balinterdi-activewarehouse-etl 0.9.2

data/test/control_test.rb

@@ -0,0 +1,43 @@
+require File.dirname(__FILE__) + '/test_helper'
+
+class ControlTest < Test::Unit::TestCase
+  # Test the ability to parse control files.
+  def test_parse
+    assert_nothing_raised do
+      Dir.glob('*.ctl') do |f|
+        ETL::Control::Control.parse(File.join(File.dirname(__FILE__), f))
+      end
+    end
+  end
+  
+  def test_bad_control_raises_error
+    assert_raise ETL::ControlError do
+      ETL::Control::Control.resolve(0)
+    end
+  end
+  
+  def test_resolve_control_object
+    assert_nothing_raised do
+      ETL::Control::Control.resolve(ETL::Control::Control.parse(File.join(File.dirname(__FILE__), 'delimited.ctl')))
+    end
+  end
+  
+  def test_set_error_threshold
+    assert_nothing_raised do
+      ETL::Engine.process(File.join(File.dirname(__FILE__), 'errors.ctl'))
+    end
+  end
+  
+  def test_bad_processor_name
+    assert_raise ETL::ControlError do
+      s = "before_write :chunky_monkey"
+      ETL::Control::Control.parse_text(s)
+    end
+  end
+  
+  def test_dependencies
+    s = "depends_on 'foo', 'bar'"
+    control = ETL::Control::Control.parse_text(s)
+    assert_equal control.dependencies, ['foo','bar']
+  end
+end

data/test/data/apache_combined_log.txt

@@ -0,0 +1,3 @@
+127.0.0.1 - frank [10/Oct/2000:13:55:36 -0700] "GET /apache_pb.gif HTTP/1.0" 200 2326 "http://www.example.com/start.html" "Mozilla/4.08 [en] (Win98; I ;Nav)"
+127.0.0.1 - bob [11/Oct/2000:05:22:02 -0700] "GET /apache_pb.gif HTTP/1.1" 200 2326 "http://www.foo.com/" "Mozilla/4.08 [en] (Win98; I ;Nav)"
+127.0.0.1 - bob [11/Oct/2000:05:52:31 -0700] "GET /apache_pb.gif HTTP/1.1" 200 2326 "-" "Mozilla/4.08 [en] (Win98; I ;Nav)"

data/test/data/bulk_import.txt

@@ -0,0 +1,3 @@
+1,Chris,Smith,111223333
+2,Jim,Foxworthy,444332222
+3,Brian,Collingsworth,123443435

data/test/data/bulk_import_with_empties.txt

@@ -0,0 +1,3 @@
+Chris,Smith,111223333
+Jim,,444332222
+Brian,Collingsworth,123443435

data/test/data/decode.txt

@@ -0,0 +1,3 @@
+M:Male
+F:Female
+:Unknown

data/test/data/delimited.txt

@@ -0,0 +1,3 @@
+Chris,Smith,111223333,24,M
+Jim,Foxworthy,444332222,51,M
+Brian,Collingsworth,123443435,10,M

data/test/data/encode_source_latin1.txt

@@ -0,0 +1,2 @@
+�ph�m�re has accents.
+let's encode them.

data/test/data/fixed_width.txt

@@ -0,0 +1,3 @@
+Bob      Smith       12344555523
+Jane     Doe         98766211145
+AbcdefghiJklmnopqrstu12345678920

data/test/data/multiple_delimited_1.txt

@@ -0,0 +1,3 @@
+Chris,Smith,111223333,24
+Jim,Foxworthy,444332222,51
+Brian,Collingsworth,123443435,10

data/test/data/multiple_delimited_2.txt

@@ -0,0 +1,3 @@
+Bob,Jones,444223333,28
+Tom,Allen,324001232,33
+Jesse,Baker,555443333,21

data/test/data/people.txt

@@ -0,0 +1,3 @@
+Bob,Smith
+Jane,Doe
+Chris,Cornell

data/test/data/sax.xml

@@ -0,0 +1,14 @@
+<?xml version="1.0"?>
+
+<people>
+  <person age="24">
+    <first_name>Bob</first_name>
+    <last_name>Smith</last_name>
+    <social_security_number>123456789</social_security_number>
+  </person>
+  <person age="31">
+    <first_name>John</first_name>
+    <last_name>Doe</last_name>
+    <social_security_number>222114545</social_security_number>
+  </person>
+</people>

data/test/data/xml.xml

@@ -0,0 +1,16 @@
+<?xml version="1.0"?>
+
+<people>
+  <person>
+    <first_name>Bob</first_name>
+    <last_name>Smith</last_name>
+    <social_security_number>123456789</social_security_number>
+    <age>24</age>
+  </person>
+  <person>
+    <first_name>John</first_name>
+    <last_name>Doe</last_name>
+    <social_security_number>222114545</social_security_number>
+    <age>31</age>
+  </person>
+</people>

data/test/database.example.yml

@@ -0,0 +1,18 @@
+operational_database:
+  adapter: mysql
+  username: root
+  host: localhost
+  database: etl_unittest
+  encoding: utf8
+data_warehouse:
+  adapter: mysql
+  username: root
+  host: localhost
+  database: etl_unittest
+  encoding: utf8
+etl_execution:
+  adapter: mysql
+  username: root
+  host: localhost
+  database: etl_unittest_execution
+  encoding: utf8

data/test/database.mysql.yml

@@ -0,0 +1,18 @@
+operational_database:
+  adapter: mysql
+  username: root
+  host: localhost
+  database: etl_unittest
+  encoding: utf8
+data_warehouse:
+  adapter: mysql
+  username: root
+  host: localhost
+  database: etl_unittest
+  encoding: utf8
+etl_execution:
+  adapter: mysql
+  username: root
+  host: localhost
+  database: etl_unittest_execution
+  encoding: utf8

data/test/database.postgres.yml

@@ -0,0 +1,18 @@
+operational_database:
+  adapter: postgresql
+  username: postgres
+  host: localhost
+  database: etl_unittest
+  encoding: utf8
+data_warehouse:
+  adapter: postgresql
+  username: postgres
+  host: localhost
+  database: etl_unittest
+  encoding: utf8
+etl_execution:
+  adapter: postgresql
+  username: postgres
+  host: localhost
+  database: etl_unittest_execution
+  encoding: utf8

data/test/database.yml

@@ -0,0 +1,18 @@
+operational_database:
+  adapter: postgresql
+  username: postgres
+  host: localhost
+  database: etl_unittest
+  encoding: utf8
+data_warehouse:
+  adapter: postgresql
+  username: postgres
+  host: localhost
+  database: etl_unittest
+  encoding: utf8
+etl_execution:
+  adapter: postgresql
+  username: postgres
+  host: localhost
+  database: etl_unittest_execution
+  encoding: utf8

data/test/date_dimension_builder_test.rb

@@ -0,0 +1,96 @@
+require File.dirname(__FILE__) + '/test_helper'
+
+class DateDimensionBuilderTest < Test::Unit::TestCase
+  
+  context "the DateDimensionBuilder" do
+    context "when initialized with defaults" do
+      setup do
+        @builder = ETL::Builder::DateDimensionBuilder.new
+      end
+      should "have a start date of 5 years ago" do
+        assert_equal Time.now.years_ago(5).to_date, @builder.start_date.to_date
+      end
+      should "have an end date of now" do
+        assert_equal Time.now.to_date, @builder.end_date.to_date
+      end
+      should "have an empty of array of holiday indicators" do
+        assert_equal [], @builder.holiday_indicators
+      end
+    end
+    context "when initialized with arguments" do
+      setup do 
+        @start_date = Time.now.years_ago(2)
+        @end_date = Time.now.years_ago(1)
+        @builder = ETL::Builder::DateDimensionBuilder.new(@start_date, @end_date)
+      end
+      should "respect a custom start date" do
+        assert_equal @start_date.to_date, @builder.start_date.to_date
+      end
+      should "respect a custom end date" do
+        assert_equal @end_date.to_date, @builder.end_date.to_date
+      end
+    end
+    context "when building a date dimension using the default settings" do
+      setup do
+        # specific dates required when testing, because leap years affect
+        # how many records are built
+        @start_date = Date.parse('2002-05-19').to_time
+        @end_date = Date.parse('2007-05-19').to_time
+        @builder = ETL::Builder::DateDimensionBuilder.new(@start_date, @end_date)
+        @records = @builder.build
+      end
+      should "build a dimension with the correct number of records" do
+        assert_equal 1827, @records.length
+      end
+      should "have the correct first date" do
+        assert_date_dimension_record_equal(@builder.start_date, @records.first)
+      end
+    end
+    context "when building a date dimension with a fiscal year offset month" do
+      should_eventually "respect the fiscal year offset month" do
+        
+      end
+    end
+  end
+  
+  def assert_date_dimension_record_equal(date, record)
+    real_date = date
+    date = date.to_time
+    assert_equal date.strftime("%m/%d/%Y"), record[:date]
+    assert_equal date.strftime("%B %d,%Y"), record[:full_date_description]
+    assert_equal date.strftime("%A"), record[:day_of_week]
+    assert_equal date.day, record[:day_number_in_calendar_month]
+    assert_equal date.yday, record[:day_number_in_calendar_year]
+    assert_equal date.day, record[:day_number_in_fiscal_month]
+    assert_equal date.fiscal_year_yday, record[:day_number_in_fiscal_year]
+    assert_equal "Week #{date.week}", record[:calendar_week]
+    assert_equal date.week, record[:calendar_week_number_in_year]
+    assert_equal date.strftime("%B"), record[:calendar_month_name]
+    assert_equal date.month, record[:calendar_month_number_in_year]
+    assert_equal date.strftime("%Y-%m"), record[:calendar_year_month]
+    assert_equal "Q#{date.quarter}", record[:calendar_quarter]
+    assert_equal date.quarter, record[:calendar_quarter_number_in_year]
+    assert_equal "#{date.strftime('%Y')}-#{record[:calendar_quarter]}", record[:calendar_year_quarter]
+    assert_equal "#{date.year}", record[:calendar_year]
+    assert_equal "FY Week #{date.fiscal_year_week}", record[:fiscal_week]
+    assert_equal date.fiscal_year_week, record[:fiscal_week_number_in_year]
+    assert_equal date.fiscal_year_month, record[:fiscal_month]
+    assert_equal date.fiscal_year_month, record[:fiscal_month_number_in_year]
+    assert_equal "FY#{date.fiscal_year}-" + date.fiscal_year_month.to_s.rjust(2, '0'), record[:fiscal_year_month]
+    assert_equal "FY Q#{date.fiscal_year_quarter}", record[:fiscal_quarter]
+    assert_equal "FY#{date.fiscal_year}-Q#{date.fiscal_year_quarter}", record[:fiscal_year_quarter]
+    assert_equal date.fiscal_year_quarter, record[:fiscal_year_quarter_number]
+    assert_equal "FY#{date.fiscal_year}", record[:fiscal_year]
+    assert_equal date.fiscal_year, record[:fiscal_year_number]
+    assert_equal 'Nonholiday', record[:holiday_indicator]
+    assert_equal weekday_indicators[date.wday], record[:weekday_indicator]
+    assert_equal 'None', record[:selling_season]
+    assert_equal 'None', record[:major_event]
+    assert_equal record[:sql_date_stamp], real_date
+  end
+  
+  private
+  def weekday_indicators
+    ['Weekend','Weekday','Weekday','Weekday','Weekday','Weekday','Weekend']
+  end
+end

data/test/delimited.ctl

@@ -0,0 +1,30 @@
+source :in, {
+  :file => 'data/delimited.txt',
+  :parser => {
+    :name => :delimited
+  }
+}, 
+[ 
+  :first_name,
+  :last_name,
+  :ssn,
+  :age,
+  :sex
+]
+
+#transform :age, :type, :type => :number
+transform :ssn, :sha1
+transform(:ssn){ |n, v, row| v[0,24] }
+transform :sex, :decode, {:decode_table_path => 'data/decode.txt'}
+
+destination :out, {
+  :file => 'output/delimited.txt'
+},
+{
+  :order => [:id, :first_name, :last_name, :ssn, :age, :sex, :test, :calc_test],
+  :virtual => {
+    :id => :surrogate_key,
+    :test => "test!",
+    :calc_test => Time.now
+  },
+}

data/test/delimited_absolute.ctl

@@ -0,0 +1,33 @@
+# puts "executing delimited.ctl"
+
+source :in, {
+  :file => '/tmp/delimited_abs.txt',
+  :parser => {
+    :name => :delimited
+  }
+}, 
+[ 
+  :first_name,
+  :last_name,
+  :ssn,
+  {
+    :name => :age,
+    :type => :integer
+  },
+  :sex
+]
+
+transform :ssn, :sha1
+transform(:ssn){ |n, v, row| v[0,24] }
+transform :sex, :decode, {:decode_table_path => 'data/decode.txt'}
+
+destination :out, {
+  :file => 'data/delimited_abs.txt'
+},
+{
+  :order => [:first_name, :last_name, :ssn, :age, :sex, :test, :calc_test],
+  :virtual => {
+    :test => "test!",
+    :calc_test => Time.now
+  }
+}

data/test/delimited_destination_db.ctl

@@ -0,0 +1,25 @@
+# puts "executing delimited.ctl"
+
+source :in, {
+  :file => 'data/delimited.txt',
+  :parser => :delimited
+}, 
+[ 
+  :id,
+  :first_name,
+  :last_name,
+  :ssn
+]
+
+transform :ssn, :sha1
+transform(:ssn){ |v| v[0,24] }
+
+destination :out, {
+  :type => :database,
+  :target => :data_warehouse,
+  :database => 'etl_unittest',
+  :table => 'people',
+},
+{
+  :order => [:id, :first_name, :last_name, :ssn]
+}

data/test/delimited_with_bulk_load.ctl

@@ -0,0 +1,34 @@
+infile = 'data/people.txt'
+outfile = 'output/people.txt'
+
+source :in, {
+  :file => infile,
+  :parser => {
+    :name => :delimited
+  }
+}, 
+[ 
+  :first_name,
+  :last_name,
+]
+
+before_write :surrogate_key, :target => :data_warehouse, :table => 'person_dimension', :column => 'id'
+before_write :check_exists, {
+  :target => :data_warehouse, 
+  :table => 'person_dimension', 
+  :columns => [:first_name, :last_name]
+}
+
+destination :out, {
+  :file => outfile
+},
+{
+  :order => [:id, :first_name, :last_name]
+}
+
+post_process :bulk_import, {
+  :file => outfile,
+  :target => :data_warehouse,
+  :table => 'person_dimension',
+  :order => [:id, :first_name, :last_name]
+}

data/test/destination_test.rb

@@ -0,0 +1,171 @@
+require File.dirname(__FILE__) + '/test_helper'
+
+class Person < ActiveRecord::Base
+end
+
+class BadDestination < ETL::Control::Destination
+  def initialize(control, configuration, mapping)
+    super
+  end
+end
+
+# Test the functionality of destinations
+class DestinationTest < Test::Unit::TestCase
+  # Test a file destination
+  def test_file_destination
+    outfile = 'output/test_file_destination.txt'
+    outfile2 = 'output/test_file_destination.2.txt'
+    row = ETL::Row[ :address => '123 SW 1st Street', :city => 'Melbourne', 
+      :state => 'Florida', :country => 'United States' ]
+    row_needs_escape = ETL::Row[ :address => "Allen's Way", 
+      :city => 'Some City', :state => 'Some State', :country => 'Mexico' ]
+    control = ETL::Control::Control.parse(File.dirname(__FILE__) + 
+      '/delimited.ctl')
+
+    # First define a basic configuration to check defaults
+    configuration = { 
+      :file => outfile, 
+      :buffer_size => 0,
+    }
+    mapping = { 
+      :order => [:address, :city, :state, :country, :country_code],
+      :virtual => {
+        :country_code => Proc.new do |r|
+          {
+            'United States' => 'US',
+            'Mexico' => 'MX'
+          }[r[:country]]
+        end
+      }
+    }
+    
+    dest = ETL::Control::FileDestination.new(control, configuration, mapping)    
+    dest.write(row)
+    dest.write(row_needs_escape)
+    dest.close
+    
+    configuration[:file] = outfile2
+    configuration[:separator] = '|'
+    configuration[:eol] = "[EOL]\n"
+    
+    dest = ETL::Control::FileDestination.new(control, configuration, mapping)
+    dest.write(row)
+    dest.write(row_needs_escape)
+    dest.close
+    
+    # Read back the resulting
+    lines = open(File.join(File.dirname(__FILE__), outfile), 'r').readlines
+    assert_equal "123 SW 1st Street,Melbourne,Florida,United States,US\n", lines[0]
+    assert_equal "Allen's Way,Some City,Some State,Mexico,MX\n", lines[1]
+    
+    lines = open(File.join(File.dirname(__FILE__), outfile2), 'r').readlines
+    assert_equal "123 SW 1st Street|Melbourne|Florida|United States|US[EOL]\n", lines[0]
+    assert_equal "Allen's Way|Some City|Some State|Mexico|MX[EOL]\n", lines[1]
+  end
+  
+  # Test a database destination
+  def test_database_destination
+    row = ETL::Row[:id => 1, :first_name => 'Bob', :last_name => 'Smith', :ssn => '111234444']
+    row_needs_escape = ETL::Row[:id => 2, :first_name => "Foo's", :last_name => "Bar", :ssn => '000000000' ]
+    control = ETL::Control::Control.parse(File.dirname(__FILE__) + 
+      '/delimited.ctl')
+    
+    Person.delete_all
+    assert_equal 0, Person.count
+    
+    # First define a basic configuration to check defaults
+    configuration = { 
+      :target => :data_warehouse,
+      :database => 'etl_unittest',
+      :table => 'people',
+      :buffer_size => 0 
+    }
+    mapping = { :order => [:id, :first_name, :last_name, :ssn] }
+    dest = ETL::Control::DatabaseDestination.new(control, configuration, mapping)
+    dest.write(row)
+    dest.close
+    
+    assert_equal 1, Person.find(:all).length
+  end
+  
+  def test_database_destination_with_control
+    row = ETL::Row[:id => 1, :first_name => 'Bob', :last_name => 'Smith', :ssn => '111234444']
+    control = ETL::Control::Control.parse(File.dirname(__FILE__) + 
+      '/delimited_destination_db.ctl')
+    Person.delete_all
+    assert_equal 0, Person.count
+    d = control.destinations.first
+    dest = ETL::Control::DatabaseDestination.new(control, d.configuration, d.mapping)
+    dest.write(row)
+    dest.close
+    assert_equal 1, Person.count
+  end
+  
+  def test_unique
+    row1 = ETL::Row[:id => 1, :first_name => 'Bob', :last_name => 'Smith', :ssn => '111234444']
+    row2 = ETL::Row[:id => 2, :first_name => 'Bob', :last_name => 'Smith', :ssn => '111234444']
+    row3 = ETL::Row[:id => 3, :first_name => 'John', :last_name => 'Smith', :ssn => '000112222']
+         
+    outfile = 'output/test_unique.txt'
+    control = ETL::Control::Control.parse(File.dirname(__FILE__) + '/delimited.ctl')
+
+    # First define a basic configuration to check defaults
+    configuration = { :file => outfile, :buffer_size => 0, :unique => [:ssn]}
+    mapping = { 
+      :order => [:first_name, :last_name, :ssn]
+    }
+    dest = ETL::Control::FileDestination.new(control, configuration, mapping)    
+    dest.write(row1)
+    dest.write(row2)
+    dest.write(row3)
+    
+    # Close (and flush) the destination
+    dest.close
+    
+    # Read back the resulting
+    lines = open(File.join(File.dirname(__FILE__), outfile), 'r').readlines
+    assert_equal "Bob,Smith,111234444\n", lines[0]
+    assert_equal "John,Smith,000112222\n", lines[1]
+  end
+  
+  def test_multiple_unique
+    row1 = ETL::Row[:id => 1, :first_name => 'Bob', :last_name => 'Smith', :ssn => '111234444']
+    row2 = ETL::Row[:id => 2, :first_name => 'Bob', :last_name => 'Smith', :ssn => '111234444']
+    row3 = ETL::Row[:id => 3, :first_name => 'Bob', :last_name => 'Smith', :ssn => '000112222']
+    
+    outfile = 'output/test_multiple_unique.txt'
+    control = ETL::Control::Control.parse(File.dirname(__FILE__) + '/delimited.ctl')
+
+    # First define a basic configuration to check defaults
+    configuration = { :file => outfile, :buffer_size => 0, :unique => [:last_name,:first_name]}
+    mapping = { 
+      :order => [:first_name, :last_name, :ssn]
+    }
+    dest = ETL::Control::FileDestination.new(control, configuration, mapping)    
+    dest.write(row1)
+    dest.write(row2)
+    dest.write(row3)
+    
+    # Close (and flush) the destination
+    dest.close
+    
+    # Read back the resulting
+    lines = open(File.join(File.dirname(__FILE__), outfile), 'r').readlines
+    assert_equal "Bob,Smith,111234444\n", lines[0]
+    assert_equal 1, lines.length
+  end
+
+  def test_bad_destination
+    control = ETL::Control::Control.parse_text('')
+    configuration = {}
+    mapping = {}
+    dest = BadDestination.new(control, configuration, mapping)
+    dest.write(nil)
+    assert_raise NotImplementedError do
+      dest.flush
+    end
+    assert_raise NotImplementedError do
+      dest.close
+    end
+  end
+end