add specs for `Pseudonymizer::Pager`

lib/pseudonymizer/dumper.rb

-require 'openssl'
-require 'digest'
-require 'csv'
-require 'yaml'
-
 module Pseudonymizer
-  class Pager
-    PAGE_SIZE = ENV.fetch('PSEUDONYMIZER_BATCH', 100_000)
-
-    def initialize(table, columns)
-      @table = table
-      @columns = columns
-    end
-
-    def pages(&block)
-      if @columns.include?("id")
-        # optimize the pagination using WHERE id > ?
-        pages_per_id(&block)
-      else
-        # fallback to `LIMIT ? OFFSET ?` when "id" is unavailable
-        pages_per_offset(&block)
-      end
-    end
-
-    def pages_per_id(&block)
-      id_offset = 0
-
-      loop do
-        # a page of results
-        results = ActiveRecord::Base.connection.exec_query(<<-SQL.squish)
-          SELECT #{@columns.join(",")}
-          FROM #{@table}
-          WHERE id > #{id_offset}
-          ORDER BY id
-          LIMIT #{PAGE_SIZE}
-        SQL
-        Rails.logger.debug("#{self.class.name} fetch ids [#{id_offset}, +#{PAGE_SIZE}[")
-        break if results.empty?
-
-        id_offset = results.last["id"].to_i
-        yield results
-
-        break if results.count < PAGE_SIZE
-      end
-    end
-
-    def pages_per_offset(&block)
-      page = 0
-
-      loop do
-        offset = page * PAGE_SIZE
-
-        # a page of results
-        results = ActiveRecord::Base.connection.exec_query(<<-SQL.squish)
-          SELECT #{@columns.join(",")}
-          FROM #{@table}
-          ORDER BY #{@columns.join(",")}
-          LIMIT #{PAGE_SIZE} OFFSET #{offset}
-        SQL
-        Rails.logger.debug("#{self.class.name} fetching offset [#{offset}, #{offset + PAGE_SIZE}[")
-        break if results.empty?
-
-        page += 1
-        yield results
-
-        break if results.count < PAGE_SIZE
-      end
-    end
-  end
-
-  class Anon
-    def initialize(table, whitelisted_fields, pseudonymized_fields)
-      @table = table
-      @pseudo_fields = pseudo_fields(whitelisted_fields, pseudonymized_fields)
-    end
-
-    def anonymize(results)
-      key = Rails.application.secrets[:secret_key_base]
-      digest = OpenSSL::Digest.new('sha256')
-
-      Enumerator.new do |yielder|
-        results.each do |result|
-          @pseudo_fields.each do |field|
-            next if result[field].nil?
-
-            result[field] = OpenSSL::HMAC.hexdigest(digest, key, String(result[field]))
-          end
-          yielder << result
-        end
-      end
-    end
-
-    private
-
-    def pseudo_fields(whitelisted, 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.")
-      end
-
-      pseudonymized & whitelisted
-    end
-  end
-
   class Dumper
     attr_accessor :config, :output_dir
 
@@ -165,12 +62,12 @@ module Pseudonymizer
 
     # yield every results, pagined, anonymized
     def table_page_results(table, whitelist_columns, pseudonymity_columns)
-      anonymizer = Anon.new(table, whitelist_columns, pseudonymity_columns)
+      filter = Filter.new(table, whitelist_columns, pseudonymity_columns)
       pager = Pager.new(table, whitelist_columns)
 
       Enumerator.new do |yielder|
         pager.pages do |page|
-          anonymizer.anonymize(page).each do |result|
+          filter.anonymize(page).each do |result|
             yielder << result
           end
         end

lib/pseudonymizer/filter.rb

+require 'openssl'
+require 'digest'
+
+module Pseudonymizer
+  class Filter
+    def initialize(table, whitelisted_fields, pseudonymized_fields)
+      @table = table
+      @pseudo_fields = pseudo_fields(whitelisted_fields, pseudonymized_fields)
+    end
+
+    def anonymize(results)
+      key = Rails.application.secrets[:secret_key_base]
+      digest = OpenSSL::Digest.new('sha256')
+
+      Enumerator.new do |yielder|
+        results.each do |result|
+          @pseudo_fields.each do |field|
+            next if result[field].nil?
+
+            result[field] = OpenSSL::HMAC.hexdigest(digest, key, String(result[field]))
+          end
+          yielder << result
+        end
+      end
+    end
+
+    private
+
+    def pseudo_fields(whitelisted, 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.")
+      end
+
+      pseudonymized & whitelisted
+    end
+  end
+end

lib/pseudonymizer/pager.rb

+module Pseudonymizer
+  class Pager
+    PAGE_SIZE = ENV.fetch('PSEUDONYMIZER_BATCH', 100_000)
+
+    def initialize(table, columns)
+      @table = table
+      @columns = columns
+    end
+
+    def pages(&block)
+      if @columns.include?("id")
+        # optimize the pagination using WHERE id > ?
+        pages_per_id(&block)
+      else
+        # fallback to `LIMIT ? OFFSET ?` when "id" is unavailable
+        pages_per_offset(&block)
+      end
+    end
+
+    def pages_per_id(&block)
+      id_offset = 0
+
+      loop do
+        # a page of results
+        results = ActiveRecord::Base.connection.exec_query(<<-SQL.squish)
+          SELECT #{@columns.join(",")}
+          FROM #{@table}
+          WHERE id > #{id_offset}
+          ORDER BY id
+          LIMIT #{PAGE_SIZE}
+        SQL
+        Rails.logger.debug("#{self.class.name} fetch ids [#{id_offset}, +#{PAGE_SIZE}[")
+        break if results.empty?
+
+        id_offset = results.last["id"].to_i
+        yield results
+
+        break if results.count < PAGE_SIZE
+      end
+    end
+
+    def pages_per_offset(&block)
+      offset = 0
+
+      loop do
+        # a page of results
+        results = ActiveRecord::Base.connection.exec_query(<<-SQL.squish)
+          SELECT #{@columns.join(",")}
+          FROM #{@table}
+          ORDER BY #{@columns.join(",")}
+          LIMIT #{PAGE_SIZE} OFFSET #{offset}
+        SQL
+        Rails.logger.debug("#{self.class.name} fetching offset [#{offset}, #{offset + PAGE_SIZE}[")
+        break if results.empty?
+
+        offset += PAGE_SIZE
+        yield results
+
+        break if results.count < PAGE_SIZE
+      end
+    end
+  end
+end

spec/lib/pseudonymizer/pager_spec.rb

+require 'spec_helper'
+
+describe Pseudonymizer::Pager do
+  class Counter
+    @count = 0
+
+    def increment(*args)
+      self.count += 1
+    end
+  end
+
+  let(:page_size) { 1 }
+  let!(:projects) { create_list(:project, 10) }
+  subject { described_class.new("projects", whitelisted_columns) }
+
+  before do
+    stub_const("Pseudonymizer::Pager::PAGE_SIZE", page_size)
+  end
+
+  shared_examples "yield results in page" do
+    it do
+      page_count = 0
+      result_count = 0
+
+      subject.pages do |page|
+        result_count += page.count
+        page_count += 1
+      end
+
+      expect(result_count).to eq(projects.count)
+      expect(page_count).to eq(projects.count / page_size)
+    end
+  end
+
+  context "`id` column is present" do
+    let(:whitelisted_columns) { %w(id name) }
+
+    describe "#pages" do
+      it "delegates to #pages_per_id" do
+        expect(subject).to receive(:pages_per_id)
+
+        subject.pages {|page| nil}
+      end
+
+      include_examples "yield results in page"
+    end
+  end
+
+  context "`id` column is missing" do
+    let(:whitelisted_columns) { %w(name) }
+
+    describe "#pages" do
+      it "delegates to #pages_per_offset" do
+        expect(subject).to receive(:pages_per_offset)
+
+        subject.pages {|page| nil}
+      end
+
+      include_examples "yield results in page"
+    end
+  end
+end