etl 0.9.5.rc1

data/test/performance/delimited.ctl

@@ -0,0 +1,30 @@
+# puts "executing delimited.ctl"
+
+source :in, {
+  :file => 'delimited.txt',
+  :parser => :csv
+}, 
+[ 
+  :first_name,
+  :last_name,
+  :ssn,
+  {
+    :name => :age,
+    :type => :integer
+  },
+  :sex
+]
+
+transform :ssn, :sha1
+transform(:ssn){ |v| v[0,24] }
+transform :sex, :decode, {:decode_table_path => 'delimited_decode.txt'}
+
+destination :out, {
+  :file => 'delimited.out.txt'
+},
+{
+  :order => [:first_name, :last_name, :name, :ssn, :age, :sex],
+  :virtual => {
+    :name => Proc.new { |row| "#{row[:first_name]} #{row[:last_name]}" }
+  }
+}

data/test/processor_test.rb

@@ -0,0 +1,44 @@
+require File.dirname(__FILE__) + '/test_helper'
+
+class Person < ActiveRecord::Base
+end
+
+# Test pre- and post-processors
+class ProcessorTest < Test::Unit::TestCase
+  # Test bulk import functionality
+  
+  context "the bulk import processor" do
+    should "should import successfully" do
+      assert_nothing_raised { do_bulk_import }
+      assert_equal 3, Person.count
+      assert_equal "Foxworthy", Person.find(2).last_name
+    end
+  end
+  
+  def test_bulk_import_with_empties
+    # this test ensure that one column with empty value will still allow
+    # the row to be imported
+    # this doesn't apply to the id column though - untested
+    assert_nothing_raised { do_bulk_import('bulk_import_with_empties.txt') }
+    assert_equal 3, Person.count
+    assert Person.find(2).last_name.blank?
+  end
+
+  def test_truncate
+    # TODO: implement test
+  end
+  
+  private
+  
+  def do_bulk_import(file = 'bulk_import.txt')
+    control = ETL::Control::Control.new(File.join(File.dirname(__FILE__), 'delimited.ctl'))
+    configuration = {
+      :file => "data/#{file}",
+      :truncate => true,
+      :target => :data_warehouse,
+      :table => 'people'
+    }
+    processor = ETL::Processor::BulkImportProcessor.new(control, configuration)
+    processor.process
+  end
+end

data/test/row_processor_test.rb

@@ -0,0 +1,17 @@
+require File.dirname(__FILE__) + '/test_helper'
+
+# Test row processors
+class RowProcessorTest < Test::Unit::TestCase
+  def test_copy_field_processor
+    
+  end
+  def test_hierarchy_exploder_processor
+    
+  end
+  def test_rename_processor
+    
+  end
+  def test_sequence_processor
+    
+  end
+end

data/test/scd_test.rb

@@ -0,0 +1,257 @@
+require File.dirname(__FILE__) + '/test_helper'
+
+class ScdTest < Test::Unit::TestCase
+  context "when working with a slowly changing dimension" do
+    setup do
+      @connection = ETL::Engine.connection(:data_warehouse)
+      @connection.delete("DELETE FROM person_dimension")
+      @end_of_time = DateTime.parse('9999-12-31 00:00:00')
+    end
+    context "of type 1" do
+      context "on run 1" do
+        setup do
+          do_type_1_run(1)
+        end
+        should "insert record" do
+          assert_equal 1, count_bobs
+        end
+        should "set the original address" do
+          assert_boston_address(find_bobs.first)
+        end
+        should "set the original id" do
+          assert_equal 1, find_bobs.first.id
+        end
+        should "skip the load if there is no change" do
+          do_type_1_run(1)
+          lines = lines_for('scd_test_type_1.txt')
+          assert lines.empty?, "scheduled load expected to be empty, was #{lines.size} records"
+        end
+      end
+      context "on run 2" do
+        setup do
+          do_type_1_run(1)
+          do_type_1_run(2)
+        end
+        should "delete the old record" do
+          assert_equal 1, count_bobs, "new record created, but old not deleted: #{find_bobs.inspect}"
+        end
+        should "update the address" do
+          assert_los_angeles_address(find_bobs.last)
+        end
+        should "keep id" do
+          assert_equal 1, find_bobs.first.id
+        end
+        should "only change once even if run again" do
+          do_type_1_run(2)
+          assert_equal 1, count_bobs
+          lines = lines_for('scd_test_type_1.txt')
+          assert lines.empty?, "scheduled load expected to be empty, was #{lines.size} records"
+        end
+        should "revert address on new record" do
+          do_type_1_run(1)
+          assert_boston_address(find_bobs.first)
+        end
+        should "keep record on revert" do
+          do_type_1_run(1)
+          assert_equal 1, count_bobs
+        end
+      end
+    end
+    context "of type 2" do
+      context "on run 1" do
+        setup do
+          do_type_2_run(1)
+        end
+        should "insert record" do
+          assert_equal 1, count_bobs
+        end
+        should "set the original record" do
+          assert_boston_address(find_bobs.first)
+        end
+        should "set the original id" do
+          assert_equal 1, find_bobs.first.id
+        end
+        should "set the effective date" do
+          # doing comparison on strings, as comparison on objects
+          # doesn't consider things equal for some yet to be understood
+          # reason
+          assert_equal current_datetime.to_s, find_bobs.first.effective_date.to_s
+        end
+        should "set the end date" do
+          assert_equal @end_of_time, find_bobs.first.end_date
+        end
+        should "set the latest version flag" do
+          assert find_bobs.first.latest_version?
+        end
+        should "skip the load if there is no change" do
+          do_type_2_run(1)
+          assert_equal 1, find_bobs.last.id, "scheduled load expected to be empty"
+        end
+        
+      end
+      context "on run 2" do
+        setup do
+          do_type_2_run(1)
+          do_type_2_run(2)
+        end
+        should "insert new record" do
+          assert_equal 2, count_bobs
+        end
+        should "keep the primary key of the original version" do
+          assert_not_nil find_bobs.detect { |bob| 1 == bob.id }
+        end
+        should "increment the primary key for the new version" do
+          assert_not_nil find_bobs.detect { |bob| 2 == bob.id }
+        end
+        should "expire the old record" do
+          original_bob = find_bobs.detect { |bob| 1 == bob.id }
+          new_bob = find_bobs.detect { |bob| 2 == bob.id }
+          assert_equal new_bob.effective_date, original_bob.end_date
+        end
+        should "keep the address for the expired record" do
+          assert_boston_address(find_bobs.detect { |bob| 1 == bob.id })
+        end
+        should "update the address on the new record" do
+          assert_los_angeles_address(find_bobs.detect { |bob| 2 == bob.id })
+        end
+        should "activate the new record" do
+          # doing comparison on strings, as comparison on objects
+          # doesn't consider things equal for some yet to be understood
+          # reason
+          assert_equal current_datetime.to_s, find_bobs.detect { |bob| 2 == bob.id }.effective_date.to_s
+        end
+        should "set the end date for the new record" do
+          assert_equal @end_of_time, find_bobs.detect { |bob| 2 == bob.id }.end_date
+        end
+        should "shift the latest version" do
+          original_bob = find_bobs.detect { |bob| 1 == bob.id }
+          new_bob = find_bobs.detect { |bob| 2 == bob.id }
+          assert !original_bob.latest_version?
+          assert new_bob.latest_version?
+        end
+        should "only execute a change once" do
+          do_type_2_run(2)
+          assert_equal 2, count_bobs, "scheduled load expected to be empty"
+        end
+        should "insert new records on revert" do
+          do_type_2_run(1)
+          assert_equal 3, count_bobs
+        end
+        should "update address on new record on revert" do
+          do_type_2_run(1)
+          assert_boston_address(find_bobs.detect { |bob| 3 == bob.id })
+        end
+        should "only delete one row on an scd change" do
+          # Two records right now
+          assert_equal 2, count_bobs
+          do_type_2_run(1) # put third version in (same as first version, but that's irrelevant)
+          # was failing because first and second versions were being deleted.
+          assert_equal 3, count_bobs
+        end
+      end
+      context "on non sdc fields that change" do
+        setup do
+          do_type_2_run_with_only_city_state_zip_scd(1)
+          do_type_2_run_with_only_city_state_zip_scd(2)
+        end
+        should "not create an extra record" do
+          do_type_2_run_with_only_city_state_zip_scd(3)
+          assert_equal 2, count_bobs
+        end
+        should "keep id" do
+          do_type_2_run_with_only_city_state_zip_scd(3)
+          assert_not_nil find_bobs.detect { |bob| 2 == bob.id }
+        end
+        should "keep dates" do
+          old_bob = find_bobs.detect { |bob| 2 == bob.id }
+          do_type_2_run_with_only_city_state_zip_scd(3)
+          new_bob = find_bobs.detect { |bob| 2 == bob.id }
+          assert_equal old_bob.end_date, new_bob.end_date
+          assert_equal old_bob.effective_date, new_bob.effective_date
+        end
+        should "keep the latest version flag" do
+          do_type_2_run_with_only_city_state_zip_scd(3)
+          assert find_bobs.detect { |bob| 2 == bob.id }.latest_version?
+        end
+        should "treat non scd fields like type 1 fields" do
+          do_type_2_run_with_only_city_state_zip_scd(3)
+          assert_los_angeles_address(find_bobs.detect { |bob| 2 == bob.id }, "280 Pine Street")
+        end
+        should "skip load when there is no change" do
+          do_type_2_run_with_only_city_state_zip_scd(2)
+          assert_equal 2, count_bobs, "scheduled load expected to be empty"
+        end
+      end
+    end
+  end
+  
+  def do_type_2_run(run_num)
+    ENV['run_number'] = run_num.to_s
+    assert_nothing_raised do
+      run_ctl_file("scd_test_type_2.ctl")
+    end
+  end
+  
+  def do_type_2_run_with_only_city_state_zip_scd(run_num)
+    ENV['type_2_scd_fields'] = Marshal.dump([:city, :state, :zip_code])
+    do_type_2_run(run_num)
+  end
+  
+  def do_type_1_run(run_num)
+    ENV['run_number'] = run_num.to_s
+    assert_nothing_raised do
+      run_ctl_file("scd_test_type_1.ctl")
+    end
+  end
+  
+  def lines_for(file)
+    File.readlines(File.dirname(__FILE__) + "/output/#{file}")
+  end
+  
+  def run_ctl_file(file)
+    ETL::Engine.process(File.dirname(__FILE__) + "/#{file}")
+  end
+  
+  def count_bobs
+    @connection.select_value(
+      "SELECT count(*) FROM person_dimension WHERE first_name = 'Bob' and last_name = 'Smith'").to_i
+  end
+  
+  def find_bobs
+    bobs = @connection.select_all(
+      "SELECT * FROM person_dimension WHERE first_name = 'Bob' and last_name = 'Smith'")
+    bobs.each do |bob|
+      def bob.id
+        self["id"].to_i
+      end
+      def bob.effective_date
+        DateTime.parse(self["effective_date"])
+      end
+      def bob.end_date
+        DateTime.parse(self["end_date"])
+      end
+      def bob.latest_version?
+        ActiveRecord::ConnectionAdapters::Column.value_to_boolean(self["latest_version"])
+      end
+    end
+    bobs
+  end
+  
+  def current_datetime
+    DateTime.parse(Time.now.to_s(:db))
+  end
+  
+  def assert_boston_address(bob, street = "200 South Drive")
+    assert_equal street, bob['address'], bob.inspect
+    assert_equal "Boston", bob['city'], bob.inspect
+    assert_equal "MA", bob['state'], bob.inspect
+    assert_equal "32123", bob['zip_code'], bob.inspect
+  end
+  
+  def assert_los_angeles_address(bob, street = "1010 SW 23rd St")
+    assert_equal street, bob['address'], bob.inspect
+    assert_equal "Los Angeles", bob['city'], bob.inspect
+    assert_equal "CA", bob['state'], bob.inspect
+    assert_equal "90392", bob['zip_code'], bob.inspect
+  end
+end

data/test/screen_test.rb

@@ -0,0 +1,9 @@
+require File.dirname(__FILE__) + '/test_helper'
+
+class ScreenTest < Test::Unit::TestCase
+  def test_screen
+    assert_raises(SystemExit) do
+      ETL::Engine.process(File.dirname(__FILE__) + '/screen_test_fatal.ctl')
+    end
+  end
+end

data/test/source_test.rb

@@ -0,0 +1,154 @@
+require File.dirname(__FILE__) + '/test_helper'
+
+class Person < ActiveRecord::Base
+end
+class SourceTest < Test::Unit::TestCase
+  
+  context "source" do
+    should "set store_locally to true by default" do
+      assert_equal true, Source.new(nil, { :store_locally => true }, nil).store_locally
+    end
+    
+    should "let the user set store_locally to true" do
+      assert_equal true, Source.new(nil, { :store_locally => true }, nil).store_locally
+    end
+
+    should "let the user set store_locally to false" do
+      assert_equal false, Source.new(nil, { :store_locally => false }, nil).store_locally
+    end
+  end
+  
+  context "a file source" do
+    context "with delimited data" do
+      setup do
+        control = ETL::Control::Control.parse(File.dirname(__FILE__) + '/delimited.ctl')
+        configuration = {
+          :file => 'data/delimited.txt',
+          :parser => :csv
+        }
+        definition = self.definition + [:sex]
+    
+        source = ETL::Control::FileSource.new(control, configuration, definition)
+        @rows = source.collect { |row| row }
+      end
+      should "find 3 rows in the delimited file" do
+        assert_equal 3, @rows.length
+      end
+    end
+  end
+  
+  context "a file source with a glob" do
+    setup do
+      control = ETL::Control::Control.parse(File.dirname(__FILE__) + '/multiple_delimited.ctl')
+      configuration = {
+        :file => 'data/multiple_delimited_*.txt',
+        :parser => :csv
+      }
+
+      source = ETL::Control::FileSource.new(control, configuration, definition)
+      @rows = source.collect { |row| row }
+    end
+    should "find 6 rows in total" do
+      assert_equal 6, @rows.length
+    end
+  end
+  
+  context "a file source with an absolute path" do
+    setup do
+      FileUtils.cp(File.dirname(__FILE__) + '/data/delimited.txt', '/tmp/delimited_abs.txt')
+
+      control = ETL::Control::Control.parse(File.dirname(__FILE__) + 
+        '/delimited_absolute.ctl')
+      configuration = {
+        :file => '/tmp/delimited_abs.txt',
+        :parser => :csv
+      }
+      definition = self.definition + [:sex]
+
+      source = ETL::Control::FileSource.new(control, configuration, definition)
+      @rows = source.collect { |row| row }
+    end
+    should "find 3 rows" do
+      assert_equal 3, @rows.length
+    end
+  end
+  
+  context "multiple sources" do
+    setup do
+      control = ETL::Control::Control.parse(File.dirname(__FILE__) + 
+        '/multiple_source_delimited.ctl')
+      @rows = control.sources.collect { |source| source.collect { |row| row }}.flatten!
+    end
+    should "find 12 rows" do
+      assert_equal 12, @rows.length
+    end
+  end
+  
+  context "a database source" do
+    setup do
+      control = ETL::Control::Control.parse(File.dirname(__FILE__) + '/delimited.ctl')
+      configuration = {
+        :database => 'etl_unittest',
+        :target => :operational_database,
+        :table => 'people',
+      }
+      definition = [ 
+        :first_name,
+        :last_name,
+        :ssn,
+      ]
+      @source = ETL::Control::DatabaseSource.new(control, configuration, definition)
+    end
+    should "set the local file for extraction storage" do
+      assert_match %r{source_data/localhost/etl_unittest/people/\d+.csv}, @source.local_file.to_s
+    end
+    should "find 1 row" do
+      Person.delete_all
+      assert_equal 0, Person.count
+      Person.create!(:first_name => 'Bob', :last_name => 'Smith', :ssn => '123456789')
+      assert_equal 1, Person.count
+      rows = @source.collect { |row| row }
+      assert_equal 1, rows.length
+    end
+  end
+  
+  context "a file source with an xml parser" do
+    setup do
+      control = ETL::Control::Control.parse(File.dirname(__FILE__) + 
+        '/xml.ctl')
+      @rows = control.sources.collect{ |source| source.collect { |row| row }}.flatten!
+    end
+    should "find 2 rows" do
+      assert_equal 2, @rows.length
+    end
+  end
+
+  context "a model source" do
+    setup do
+      control = ETL::Control::Control.parse(File.dirname(__FILE__) + '/model_source.ctl')
+      configuration = {
+
+      }
+      definition = [
+        :first_name,
+        :last_name,
+        :ssn
+      ]
+    end
+    should_eventually "find n rows" do
+      
+    end
+  end
+  
+  def definition
+    [ 
+      :first_name,
+      :last_name,
+      :ssn,
+      {
+        :name => :age,
+        :type => :integer
+      }
+    ]
+  end
+end