  after(:each) do
+    `rm tmp/*`
+  end
+end