activewarehouse-etl-sgonyea 0.9.6
Sign up to get free protection for your applications and to get access to all the features.
- data/.gitignore +9 -0
- data/0.9-UPGRADE +6 -0
- data/CHANGELOG +236 -0
- data/Gemfile +4 -0
- data/HOW_TO_RELEASE +13 -0
- data/LICENSE +7 -0
- data/README.textile +111 -0
- data/Rakefile +103 -0
- data/TODO +28 -0
- data/active_support_logger.patch +78 -0
- data/activewarehouse-etl.gemspec +36 -0
- data/bin/etl +28 -0
- data/bin/etl.cmd +8 -0
- data/examples/database.example.yml +16 -0
- data/lib/etl.rb +97 -0
- data/lib/etl/batch.rb +2 -0
- data/lib/etl/batch/batch.rb +111 -0
- data/lib/etl/batch/directives.rb +65 -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 +438 -0
- data/lib/etl/control/destination/csv_destination.rb +113 -0
- data/lib/etl/control/destination/database_destination.rb +97 -0
- data/lib/etl/control/destination/excel_destination.rb +91 -0
- data/lib/etl/control/destination/file_destination.rb +126 -0
- data/lib/etl/control/destination/insert_update_database_destination.rb +136 -0
- data/lib/etl/control/destination/update_database_destination.rb +109 -0
- data/lib/etl/control/destination/yaml_destination.rb +74 -0
- data/lib/etl/control/source.rb +132 -0
- data/lib/etl/control/source/database_source.rb +224 -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 +582 -0
- data/lib/etl/execution.rb +19 -0
- data/lib/etl/execution/base.rb +8 -0
- data/lib/etl/execution/batch.rb +10 -0
- data/lib/etl/execution/job.rb +8 -0
- data/lib/etl/execution/migration.rb +90 -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/csv_parser.rb +93 -0
- data/lib/etl/parser/excel_parser.rb +112 -0
- data/lib/etl/parser/fixed_width_parser.rb +65 -0
- data/lib/etl/parser/nokogiri_xml_parser.rb +83 -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 +94 -0
- data/lib/etl/processor/check_exist_processor.rb +80 -0
- data/lib/etl/processor/check_unique_processor.rb +39 -0
- data/lib/etl/processor/copy_field_processor.rb +26 -0
- data/lib/etl/processor/database_join_processor.rb +82 -0
- data/lib/etl/processor/encode_processor.rb +55 -0
- data/lib/etl/processor/ensure_fields_presence_processor.rb +24 -0
- data/lib/etl/processor/escape_csv_processor.rb +77 -0
- data/lib/etl/processor/filter_row_processor.rb +51 -0
- data/lib/etl/processor/ftp_downloader_processor.rb +68 -0
- data/lib/etl/processor/ftp_uploader_processor.rb +65 -0
- data/lib/etl/processor/hierarchy_exploder_processor.rb +55 -0
- data/lib/etl/processor/imapattachment_downloader_processor.rb +91 -0
- data/lib/etl/processor/pop3attachment_downloader_processor.rb +90 -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 +27 -0
- data/lib/etl/processor/sequence_processor.rb +23 -0
- data/lib/etl/processor/sftp_downloader_processor.rb +63 -0
- data/lib/etl/processor/sftp_uploader_processor.rb +63 -0
- data/lib/etl/processor/surrogate_key_processor.rb +53 -0
- data/lib/etl/processor/truncate_processor.rb +40 -0
- data/lib/etl/processor/zip_file_processor.rb +27 -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/calculation_transform.rb +71 -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 +211 -0
- data/lib/etl/transform/hierarchy_lookup_transform.rb +49 -0
- data/lib/etl/transform/md5_transform.rb +13 -0
- data/lib/etl/transform/ordinalize_transform.rb +14 -0
- data/lib/etl/transform/sha1_transform.rb +13 -0
- data/lib/etl/transform/split_fields_transform.rb +27 -0
- data/lib/etl/transform/string_to_date_time_transform.rb +14 -0
- data/lib/etl/transform/string_to_date_transform.rb +16 -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 +3 -0
- data/test-matrix.yml +10 -0
- data/test/.gitignore +1 -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/check_exist_processor_test.rb +92 -0
- data/test/check_unique_processor_test.rb +40 -0
- data/test/config/Gemfile.rails-2.3.x +3 -0
- data/test/config/Gemfile.rails-2.3.x.lock +53 -0
- data/test/config/Gemfile.rails-3.0.x +3 -0
- data/test/config/Gemfile.rails-3.0.x.lock +61 -0
- data/test/config/common.rb +29 -0
- data/test/connection/mysql/connection.rb +9 -0
- data/test/connection/mysql/schema.sql +37 -0
- data/test/connection/postgresql/connection.rb +13 -0
- data/test/connection/postgresql/schema.sql +40 -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/excel.xls +0 -0
- data/test/data/excel2.xls +0 -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/nokogiri.xml +38 -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_join_processor_test.rb +43 -0
- data/test/date_dimension_builder_test.rb +96 -0
- data/test/delimited.ctl +30 -0
- data/test/delimited_absolute.ctl +31 -0
- data/test/delimited_destination_db.ctl +23 -0
- data/test/delimited_excel.ctl +31 -0
- data/test/delimited_insert_update.ctl +34 -0
- data/test/delimited_update.ctl +34 -0
- data/test/delimited_with_bulk_load.ctl +34 -0
- data/test/destination_test.rb +275 -0
- data/test/directive_test.rb +23 -0
- data/test/encode_processor_test.rb +32 -0
- data/test/engine_test.rb +78 -0
- data/test/ensure_fields_presence_processor_test.rb +28 -0
- data/test/errors.ctl +24 -0
- data/test/etl_test.rb +42 -0
- data/test/excel.ctl +24 -0
- data/test/excel2.ctl +25 -0
- data/test/fixed_width.ctl +35 -0
- data/test/foreign_key_lookup_transform_test.rb +50 -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/nokogiri_all.ctl +35 -0
- data/test/nokogiri_select.ctl +35 -0
- data/test/nokogiri_test.rb +35 -0
- data/test/parser_test.rb +224 -0
- data/test/performance/delimited.ctl +30 -0
- data/test/processor_test.rb +44 -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 +257 -0
- data/test/scd_test_type_1.ctl +43 -0
- data/test/scd_test_type_2.ctl +34 -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 +154 -0
- data/test/test_helper.rb +37 -0
- data/test/transform_test.rb +101 -0
- data/test/truncate_processor_test.rb +37 -0
- data/test/xml.ctl +31 -0
- metadata +370 -0
@@ -0,0 +1,14 @@
|
|
1
|
+
require 'active_support/core_ext/integer/inflections.rb'
|
2
|
+
|
3
|
+
module ETL #:nodoc:
|
4
|
+
module Transform #:nodoc:
|
5
|
+
# Transform a number to an ordinalized version using the ActiveSupport ordinalize
|
6
|
+
# core extension
|
7
|
+
class OrdinalizeTransform < ETL::Transform::Transform
|
8
|
+
# Transform the value from a number to an ordinalized number
|
9
|
+
def transform(name, value, row)
|
10
|
+
value.ordinalize
|
11
|
+
end
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
@@ -0,0 +1,13 @@
|
|
1
|
+
require 'digest/sha1'
|
2
|
+
|
3
|
+
module ETL #:nodoc:
|
4
|
+
module Transform #:nodoc:
|
5
|
+
# Transform which hashes the original value with a SHA-1 hash algorithm
|
6
|
+
class Sha1Transform < ETL::Transform::Transform
|
7
|
+
# Transform the value with a SHA1 digest algorithm.
|
8
|
+
def transform(name, value, row)
|
9
|
+
Digest::SHA1.hexdigest(value)
|
10
|
+
end
|
11
|
+
end
|
12
|
+
end
|
13
|
+
end
|
@@ -0,0 +1,27 @@
|
|
1
|
+
module ETL
|
2
|
+
module Transform
|
3
|
+
class SplitFieldsTransform < ETL::Transform::Transform
|
4
|
+
attr_reader :delimiter
|
5
|
+
attr_reader :new_fields
|
6
|
+
|
7
|
+
def initialize(control, name, configuration)
|
8
|
+
@delimiter = configuration[:delimiter] || ','
|
9
|
+
@new_fields = configuration[:new_fields]
|
10
|
+
super
|
11
|
+
end
|
12
|
+
|
13
|
+
def transform(name, value, row)
|
14
|
+
return nil if row.nil?
|
15
|
+
return nil if row[name].nil?
|
16
|
+
|
17
|
+
fields = row[name].split(@delimiter)
|
18
|
+
@new_fields.each_with_index do |new, index|
|
19
|
+
row[new] = fields[index]
|
20
|
+
end
|
21
|
+
|
22
|
+
row[name]
|
23
|
+
end
|
24
|
+
|
25
|
+
end
|
26
|
+
end
|
27
|
+
end
|
@@ -0,0 +1,14 @@
|
|
1
|
+
module ETL #:nodoc:
|
2
|
+
module Transform #:nodoc:
|
3
|
+
# Transform a String representation of a date to a DateTime instance
|
4
|
+
class StringToDateTimeTransform < ETL::Transform::Transform
|
5
|
+
# Transform the value using DateTime.parse.
|
6
|
+
#
|
7
|
+
# WARNING: This transform is slow (due to the Ruby implementation), but if you need to
|
8
|
+
# parse timestamps before or after the values supported by the Time.parse.
|
9
|
+
def transform(name, value, row)
|
10
|
+
DateTime.parse(value) unless value.nil?
|
11
|
+
end
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
module ETL #:nodoc:
|
2
|
+
module Transform #:nodoc:
|
3
|
+
# Transform a String representation of a date to a Date instance
|
4
|
+
class StringToDateTransform < ETL::Transform::Transform
|
5
|
+
# Transform the value using Date.parse
|
6
|
+
def transform(name, value, row)
|
7
|
+
return value if value.nil?
|
8
|
+
begin
|
9
|
+
Date.parse(value)
|
10
|
+
rescue => e
|
11
|
+
return value
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,11 @@
|
|
1
|
+
module ETL #:nodoc:
|
2
|
+
module Transform #:nodoc:
|
3
|
+
# Transform a String representation of a date to a Time instance
|
4
|
+
class StringToTimeTransform < ETL::Transform::Transform
|
5
|
+
# Transform the value using Time.parse
|
6
|
+
def transform(name, value, row)
|
7
|
+
Time.parse(value) unless value.nil?
|
8
|
+
end
|
9
|
+
end
|
10
|
+
end
|
11
|
+
end
|
@@ -0,0 +1,61 @@
|
|
1
|
+
module ETL#:nodoc:
|
2
|
+
module Transform#:nodoc:
|
3
|
+
# Base class for transforms.
|
4
|
+
#
|
5
|
+
# A transform converts one value to another value using some sort of algorithm.
|
6
|
+
#
|
7
|
+
# A simple transform has two arguments, the field to transform and the name of the transform:
|
8
|
+
#
|
9
|
+
# transform :ssn, :sha1
|
10
|
+
#
|
11
|
+
# Transforms can also be blocks:
|
12
|
+
#
|
13
|
+
# transform(:ssn){ |v| v[0,24] }
|
14
|
+
#
|
15
|
+
# Finally, a transform can include a configuration hash:
|
16
|
+
#
|
17
|
+
# transform :sex, :decode, {:decode_table_path => 'delimited_decode.txt'}
|
18
|
+
class Transform
|
19
|
+
class << self
|
20
|
+
# Transform the specified value using the given transforms. The transforms can either be
|
21
|
+
# Proc objects or objects which extend from Transform and implement the method <tt>transform(value)</tt>.
|
22
|
+
# Any other object will result in a ControlError being raised.
|
23
|
+
def transform(name, value, row, transforms)
|
24
|
+
transforms.each do |transform|
|
25
|
+
benchmarks[transform.class] ||= 0
|
26
|
+
benchmarks[transform.class] += Benchmark.realtime do
|
27
|
+
Engine.logger.debug "Transforming field #{name} with #{transform.inspect}"
|
28
|
+
case transform
|
29
|
+
when Proc
|
30
|
+
value = transform.call([name, value, row])
|
31
|
+
when Transform
|
32
|
+
value = transform.transform(name, value, row)
|
33
|
+
else
|
34
|
+
raise ControlError, "Unsupported transform configuration type: #{transform}"
|
35
|
+
end
|
36
|
+
end
|
37
|
+
end
|
38
|
+
value
|
39
|
+
end
|
40
|
+
|
41
|
+
def benchmarks
|
42
|
+
@benchmarks ||= {}
|
43
|
+
end
|
44
|
+
end
|
45
|
+
|
46
|
+
attr_reader :control, :name, :configuration
|
47
|
+
|
48
|
+
# Initialize the transform object with the given control object, field name and
|
49
|
+
# configuration hash
|
50
|
+
def initialize(control, name, configuration={})
|
51
|
+
@control = control
|
52
|
+
@name = name
|
53
|
+
@configuration = configuration
|
54
|
+
end
|
55
|
+
|
56
|
+
def transform(name, value, row)
|
57
|
+
raise "transform is an abstract method"
|
58
|
+
end
|
59
|
+
end
|
60
|
+
end
|
61
|
+
end
|
@@ -0,0 +1,26 @@
|
|
1
|
+
module ETL #:nodoc:
|
2
|
+
module Transform #:nodoc:
|
3
|
+
# Transform to trim string
|
4
|
+
class TrimTransform < ETL::Transform::Transform
|
5
|
+
# Configuration options:
|
6
|
+
# * <tt>:type</tt>: :left, :right or :both. Default is :both
|
7
|
+
def initialize(control, name, configuration={})
|
8
|
+
super
|
9
|
+
@type = (configuration[:type] || :both).to_sym
|
10
|
+
end
|
11
|
+
# Transform the value
|
12
|
+
def transform(name, value, row)
|
13
|
+
case @type
|
14
|
+
when :left
|
15
|
+
value.lstrip
|
16
|
+
when :right
|
17
|
+
value.rstrip
|
18
|
+
when :both
|
19
|
+
value.strip
|
20
|
+
else
|
21
|
+
raise "Trim type, if specified, must be :left, :right or :both"
|
22
|
+
end
|
23
|
+
end
|
24
|
+
end
|
25
|
+
end
|
26
|
+
end
|
@@ -0,0 +1,35 @@
|
|
1
|
+
module ETL #:nodoc:
|
2
|
+
module Transform #:nodoc:
|
3
|
+
# Transform from one type to another
|
4
|
+
class TypeTransform < ETL::Transform::Transform
|
5
|
+
# Initialize the transformer.
|
6
|
+
#
|
7
|
+
# Configuration options:
|
8
|
+
# * <tt>:type</tt>: The type to convert to. Supported types:
|
9
|
+
# ** :string
|
10
|
+
# ** :number,:integer
|
11
|
+
# ** :float
|
12
|
+
# ** :decimal
|
13
|
+
def initialize(control, name, configuration={})
|
14
|
+
super
|
15
|
+
@type = configuration[:type]
|
16
|
+
@significant = configuration[:significant] ||= 0
|
17
|
+
end
|
18
|
+
# Transform the value
|
19
|
+
def transform(name, value, row)
|
20
|
+
case @type
|
21
|
+
when :string
|
22
|
+
value.to_s
|
23
|
+
when :number, :integer
|
24
|
+
value.to_i
|
25
|
+
when :float
|
26
|
+
value.to_f
|
27
|
+
when :decimal
|
28
|
+
BigDecimal.new(value.to_s, @significant)
|
29
|
+
else
|
30
|
+
raise "Unsupported type: #{@type}"
|
31
|
+
end
|
32
|
+
end
|
33
|
+
end
|
34
|
+
end
|
35
|
+
end
|
data/lib/etl/util.rb
ADDED
@@ -0,0 +1,59 @@
|
|
1
|
+
module ETL
|
2
|
+
module Util
|
3
|
+
# Return the distance of time in words from the given from_time to the specified to_time. If to_time
|
4
|
+
# is not specified then Time.now is used. By default seconds are included...set the include_seconds
|
5
|
+
# argument to false to disable the seconds.
|
6
|
+
def distance_of_time_in_words(from_time, to_time=Time.now)
|
7
|
+
from_time = from_time.to_time if from_time.respond_to?(:to_time)
|
8
|
+
to_time = to_time.to_time if to_time.respond_to?(:to_time)
|
9
|
+
seconds = (to_time - from_time).round
|
10
|
+
distance_in_days = (seconds/(60*60*24)).round
|
11
|
+
seconds = seconds % (60*60*24)
|
12
|
+
distance_in_hours = (seconds/(60*60)).round
|
13
|
+
seconds = seconds % (60*60)
|
14
|
+
distance_in_minutes = (seconds/60).round
|
15
|
+
seconds = seconds % 60
|
16
|
+
distance_in_seconds = seconds
|
17
|
+
|
18
|
+
s = ''
|
19
|
+
s << "#{distance_in_days} days," if distance_in_days > 0
|
20
|
+
s << "#{distance_in_hours} hours, " if distance_in_hours > 0
|
21
|
+
s << "#{distance_in_minutes} minutes, " if distance_in_minutes > 0
|
22
|
+
s << "#{distance_in_seconds} seconds"
|
23
|
+
s
|
24
|
+
end
|
25
|
+
|
26
|
+
# Get the approximate disntance of time in words from the given from_time
|
27
|
+
# to the the given to_time. If to_time is not specified then it is set
|
28
|
+
# to Time.now. By default seconds are included...set the include_seconds
|
29
|
+
# argument to false to disable the seconds.
|
30
|
+
def approximate_distance_of_time_in_words(from_time, to_time=Time.now, include_seconds=true)
|
31
|
+
from_time = from_time.to_time if from_time.respond_to?(:to_time)
|
32
|
+
to_time = to_time.to_time if to_time.respond_to?(:to_time)
|
33
|
+
distance_in_minutes = (((to_time - from_time).abs)/60).round
|
34
|
+
distance_in_seconds = ((to_time - from_time).abs).round
|
35
|
+
|
36
|
+
case distance_in_minutes
|
37
|
+
when 0..1
|
38
|
+
return (distance_in_minutes == 0) ? 'less than a minute' : '1 minute' unless include_seconds
|
39
|
+
case distance_in_seconds
|
40
|
+
when 0..4 then 'less than 5 seconds'
|
41
|
+
when 5..9 then 'less than 10 seconds'
|
42
|
+
when 10..19 then 'less than 20 seconds'
|
43
|
+
when 20..39 then 'half a minute'
|
44
|
+
when 40..59 then 'less than a minute'
|
45
|
+
else '1 minute'
|
46
|
+
end
|
47
|
+
when 2..44 then "#{distance_in_minutes} minutes"
|
48
|
+
when 45..89 then 'about 1 hour'
|
49
|
+
when 90..1439 then "about #{(distance_in_minutes.to_f / 60.0).round} hours"
|
50
|
+
when 1440..2879 then '1 day'
|
51
|
+
when 2880..43199 then "#{(distance_in_minutes / 1440).round} days"
|
52
|
+
when 43200..86399 then 'about 1 month'
|
53
|
+
when 86400..525959 then "#{(distance_in_minutes / 43200).round} months"
|
54
|
+
when 525960..1051919 then 'about 1 year'
|
55
|
+
else "over #{(distance_in_minutes / 525960).round} years"
|
56
|
+
end
|
57
|
+
end
|
58
|
+
end
|
59
|
+
end
|
data/lib/etl/version.rb
ADDED
data/test-matrix.yml
ADDED
data/test/.gitignore
ADDED
@@ -0,0 +1 @@
|
|
1
|
+
database*.yml
|
data/test/.ignore
ADDED
data/test/all.ebf
ADDED
data/test/batch_test.rb
ADDED
@@ -0,0 +1,41 @@
|
|
1
|
+
require File.dirname(__FILE__) + '/test_helper'
|
2
|
+
|
3
|
+
class BatchTest < Test::Unit::TestCase
|
4
|
+
attr_reader :file, :db_yaml, :engine
|
5
|
+
def setup
|
6
|
+
@file = File.dirname(__FILE__) + '/all.ebf'
|
7
|
+
@db_yaml = File.dirname(__FILE__) + '/database.yml'
|
8
|
+
@engine = ETL::Engine.new
|
9
|
+
end
|
10
|
+
def teardown
|
11
|
+
|
12
|
+
end
|
13
|
+
def test_etl_batch_file
|
14
|
+
#`etl #{file} -c #{db_yaml}`
|
15
|
+
end
|
16
|
+
def test_batch
|
17
|
+
assert_nothing_raised do
|
18
|
+
batch = ETL::Batch::Batch.resolve(file, engine)
|
19
|
+
batch.execute
|
20
|
+
end
|
21
|
+
end
|
22
|
+
def test_batch_with_file
|
23
|
+
assert_nothing_raised do
|
24
|
+
batch = ETL::Batch::Batch.resolve(File.new(file), engine)
|
25
|
+
batch.execute
|
26
|
+
end
|
27
|
+
end
|
28
|
+
def test_batch_with_batch_object
|
29
|
+
assert_nothing_raised do
|
30
|
+
batch_instance = ETL::Batch::Batch.new(File.new(file))
|
31
|
+
batch_instance.engine = engine
|
32
|
+
batch = ETL::Batch::Batch.resolve(batch_instance, engine)
|
33
|
+
batch.execute
|
34
|
+
end
|
35
|
+
end
|
36
|
+
def test_batch_with_object_should_fail
|
37
|
+
assert_raise(RuntimeError) do
|
38
|
+
batch = ETL::Batch::Batch.resolve(0, engine)
|
39
|
+
end
|
40
|
+
end
|
41
|
+
end
|
data/test/batched1.ctl
ADDED
File without changes
|
data/test/batched2.ctl
ADDED
File without changes
|
@@ -0,0 +1,6 @@
|
|
1
|
+
source :in, { :type => :mock, :name => :block_processed_input }
|
2
|
+
|
3
|
+
after_read { |row| row[:added_by_after_read] = "after-" +row[:first_name]; row }
|
4
|
+
before_write { |row| row[:added_by_before_write] = "Row #{Engine.current_source_row}"; [row,{:new_row => 'added by post_processor'}] }
|
5
|
+
|
6
|
+
destination :out, { :type => :mock, :name => :block_processed_output }
|
@@ -0,0 +1 @@
|
|
1
|
+
pre_process { raise ControlError.new( "Cough!") }
|
@@ -0,0 +1,38 @@
|
|
1
|
+
require File.dirname(__FILE__) + '/test_helper'
|
2
|
+
include ETL
|
3
|
+
include ETL::Control
|
4
|
+
|
5
|
+
class TestWitness
|
6
|
+
end
|
7
|
+
|
8
|
+
class BlockProcessorTest < Test::Unit::TestCase
|
9
|
+
|
10
|
+
def test_block_processor_should_work_as_both_after_read_and_before_write_row_processor
|
11
|
+
MockSource[:block_processed_input] = [{ :first_name => 'John'},{:first_name => 'Gary'}]
|
12
|
+
process 'block_processor.ctl'
|
13
|
+
assert_equal 4, MockDestination[:block_processed_output].size
|
14
|
+
assert_equal({ :first_name => 'John', :added_by_after_read => 'after-John', :added_by_before_write => "Row 1" }, MockDestination[:block_processed_output][0])
|
15
|
+
assert_equal({ :new_row => 'added by post_processor' }, MockDestination[:block_processed_output][1])
|
16
|
+
assert_equal({ :first_name => 'Gary', :added_by_after_read => 'after-Gary', :added_by_before_write => "Row 2" }, MockDestination[:block_processed_output][2])
|
17
|
+
assert_equal({ :new_row => 'added by post_processor' }, MockDestination[:block_processed_output][3])
|
18
|
+
end
|
19
|
+
|
20
|
+
def test_block_processor_should_let_rows_be_removed_by_setting_it_to_nil
|
21
|
+
MockSource[:block_input] = [{ :obsolete => true, :name => 'John'},{ :obsolete => false, :name => 'Gary'}]
|
22
|
+
process 'block_processor_remove_rows.ctl'
|
23
|
+
assert_equal([{ :obsolete => false, :name => 'Gary' }], MockDestination[:block_output]) # only one record should be kept
|
24
|
+
end
|
25
|
+
|
26
|
+
def test_block_processor_should_work_as_pre_or_post_processor
|
27
|
+
flexmock(TestWitness).should_receive(:call).with("I'm called from pre_process")
|
28
|
+
flexmock(TestWitness).should_receive(:call).with("I'm called from post_process")
|
29
|
+
MockSource[:another_input] = [{ :obsolete => true, :name => 'John'},{ :obsolete => false, :name => 'Gary'}]
|
30
|
+
process 'block_processor_pre_post_process.ctl'
|
31
|
+
assert_equal(MockSource[:another_input], MockDestination[:another_output])
|
32
|
+
end
|
33
|
+
|
34
|
+
def test_block_error_should_be_propagated
|
35
|
+
assert_raise(ControlError) { process 'block_processor_error.ctl' }
|
36
|
+
end
|
37
|
+
|
38
|
+
end
|