rebase

bin/cluster_validator.rb

@@ -0,0 +1,57 @@
+require "cluster"
+require "services"
+Services.mongo!
+
+# Goes through all clusters, checks if they are valid,
+# and prints the first ocn of any invalid cluster to a file.
+
+class ClusterValidator
+  attr_reader :output_path # file name
+  def initialize(buffer_max: 500)
+    @buffer = []
+    @buffer_max = buffer_max
+    # Make an output file in the right place
+    ymd = Time.now.strftime("%Y-%m-%d")
+    dir = Settings.local_report_path
+    FileUtils.mkdir_p(dir)
+    @output_path = "#{dir}/cluster_validator_#{ymd}.txt"
+  end
+
+  def run
+    puts "Logging to #{output_path}"    
+    @outf = File.open(output_path, "w")
+    # Go through each cluster and check if valid.
+    comment "These are ocns of invalid clusters:"
+    Cluster.each do |c|
+      unless c.valid?
+        @buffer << c.ocns.first
+      end
+      empty_buffer?
+    end
+  ensure
+    # Empty bufferfer one last time and be done.
+    empty_buffer?
+    comment "Done"
+    @outf.close
+  end
+
+  private
+
+  # Make a comment in the outfile
+  def comment(msg)
+    @outf.puts "# " + msg
+  end
+
+  # So we don't write to file every iteration.
+  def empty_buffer?
+    if @buffer.size > @buffer_max
+      puts "empty buffer..."
+      @outf.puts(@buffer.join("\n"))
+      @buffer = []
+    end
+  end
+end
+
+if __FILE__ == $0 then
+  ClusterValidator.new.run
+end

spec/cluster_validator_spec.rb

@@ -0,0 +1,65 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+require_relative "../bin/cluster_validator"
+require "mongo_updater"
+
+RSpec.describe ClusterValidator do
+  let(:cluster_validator) { described_class.new }
+  # The output file will have 2 lines, header and footer, even if no body.
+  # So if the output file has 2 lines it is "empty" for the purposes of these tests.
+  let(:empty_file_line_count) { 2 }
+  let(:one_invalid_cluster_line_count) { 3 }
+
+  before(:each) do
+    Cluster.collection.find.delete_many
+  end
+
+  def get_output_lines
+    described_class.new.run
+    File.read(cluster_validator.output_path).split("\n")
+  end
+
+  it "makes an outfile when it runs" do
+    expect(File.exist?(cluster_validator.output_path)).to be false
+    cluster_validator.run
+    expect(File.exist?(cluster_validator.output_path)).to be true
+  end
+
+  it "makes an empty-ish outfile if there are no clusters" do
+    # empty-ish meaning it'll only have the header and footer, which begin with "#".
+    lines = get_output_lines
+    expect(lines.count).to eq empty_file_line_count
+    expect(lines[0]).to start_with("#")
+    expect(lines[1]).to start_with("#")
+  end
+
+  it "does not count valid clusters" do
+    # Start with building a valid cluster, and verify.
+    ocn = 5
+    org = "umich"
+    loc = "loc_1"
+    cluster_tap_save(
+      build(:holding, ocn: ocn, organization: org, local_id: loc),
+      build(:ht_item, :spm, ocns: [ocn], rights: "pd", access: "allow"),
+      build(:commitment, ocn: ocn, organization: org, local_id: loc)
+    )
+    # Verify we have one valid cluster.
+    # Verify it does not count towards the report.
+    expect(Cluster.count).to eq 1
+    expect(Cluster.first.valid?).to be true
+    expect(get_output_lines.count).to eq empty_file_line_count
+    # OK, verified.
+    # Now try to mess up the cluster so it's not valid...
+    # This should bypass the mongoid schema and let us do something "illegal"
+    # and thus create an invalid cluster.
+    MongoUpdater.update_embedded(
+      clusterable: "commitments",
+      matcher: {organization: "umich"},
+      updater: {local_shelving_type: "VALIDATE ME"}
+    )
+    expect(Cluster.count).to eq 1
+    expect(Cluster.first.valid?).to be false
+    expect(get_output_lines.count).to eq one_invalid_cluster_line_count
+  end
+end