Pseudonymizer: Gracefully handle empty pseudo entries

https://gitlab.com/gitlab-org/gitlab-ee/merge_requests/8213
removed the pseudo fields for all entries, but the code was not
updated to handle this.

Discovered in https://gitlab.com/gitlab-com/gl-infra/infrastructure/issues/5744

ee/changelogs/unreleased/sh-fix-pseudonymizer.yml

+---
+title: 'Pseudonymizer: Gracefully handle empty pseudo entries'
+merge_request: 9044
+author:
+type: fixed

ee/lib/pseudonymizer/dumper.rb

@@ -86,7 +86,7 @@ module Pseudonymizer
       type_results = type_results.map do |c|
         data_type = c.sql_type
 
-        if table_config[:pseudo].include?(c.name)
+        if table_config[:pseudo]&.include?(c.name)
           data_type = "character varying"
         end
 

ee/lib/pseudonymizer/filter.rb

@@ -27,6 +27,8 @@ module Pseudonymizer
     private
 
     def pseudo_fields(whitelisted, pseudonymized)
+      return [] unless pseudonymized
+
       pseudo_extra_fields = pseudonymized - whitelisted
       pseudo_extra_fields.each do |field|
         Rails.logger.warn("#{self.class.name} extraneous pseudo: #{@table}.#{field} is not whitelisted and will be ignored.")

ee/spec/lib/pseudonymizer/dumper_spec.rb

@@ -18,9 +18,48 @@ describe Pseudonymizer::Dumper do
     FileUtils.rm_rf(base_dir)
   end
 
-  describe 'Pseudo tables' do
+  describe '#tables_to_csv' do
+    let(:column_names) { %w(id name path description) }
+
+    def decode_project_csv(project_table_file)
+      columns = []
+      project_data = []
+
+      Zlib::GzipReader.open(project_table_file) do |gz|
+        csv = CSV.new(gz, headers: true)
+        # csv.shift # read the header row
+        project_data = csv.gets
+        columns = csv.headers
+      end
+
+      [columns, project_data]
+    end
+
+    context 'with nil pseudo fields' do
+      before do
+        pseudo.config[:tables] = {
+          projects: {
+            whitelist: column_names,
+            pseudo: nil
+          }
+        }
+      end
+
+      it 'outputs valid values' do
+        project_table_file = pseudo.tables_to_csv[0]
+        columns, project_data = decode_project_csv(project_table_file)
+
+        # check if CSV columns are correct
+        expect(columns).to include(*column_names)
+
+        column_names.each do |column|
+          expect(project_data[column].to_s).to eq(project[column].to_s)
+        end
+      end
+    end
+
+    context 'with pseudo fields' do
       it 'outputs project tables to csv' do
-      column_names = %w(id name path description)
         pseudo.config[:tables] = {
           projects: {
             whitelist: column_names,
@@ -34,14 +73,7 @@ describe Pseudonymizer::Dumper do
         project_table_file = pseudo.tables_to_csv[0]
         expect(project_table_file).to end_with("projects.csv.gz")
 
-      columns = []
-      project_data = []
-      Zlib::GzipReader.open(project_table_file) do |gz|
-        csv = CSV.new(gz, headers: true)
-        # csv.shift # read the header row
-        project_data = csv.gets
-        columns = csv.headers
-      end
+        columns, project_data = decode_project_csv(project_table_file)
 
         # check if CSV columns are correct
         expect(columns).to include(*column_names)
@@ -65,6 +97,7 @@ describe Pseudonymizer::Dumper do
         pseudo.tables_to_csv
       end
     end
+  end
 
   describe "manifest is valid" do
     it "all tables exist" do