balinterdi-activewarehouse-etl 0.9.2
Sign up to get free protection for your applications and to get access to all the features.
- data/.gitignore +4 -0
- data/0.9-UPGRADE +6 -0
- data/CHANGELOG +198 -0
- data/HOW_TO_RELEASE +4 -0
- data/LICENSE +7 -0
- data/README +85 -0
- data/Rakefile +169 -0
- data/TODO +28 -0
- data/VERSION +1 -0
- data/active_support_logger.patch +78 -0
- data/balinterdi-activewarehouse-etl.gemspec +221 -0
- data/bin/etl +28 -0
- data/bin/etl.cmd +8 -0
- data/examples/database.example.yml +16 -0
- data/lib/etl.rb +78 -0
- data/lib/etl/batch.rb +2 -0
- data/lib/etl/batch/batch.rb +111 -0
- data/lib/etl/batch/directives.rb +55 -0
- data/lib/etl/builder.rb +2 -0
- data/lib/etl/builder/date_dimension_builder.rb +96 -0
- data/lib/etl/builder/time_dimension_builder.rb +31 -0
- data/lib/etl/commands/etl.rb +89 -0
- data/lib/etl/control.rb +3 -0
- data/lib/etl/control/control.rb +405 -0
- data/lib/etl/control/destination.rb +420 -0
- data/lib/etl/control/destination/database_destination.rb +95 -0
- data/lib/etl/control/destination/file_destination.rb +124 -0
- data/lib/etl/control/source.rb +109 -0
- data/lib/etl/control/source/database_source.rb +220 -0
- data/lib/etl/control/source/enumerable_source.rb +11 -0
- data/lib/etl/control/source/file_source.rb +90 -0
- data/lib/etl/control/source/model_source.rb +39 -0
- data/lib/etl/core_ext.rb +1 -0
- data/lib/etl/core_ext/time.rb +5 -0
- data/lib/etl/core_ext/time/calculations.rb +42 -0
- data/lib/etl/engine.rb +556 -0
- data/lib/etl/execution.rb +20 -0
- data/lib/etl/execution/base.rb +9 -0
- data/lib/etl/execution/batch.rb +8 -0
- data/lib/etl/execution/job.rb +8 -0
- data/lib/etl/execution/migration.rb +85 -0
- data/lib/etl/generator.rb +2 -0
- data/lib/etl/generator/generator.rb +20 -0
- data/lib/etl/generator/surrogate_key_generator.rb +39 -0
- data/lib/etl/http_tools.rb +139 -0
- data/lib/etl/parser.rb +11 -0
- data/lib/etl/parser/apache_combined_log_parser.rb +49 -0
- data/lib/etl/parser/delimited_parser.rb +75 -0
- data/lib/etl/parser/fixed_width_parser.rb +65 -0
- data/lib/etl/parser/parser.rb +41 -0
- data/lib/etl/parser/sax_parser.rb +218 -0
- data/lib/etl/parser/xml_parser.rb +65 -0
- data/lib/etl/processor.rb +11 -0
- data/lib/etl/processor/block_processor.rb +14 -0
- data/lib/etl/processor/bulk_import_processor.rb +81 -0
- data/lib/etl/processor/check_exist_processor.rb +80 -0
- data/lib/etl/processor/check_unique_processor.rb +35 -0
- data/lib/etl/processor/copy_field_processor.rb +26 -0
- data/lib/etl/processor/encode_processor.rb +55 -0
- data/lib/etl/processor/hierarchy_exploder_processor.rb +55 -0
- data/lib/etl/processor/print_row_processor.rb +12 -0
- data/lib/etl/processor/processor.rb +25 -0
- data/lib/etl/processor/rename_processor.rb +24 -0
- data/lib/etl/processor/require_non_blank_processor.rb +26 -0
- data/lib/etl/processor/row_processor.rb +17 -0
- data/lib/etl/processor/sequence_processor.rb +23 -0
- data/lib/etl/processor/surrogate_key_processor.rb +53 -0
- data/lib/etl/processor/truncate_processor.rb +35 -0
- data/lib/etl/row.rb +20 -0
- data/lib/etl/screen.rb +14 -0
- data/lib/etl/screen/row_count_screen.rb +20 -0
- data/lib/etl/transform.rb +2 -0
- data/lib/etl/transform/block_transform.rb +13 -0
- data/lib/etl/transform/date_to_string_transform.rb +20 -0
- data/lib/etl/transform/decode_transform.rb +51 -0
- data/lib/etl/transform/default_transform.rb +20 -0
- data/lib/etl/transform/foreign_key_lookup_transform.rb +151 -0
- data/lib/etl/transform/hierarchy_lookup_transform.rb +49 -0
- data/lib/etl/transform/ordinalize_transform.rb +12 -0
- data/lib/etl/transform/sha1_transform.rb +13 -0
- data/lib/etl/transform/string_to_date_transform.rb +16 -0
- data/lib/etl/transform/string_to_datetime_transform.rb +14 -0
- data/lib/etl/transform/string_to_time_transform.rb +11 -0
- data/lib/etl/transform/transform.rb +61 -0
- data/lib/etl/transform/trim_transform.rb +26 -0
- data/lib/etl/transform/type_transform.rb +35 -0
- data/lib/etl/util.rb +59 -0
- data/lib/etl/version.rb +9 -0
- data/test/.ignore +2 -0
- data/test/all.ebf +6 -0
- data/test/apache_combined_log.ctl +11 -0
- data/test/batch_test.rb +41 -0
- data/test/batch_with_error.ebf +6 -0
- data/test/batched1.ctl +0 -0
- data/test/batched2.ctl +0 -0
- data/test/block_processor.ctl +6 -0
- data/test/block_processor_error.ctl +1 -0
- data/test/block_processor_pre_post_process.ctl +4 -0
- data/test/block_processor_remove_rows.ctl +5 -0
- data/test/block_processor_test.rb +38 -0
- data/test/connection/native_mysql/connection.rb +9 -0
- data/test/connection/native_mysql/schema.sql +36 -0
- data/test/connection/postgresql/connection.rb +13 -0
- data/test/connection/postgresql/schema.sql +43 -0
- data/test/control_test.rb +43 -0
- data/test/data/apache_combined_log.txt +3 -0
- data/test/data/bulk_import.txt +3 -0
- data/test/data/bulk_import_with_empties.txt +3 -0
- data/test/data/decode.txt +3 -0
- data/test/data/delimited.txt +3 -0
- data/test/data/encode_source_latin1.txt +2 -0
- data/test/data/fixed_width.txt +3 -0
- data/test/data/multiple_delimited_1.txt +3 -0
- data/test/data/multiple_delimited_2.txt +3 -0
- data/test/data/people.txt +3 -0
- data/test/data/sax.xml +14 -0
- data/test/data/xml.xml +16 -0
- data/test/database.example.yml +18 -0
- data/test/database.mysql.yml +18 -0
- data/test/database.postgres.yml +18 -0
- data/test/database.yml +18 -0
- data/test/date_dimension_builder_test.rb +96 -0
- data/test/delimited.ctl +30 -0
- data/test/delimited_absolute.ctl +33 -0
- data/test/delimited_destination_db.ctl +25 -0
- data/test/delimited_with_bulk_load.ctl +34 -0
- data/test/destination_test.rb +171 -0
- data/test/directive_test.rb +23 -0
- data/test/encode_processor_test.rb +31 -0
- data/test/engine_test.rb +32 -0
- data/test/errors.ctl +24 -0
- data/test/etl_test.rb +42 -0
- data/test/fixed_width.ctl +35 -0
- data/test/generator_test.rb +14 -0
- data/test/inline_parser.ctl +17 -0
- data/test/mocks/mock_destination.rb +26 -0
- data/test/mocks/mock_source.rb +25 -0
- data/test/model_source.ctl +14 -0
- data/test/multiple_delimited.ctl +22 -0
- data/test/multiple_source_delimited.ctl +39 -0
- data/test/parser_test.rb +200 -0
- data/test/performance/delimited.ctl +30 -0
- data/test/processor_test.rb +38 -0
- data/test/row_processor_test.rb +17 -0
- data/test/sax.ctl +26 -0
- data/test/scd/1.txt +1 -0
- data/test/scd/2.txt +1 -0
- data/test/scd/3.txt +1 -0
- data/test/scd_test.rb +263 -0
- data/test/scd_test_type_1.ctl +43 -0
- data/test/scd_test_type_2.ctl +42 -0
- data/test/screen_test.rb +9 -0
- data/test/screen_test_error.ctl +3 -0
- data/test/screen_test_fatal.ctl +3 -0
- data/test/source_test.rb +139 -0
- data/test/test_helper.rb +33 -0
- data/test/transform_test.rb +101 -0
- data/test/xml.ctl +31 -0
- metadata +237 -0
@@ -0,0 +1,30 @@
|
|
1
|
+
# puts "executing delimited.ctl"
|
2
|
+
|
3
|
+
source :in, {
|
4
|
+
:file => 'delimited.txt',
|
5
|
+
:parser => :delimited
|
6
|
+
},
|
7
|
+
[
|
8
|
+
:first_name,
|
9
|
+
:last_name,
|
10
|
+
:ssn,
|
11
|
+
{
|
12
|
+
:name => :age,
|
13
|
+
:type => :integer
|
14
|
+
},
|
15
|
+
:sex
|
16
|
+
]
|
17
|
+
|
18
|
+
transform :ssn, :sha1
|
19
|
+
transform(:ssn){ |v| v[0,24] }
|
20
|
+
transform :sex, :decode, {:decode_table_path => 'delimited_decode.txt'}
|
21
|
+
|
22
|
+
destination :out, {
|
23
|
+
:file => 'delimited.out.txt'
|
24
|
+
},
|
25
|
+
{
|
26
|
+
:order => [:first_name, :last_name, :name, :ssn, :age, :sex],
|
27
|
+
:virtual => {
|
28
|
+
:name => Proc.new { |row| "#{row[:first_name]} #{row[:last_name]}" }
|
29
|
+
}
|
30
|
+
}
|
@@ -0,0 +1,38 @@
|
|
1
|
+
require File.dirname(__FILE__) + '/test_helper'
|
2
|
+
|
3
|
+
class Person < ActiveRecord::Base
|
4
|
+
end
|
5
|
+
|
6
|
+
# Test pre- and post-processors
|
7
|
+
class ProcessorTest < Test::Unit::TestCase
|
8
|
+
# Test bulk import functionality
|
9
|
+
|
10
|
+
context "the bulk import processor" do
|
11
|
+
should "should import successfully" do
|
12
|
+
assert_nothing_raised { do_bulk_import }
|
13
|
+
assert_equal 3, Person.count
|
14
|
+
end
|
15
|
+
end
|
16
|
+
|
17
|
+
def test_bulk_import_with_empties
|
18
|
+
assert_raise(ActiveRecord::StatementInvalid) { do_bulk_import('bulk_import_with_empties.txt') }
|
19
|
+
end
|
20
|
+
|
21
|
+
def test_truncate
|
22
|
+
# TODO: implement test
|
23
|
+
end
|
24
|
+
|
25
|
+
private
|
26
|
+
|
27
|
+
def do_bulk_import(file = 'bulk_import.txt')
|
28
|
+
control = ETL::Control::Control.new(File.join(File.dirname(__FILE__), 'delimited.ctl'))
|
29
|
+
configuration = {
|
30
|
+
:file => "data/#{file}",
|
31
|
+
:truncate => true,
|
32
|
+
:target => :data_warehouse,
|
33
|
+
:table => 'people'
|
34
|
+
}
|
35
|
+
processor = ETL::Processor::BulkImportProcessor.new(control, configuration)
|
36
|
+
processor.process
|
37
|
+
end
|
38
|
+
end
|
@@ -0,0 +1,17 @@
|
|
1
|
+
require File.dirname(__FILE__) + '/test_helper'
|
2
|
+
|
3
|
+
# Test row processors
|
4
|
+
class RowProcessorTest < Test::Unit::TestCase
|
5
|
+
def test_copy_field_processor
|
6
|
+
|
7
|
+
end
|
8
|
+
def test_hierarchy_exploder_processor
|
9
|
+
|
10
|
+
end
|
11
|
+
def test_rename_processor
|
12
|
+
|
13
|
+
end
|
14
|
+
def test_sequence_processor
|
15
|
+
|
16
|
+
end
|
17
|
+
end
|
data/test/sax.ctl
ADDED
@@ -0,0 +1,26 @@
|
|
1
|
+
# puts "executing fixed_width.ctl"
|
2
|
+
|
3
|
+
source :in, {
|
4
|
+
:file => 'data/sax.xml',
|
5
|
+
:parser => :sax
|
6
|
+
},
|
7
|
+
{
|
8
|
+
:write_trigger => 'people/person',
|
9
|
+
:fields => {
|
10
|
+
:first_name => 'people/person/first_name',
|
11
|
+
:last_name => 'people/person/last_name',
|
12
|
+
:ssn => 'people/person/social_security_number',
|
13
|
+
:age => 'people/person[age]'
|
14
|
+
}
|
15
|
+
}
|
16
|
+
|
17
|
+
transform :ssn, :sha1
|
18
|
+
transform(:ssn){ |v| v[0,24] }
|
19
|
+
transform :age, :type, {:type => :number}
|
20
|
+
|
21
|
+
destination :out, {
|
22
|
+
:file => 'output/sax.out.txt'
|
23
|
+
},
|
24
|
+
{
|
25
|
+
:order => [:first_name, :last_name, :ssn, :age]
|
26
|
+
}
|
data/test/scd/1.txt
ADDED
@@ -0,0 +1 @@
|
|
1
|
+
Bob,Smith,200 South Drive,Boston,MA,32123
|
data/test/scd/2.txt
ADDED
@@ -0,0 +1 @@
|
|
1
|
+
Bob,Smith,1010 SW 23rd St,Los Angeles,CA,90392
|
data/test/scd/3.txt
ADDED
@@ -0,0 +1 @@
|
|
1
|
+
Bob,Smith,280 Pine Street,Los Angeles,CA,90392
|
data/test/scd_test.rb
ADDED
@@ -0,0 +1,263 @@
|
|
1
|
+
require File.dirname(__FILE__) + '/test_helper'
|
2
|
+
|
3
|
+
class ScdTest < Test::Unit::TestCase
|
4
|
+
context "when working with a slowly changing dimension" do
|
5
|
+
setup do
|
6
|
+
@connection = ETL::Engine.connection(:data_warehouse)
|
7
|
+
@connection.delete("DELETE FROM person_dimension")
|
8
|
+
@end_of_time = DateTime.parse('9999-12-31 00:00:00')
|
9
|
+
end
|
10
|
+
context "of type 1" do
|
11
|
+
context "on run 1" do
|
12
|
+
setup do
|
13
|
+
do_type_1_run(1)
|
14
|
+
end
|
15
|
+
should "insert record" do
|
16
|
+
assert_equal 1, count_bobs
|
17
|
+
end
|
18
|
+
should "set the original address" do
|
19
|
+
assert_boston_address(find_bobs.first)
|
20
|
+
end
|
21
|
+
should "set the original id" do
|
22
|
+
assert_equal 1, find_bobs.first.id
|
23
|
+
end
|
24
|
+
should "skip the load if there is no change" do
|
25
|
+
do_type_1_run(1)
|
26
|
+
lines = lines_for('scd_test_type_1.txt')
|
27
|
+
assert lines.empty?, "scheduled load expected to be empty, was #{lines.size} records"
|
28
|
+
end
|
29
|
+
end
|
30
|
+
context "on run 2" do
|
31
|
+
setup do
|
32
|
+
do_type_1_run(1)
|
33
|
+
do_type_1_run(2)
|
34
|
+
end
|
35
|
+
should "delete the old record" do
|
36
|
+
assert_equal 1, count_bobs, "new record created, but old not deleted: #{find_bobs.inspect}"
|
37
|
+
end
|
38
|
+
should "update the address" do
|
39
|
+
assert_los_angeles_address(find_bobs.last)
|
40
|
+
end
|
41
|
+
should "keep id" do
|
42
|
+
assert_equal 1, find_bobs.first.id
|
43
|
+
end
|
44
|
+
should "only change once even if run again" do
|
45
|
+
do_type_1_run(2)
|
46
|
+
assert_equal 1, count_bobs
|
47
|
+
lines = lines_for('scd_test_type_1.txt')
|
48
|
+
assert lines.empty?, "scheduled load expected to be empty, was #{lines.size} records"
|
49
|
+
end
|
50
|
+
should "revert address on new record" do
|
51
|
+
do_type_1_run(1)
|
52
|
+
assert_boston_address(find_bobs.first)
|
53
|
+
end
|
54
|
+
should "keep record on revert" do
|
55
|
+
do_type_1_run(1)
|
56
|
+
assert_equal 1, count_bobs
|
57
|
+
end
|
58
|
+
end
|
59
|
+
end
|
60
|
+
context "of type 2" do
|
61
|
+
context "on run 1" do
|
62
|
+
setup do
|
63
|
+
do_type_2_run(1)
|
64
|
+
end
|
65
|
+
should "insert record" do
|
66
|
+
assert_equal 1, count_bobs
|
67
|
+
end
|
68
|
+
should "set the original record" do
|
69
|
+
assert_boston_address(find_bobs.first)
|
70
|
+
end
|
71
|
+
should "set the original id" do
|
72
|
+
assert_equal 1, find_bobs.first.id
|
73
|
+
end
|
74
|
+
should "set the effective date" do
|
75
|
+
# TODO: This is a test bug - if the tests don't run at the correct
|
76
|
+
# time, this timestamp may not match the timestamp used during the
|
77
|
+
# creation of the SCD row
|
78
|
+
assert_equal current_datetime, find_bobs.first.effective_date,
|
79
|
+
"failure might be a test bug - see test notes"
|
80
|
+
end
|
81
|
+
should "set the end date" do
|
82
|
+
assert_equal @end_of_time, find_bobs.first.end_date
|
83
|
+
end
|
84
|
+
should "set the latest version flag" do
|
85
|
+
assert find_bobs.first.latest_version?
|
86
|
+
end
|
87
|
+
should "skip the load if there is no change" do
|
88
|
+
do_type_2_run(1)
|
89
|
+
lines = lines_for('scd_test_type_2.txt')
|
90
|
+
assert lines.empty?, "scheduled load expected to be empty, was #{lines.size} records"
|
91
|
+
end
|
92
|
+
|
93
|
+
end
|
94
|
+
context "on run 2" do
|
95
|
+
setup do
|
96
|
+
do_type_2_run(1)
|
97
|
+
do_type_2_run(2)
|
98
|
+
end
|
99
|
+
should "insert new record" do
|
100
|
+
assert_equal 2, count_bobs
|
101
|
+
end
|
102
|
+
should "keep the primary key of the original version" do
|
103
|
+
assert_not_nil find_bobs.detect { |bob| 1 == bob.id }
|
104
|
+
end
|
105
|
+
should "increment the primary key for the new version" do
|
106
|
+
assert_not_nil find_bobs.detect { |bob| 2 == bob.id }
|
107
|
+
end
|
108
|
+
should "expire the old record" do
|
109
|
+
original_bob = find_bobs.detect { |bob| 1 == bob.id }
|
110
|
+
new_bob = find_bobs.detect { |bob| 2 == bob.id }
|
111
|
+
assert_equal new_bob.effective_date, original_bob.end_date
|
112
|
+
end
|
113
|
+
should "keep the address for the expired record" do
|
114
|
+
assert_boston_address(find_bobs.detect { |bob| 1 == bob.id })
|
115
|
+
end
|
116
|
+
should "update the address on the new record" do
|
117
|
+
assert_los_angeles_address(find_bobs.detect { |bob| 2 == bob.id })
|
118
|
+
end
|
119
|
+
should "activate the new record" do
|
120
|
+
# TODO: This is a test bug - if the tests don't run at the correct
|
121
|
+
# time, this timestamp may not match the timestamp used during the
|
122
|
+
# creation of the SCD row
|
123
|
+
assert_equal current_datetime, find_bobs.detect { |bob| 2 == bob.id }.effective_date,
|
124
|
+
"failure might be a test bug - see test notes"
|
125
|
+
end
|
126
|
+
should "set the end date for the new record" do
|
127
|
+
assert_equal @end_of_time, find_bobs.detect { |bob| 2 == bob.id }.end_date
|
128
|
+
end
|
129
|
+
should "shift the latest version" do
|
130
|
+
original_bob = find_bobs.detect { |bob| 1 == bob.id }
|
131
|
+
new_bob = find_bobs.detect { |bob| 2 == bob.id }
|
132
|
+
assert !original_bob.latest_version?
|
133
|
+
assert new_bob.latest_version?
|
134
|
+
end
|
135
|
+
should "only execute a change once" do
|
136
|
+
do_type_2_run(2)
|
137
|
+
assert_equal 2, count_bobs
|
138
|
+
lines = lines_for('scd_test_type_2.txt')
|
139
|
+
assert lines.empty?, "scheduled load expected to be empty, was #{lines.size} records"
|
140
|
+
end
|
141
|
+
should "insert new records on revert" do
|
142
|
+
do_type_2_run(1)
|
143
|
+
assert_equal 3, count_bobs
|
144
|
+
end
|
145
|
+
should "update address on new record on revert" do
|
146
|
+
do_type_2_run(1)
|
147
|
+
assert_boston_address(find_bobs.detect { |bob| 3 == bob.id })
|
148
|
+
end
|
149
|
+
should "only delete one row on an scd change" do
|
150
|
+
# Two records right now
|
151
|
+
assert_equal 2, count_bobs
|
152
|
+
do_type_2_run(1) # put third version in (same as first version, but that's irrelevant)
|
153
|
+
# was failing because first and second versions were being deleted.
|
154
|
+
assert_equal 3, count_bobs
|
155
|
+
end
|
156
|
+
end
|
157
|
+
context "on non sdc fields that change" do
|
158
|
+
setup do
|
159
|
+
do_type_2_run_with_only_city_state_zip_scd(1)
|
160
|
+
do_type_2_run_with_only_city_state_zip_scd(2)
|
161
|
+
end
|
162
|
+
should "not create an extra record" do
|
163
|
+
do_type_2_run_with_only_city_state_zip_scd(3)
|
164
|
+
assert_equal 2, count_bobs
|
165
|
+
end
|
166
|
+
should "keep id" do
|
167
|
+
do_type_2_run_with_only_city_state_zip_scd(3)
|
168
|
+
assert_not_nil find_bobs.detect { |bob| 2 == bob.id }
|
169
|
+
end
|
170
|
+
should "keep dates" do
|
171
|
+
old_bob = find_bobs.detect { |bob| 2 == bob.id }
|
172
|
+
do_type_2_run_with_only_city_state_zip_scd(3)
|
173
|
+
new_bob = find_bobs.detect { |bob| 2 == bob.id }
|
174
|
+
assert_equal old_bob.end_date, new_bob.end_date
|
175
|
+
assert_equal old_bob.effective_date, new_bob.effective_date
|
176
|
+
end
|
177
|
+
should "keep the latest version flag" do
|
178
|
+
do_type_2_run_with_only_city_state_zip_scd(3)
|
179
|
+
assert find_bobs.detect { |bob| 2 == bob.id }.latest_version?
|
180
|
+
end
|
181
|
+
should "treat non scd fields like type 1 fields" do
|
182
|
+
do_type_2_run_with_only_city_state_zip_scd(3)
|
183
|
+
assert_los_angeles_address(find_bobs.detect { |bob| 2 == bob.id }, "280 Pine Street")
|
184
|
+
end
|
185
|
+
should "skip load when there is no change" do
|
186
|
+
do_type_2_run_with_only_city_state_zip_scd(2)
|
187
|
+
lines = lines_for('scd_test_type_2.txt')
|
188
|
+
assert lines.empty?, "scheduled load expected to be empty, was #{lines.size} records"
|
189
|
+
end
|
190
|
+
end
|
191
|
+
end
|
192
|
+
end
|
193
|
+
|
194
|
+
def do_type_2_run(run_num)
|
195
|
+
ENV['run_number'] = run_num.to_s
|
196
|
+
assert_nothing_raised do
|
197
|
+
run_ctl_file("scd_test_type_2.ctl")
|
198
|
+
end
|
199
|
+
end
|
200
|
+
|
201
|
+
def do_type_2_run_with_only_city_state_zip_scd(run_num)
|
202
|
+
ENV['type_2_scd_fields'] = Marshal.dump([:city, :state, :zip_code])
|
203
|
+
do_type_2_run(run_num)
|
204
|
+
end
|
205
|
+
|
206
|
+
def do_type_1_run(run_num)
|
207
|
+
ENV['run_number'] = run_num.to_s
|
208
|
+
assert_nothing_raised do
|
209
|
+
run_ctl_file("scd_test_type_1.ctl")
|
210
|
+
end
|
211
|
+
end
|
212
|
+
|
213
|
+
def lines_for(file)
|
214
|
+
File.readlines(File.dirname(__FILE__) + "/output/#{file}")
|
215
|
+
end
|
216
|
+
|
217
|
+
def run_ctl_file(file)
|
218
|
+
ETL::Engine.process(File.dirname(__FILE__) + "/#{file}")
|
219
|
+
end
|
220
|
+
|
221
|
+
def count_bobs
|
222
|
+
@connection.select_value(
|
223
|
+
"SELECT count(*) FROM person_dimension WHERE first_name = 'Bob' and last_name = 'Smith'").to_i
|
224
|
+
end
|
225
|
+
|
226
|
+
def find_bobs
|
227
|
+
bobs = @connection.select_all(
|
228
|
+
"SELECT * FROM person_dimension WHERE first_name = 'Bob' and last_name = 'Smith'")
|
229
|
+
bobs.each do |bob|
|
230
|
+
def bob.id
|
231
|
+
self["id"].to_i
|
232
|
+
end
|
233
|
+
def bob.effective_date
|
234
|
+
DateTime.parse(self["effective_date"])
|
235
|
+
end
|
236
|
+
def bob.end_date
|
237
|
+
DateTime.parse(self["end_date"])
|
238
|
+
end
|
239
|
+
def bob.latest_version?
|
240
|
+
ActiveRecord::ConnectionAdapters::Column.value_to_boolean(self["latest_version"])
|
241
|
+
end
|
242
|
+
end
|
243
|
+
bobs
|
244
|
+
end
|
245
|
+
|
246
|
+
def current_datetime
|
247
|
+
DateTime.parse(Time.now.to_s(:db))
|
248
|
+
end
|
249
|
+
|
250
|
+
def assert_boston_address(bob, street = "200 South Drive")
|
251
|
+
assert_equal street, bob['address'], bob.inspect
|
252
|
+
assert_equal "Boston", bob['city'], bob.inspect
|
253
|
+
assert_equal "MA", bob['state'], bob.inspect
|
254
|
+
assert_equal "32123", bob['zip_code'], bob.inspect
|
255
|
+
end
|
256
|
+
|
257
|
+
def assert_los_angeles_address(bob, street = "1010 SW 23rd St")
|
258
|
+
assert_equal street, bob['address'], bob.inspect
|
259
|
+
assert_equal "Los Angeles", bob['city'], bob.inspect
|
260
|
+
assert_equal "CA", bob['state'], bob.inspect
|
261
|
+
assert_equal "90392", bob['zip_code'], bob.inspect
|
262
|
+
end
|
263
|
+
end
|
@@ -0,0 +1,43 @@
|
|
1
|
+
source :in, {
|
2
|
+
:file => "scd/#{ENV['run_number']}.txt",
|
3
|
+
:parser => :delimited
|
4
|
+
},
|
5
|
+
[
|
6
|
+
:first_name,
|
7
|
+
:last_name,
|
8
|
+
:address,
|
9
|
+
:city,
|
10
|
+
:state,
|
11
|
+
:zip_code
|
12
|
+
]
|
13
|
+
|
14
|
+
# NOTE: These are not usually required for a type 1 SCD dimension, but since
|
15
|
+
# we're sharing this table with the type 2 tests, they're necessary.
|
16
|
+
transform :effective_date, :default, :default_value => Time.now.to_s(:db)
|
17
|
+
transform :end_date, :default, :default_value => '9999-12-31 00:00:00'
|
18
|
+
transform :latest_version, :default, :default_value => true
|
19
|
+
|
20
|
+
destination :out, {
|
21
|
+
:file => 'output/scd_test_type_1.txt',
|
22
|
+
:natural_key => [:first_name, :last_name],
|
23
|
+
:scd => {
|
24
|
+
:type => 1,
|
25
|
+
:dimension_target => :data_warehouse,
|
26
|
+
:dimension_table => 'person_dimension'
|
27
|
+
},
|
28
|
+
:scd_fields => [:address, :city, :state, :zip_code]
|
29
|
+
},
|
30
|
+
{
|
31
|
+
:order => [
|
32
|
+
:id, :first_name, :last_name, :address, :city, :state, :zip_code, :effective_date, :end_date, :latest_version
|
33
|
+
],
|
34
|
+
:virtual => {
|
35
|
+
:id => ETL::Generator::SurrogateKeyGenerator.new(:target => :data_warehouse, :table => 'person_dimension')
|
36
|
+
}
|
37
|
+
}
|
38
|
+
|
39
|
+
post_process :bulk_import, {
|
40
|
+
:file => 'output/scd_test_type_1.txt',
|
41
|
+
:target => :data_warehouse,
|
42
|
+
:table => 'person_dimension'
|
43
|
+
}
|
@@ -0,0 +1,42 @@
|
|
1
|
+
source :in, {
|
2
|
+
:file => "scd/#{ENV['run_number']}.txt",
|
3
|
+
:parser => :delimited
|
4
|
+
},
|
5
|
+
[
|
6
|
+
:first_name,
|
7
|
+
:last_name,
|
8
|
+
:address,
|
9
|
+
:city,
|
10
|
+
:state,
|
11
|
+
:zip_code
|
12
|
+
]
|
13
|
+
|
14
|
+
|
15
|
+
|
16
|
+
|
17
|
+
|
18
|
+
|
19
|
+
destination :out, {
|
20
|
+
:file => 'output/scd_test_type_2.txt',
|
21
|
+
:natural_key => [:first_name, :last_name],
|
22
|
+
:scd => {
|
23
|
+
:type => 2,
|
24
|
+
:dimension_target => :data_warehouse,
|
25
|
+
:dimension_table => 'person_dimension'
|
26
|
+
},
|
27
|
+
:scd_fields => ENV['type_2_scd_fields'] ? Marshal.load(ENV['type_2_scd_fields']) : [:address, :city, :state, :zip_code]
|
28
|
+
},
|
29
|
+
{
|
30
|
+
:order => [
|
31
|
+
:id, :first_name, :last_name, :address, :city, :state, :zip_code, :effective_date, :end_date, :latest_version
|
32
|
+
],
|
33
|
+
:virtual => {
|
34
|
+
:id => ETL::Generator::SurrogateKeyGenerator.new(:target => :data_warehouse, :table => 'person_dimension')
|
35
|
+
}
|
36
|
+
}
|
37
|
+
|
38
|
+
post_process :bulk_import, {
|
39
|
+
:file => 'output/scd_test_type_2.txt',
|
40
|
+
:target => :data_warehouse,
|
41
|
+
:table => 'person_dimension'
|
42
|
+
}
|