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 +1,97 @@
|
|
1
|
+
module ETL #:nodoc:
|
2
|
+
module Control #:nodoc:
|
3
|
+
# Destination which writes directly to a database. This is useful when you are dealing with
|
4
|
+
# a small amount of data. For larger amounts of data you should probably use the bulk
|
5
|
+
# loader if it is supported with your target database as it will use a much faster load
|
6
|
+
# method.
|
7
|
+
class DatabaseDestination < Destination
|
8
|
+
# The target connection
|
9
|
+
attr_reader :target
|
10
|
+
|
11
|
+
# The table
|
12
|
+
attr_reader :table
|
13
|
+
|
14
|
+
# Specify the order from the source
|
15
|
+
attr_reader :order
|
16
|
+
|
17
|
+
# Set to true to truncate the destination table first
|
18
|
+
attr_reader :truncate
|
19
|
+
|
20
|
+
# Initialize the database destination
|
21
|
+
#
|
22
|
+
# * <tt>control</tt>: The ETL::Control::Control instance
|
23
|
+
# * <tt>configuration</tt>: The configuration Hash
|
24
|
+
# * <tt>mapping</tt>: The mapping
|
25
|
+
#
|
26
|
+
# Configuration options:
|
27
|
+
# * <tt>:database</tt>: The database name (REQUIRED)
|
28
|
+
# * <tt>:target</tt>: The target connection (REQUIRED)
|
29
|
+
# * <tt>:table</tt>: The table to write to (REQUIRED)
|
30
|
+
# * <tt>:truncate</tt>: Set to true to truncate before writing (defaults to false)
|
31
|
+
# * <tt>:unique</tt>: Set to true to only insert unique records (defaults to false)
|
32
|
+
# * <tt>:append_rows</tt>: Array of rows to append
|
33
|
+
#
|
34
|
+
# Mapping options:
|
35
|
+
# * <tt>:order</tt>: The order of fields to write (REQUIRED)
|
36
|
+
def initialize(control, configuration, mapping={})
|
37
|
+
super
|
38
|
+
@target = configuration[:target]
|
39
|
+
@table = configuration[:table]
|
40
|
+
@truncate = configuration[:truncate] ||= false
|
41
|
+
@unique = configuration[:unique] ? configuration[:unique] + [scd_effective_date_field] : configuration[:unique]
|
42
|
+
@unique.uniq! unless @unique.nil?
|
43
|
+
@order = mapping[:order] ? mapping[:order] + scd_required_fields : order_from_source
|
44
|
+
@order.uniq! unless @order.nil?
|
45
|
+
raise ControlError, "Order required in mapping" unless @order
|
46
|
+
raise ControlError, "Table required" unless @table
|
47
|
+
raise ControlError, "Target required" unless @target
|
48
|
+
end
|
49
|
+
|
50
|
+
# Flush the currently buffered data
|
51
|
+
def flush
|
52
|
+
conn.transaction do
|
53
|
+
buffer.flatten.each do |row|
|
54
|
+
# check to see if this row's compound key constraint already exists
|
55
|
+
# note that the compound key constraint may not utilize virtual fields
|
56
|
+
next unless row_allowed?(row)
|
57
|
+
|
58
|
+
# add any virtual fields
|
59
|
+
add_virtuals!(row)
|
60
|
+
|
61
|
+
names = []
|
62
|
+
values = []
|
63
|
+
order.each do |name|
|
64
|
+
names << conn.quote_column_name(name)
|
65
|
+
values << conn.quote(row[name])
|
66
|
+
end
|
67
|
+
q = "INSERT INTO #{conn.quote_table_name(table_name)} (#{names.join(',')}) VALUES (#{values.join(',')})"
|
68
|
+
ETL::Engine.logger.debug("Executing insert: #{q}")
|
69
|
+
conn.insert(q, "Insert row #{current_row}")
|
70
|
+
@current_row += 1
|
71
|
+
end
|
72
|
+
buffer.clear
|
73
|
+
end
|
74
|
+
end
|
75
|
+
|
76
|
+
# Close the connection
|
77
|
+
def close
|
78
|
+
buffer << append_rows if append_rows
|
79
|
+
flush
|
80
|
+
end
|
81
|
+
|
82
|
+
private
|
83
|
+
def conn
|
84
|
+
@conn ||= begin
|
85
|
+
conn = ETL::Engine.connection(target)
|
86
|
+
conn.truncate(table_name) if truncate
|
87
|
+
conn
|
88
|
+
end
|
89
|
+
end
|
90
|
+
|
91
|
+
def table_name
|
92
|
+
ETL::Engine.table(table, ETL::Engine.connection(target))
|
93
|
+
end
|
94
|
+
|
95
|
+
end
|
96
|
+
end
|
97
|
+
end
|
@@ -0,0 +1,91 @@
|
|
1
|
+
optional_require 'spreadsheet'
|
2
|
+
|
3
|
+
module ETL
|
4
|
+
module Control
|
5
|
+
# Excel as the final destination.
|
6
|
+
class ExcelDestination < Destination
|
7
|
+
# The File to write to
|
8
|
+
attr_reader :file
|
9
|
+
|
10
|
+
# The output order
|
11
|
+
attr_reader :order
|
12
|
+
|
13
|
+
# Flag which indicates to append (default is to overwrite)
|
14
|
+
attr_accessor :append
|
15
|
+
|
16
|
+
# Initialize the object.
|
17
|
+
# * <tt>control</tt>: The Control object
|
18
|
+
# * <tt>configuration</tt>: The configuration map
|
19
|
+
# * <tt>mapping</tt>: The output mapping
|
20
|
+
#
|
21
|
+
# Configuration options:
|
22
|
+
# * <tt>:file<tt>: The file to write to (REQUIRED)
|
23
|
+
# * <tt>:append</tt>: Set to true to append to the file (default is to overwrite)
|
24
|
+
# * <tt>:unique</tt>: Set to true to only write unique records
|
25
|
+
# * <tt>:append_rows</tt>: Array of rows to append
|
26
|
+
#
|
27
|
+
# Mapping options:
|
28
|
+
# * <tt>:order</tt>: The order array
|
29
|
+
def initialize(control, configuration, mapping={})
|
30
|
+
super
|
31
|
+
path = Pathname.new(configuration[:file])
|
32
|
+
@file = path.absolute? ? path : Pathname.new(File.dirname(File.expand_path(control.file))) + path
|
33
|
+
@append = configuration[:append] ||= false
|
34
|
+
@unique = configuration[:unique] ? configuration[:unique] + scd_required_fields : configuration[:unique]
|
35
|
+
@unique.uniq! unless @unique.nil?
|
36
|
+
@order = mapping[:order] ? mapping[:order] + scd_required_fields : order_from_source
|
37
|
+
@order.uniq! unless @order.nil?
|
38
|
+
raise ControlError, "Order required in mapping" unless @order
|
39
|
+
end
|
40
|
+
|
41
|
+
# Close the destination. This will flush the buffer and close the underlying stream or connection.
|
42
|
+
def close
|
43
|
+
buffer << append_rows if append_rows
|
44
|
+
flush
|
45
|
+
book.write(file)
|
46
|
+
end
|
47
|
+
|
48
|
+
# Flush the destination buffer
|
49
|
+
def flush
|
50
|
+
#puts "Flushing buffer (#{file}) with #{buffer.length} rows"
|
51
|
+
buffer.flatten.each_with_index do |row, index|
|
52
|
+
#puts "row change type: #{row.change_type}"
|
53
|
+
# check to see if this row's compound key constraint already exists
|
54
|
+
# note that the compound key constraint may not utilize virtual fields
|
55
|
+
next unless row_allowed?(row)
|
56
|
+
|
57
|
+
# add any virtual fields
|
58
|
+
add_virtuals!(row)
|
59
|
+
|
60
|
+
# collect all of the values using the order designated in the configuration
|
61
|
+
values = order.collect do |name|
|
62
|
+
value = row[name]
|
63
|
+
case value
|
64
|
+
when Date, Time, DateTime
|
65
|
+
value.to_s(:db)
|
66
|
+
else
|
67
|
+
value.to_s
|
68
|
+
end
|
69
|
+
end
|
70
|
+
|
71
|
+
# write the values
|
72
|
+
sheet.insert_row(index, values)
|
73
|
+
end
|
74
|
+
buffer.clear
|
75
|
+
#puts "After flush there are #{buffer.length} rows"
|
76
|
+
end
|
77
|
+
|
78
|
+
private
|
79
|
+
# Get the open file excel
|
80
|
+
def book
|
81
|
+
@book ||= ( append ? Spreadsheet.open(file) : Spreadsheet::Workbook.new(file) )
|
82
|
+
end
|
83
|
+
|
84
|
+
private
|
85
|
+
# Get the open sheet
|
86
|
+
def sheet
|
87
|
+
@sheet ||= ( append ? book.worksheet(0) : book.create_worksheet() )
|
88
|
+
end
|
89
|
+
end
|
90
|
+
end
|
91
|
+
end
|
@@ -0,0 +1,126 @@
|
|
1
|
+
# This source file contains the ETL::Control::FileDestination
|
2
|
+
|
3
|
+
module ETL #:nodoc:
|
4
|
+
module Control #:nodoc:
|
5
|
+
# File as the final destination.
|
6
|
+
class FileDestination < Destination
|
7
|
+
# The File to write to
|
8
|
+
attr_reader :file
|
9
|
+
|
10
|
+
# The output order
|
11
|
+
attr_reader :order
|
12
|
+
|
13
|
+
# Flag which indicates to append (default is to overwrite)
|
14
|
+
attr_accessor :append
|
15
|
+
|
16
|
+
# The separator
|
17
|
+
attr_accessor :separator
|
18
|
+
|
19
|
+
# The end of line marker
|
20
|
+
attr_accessor :eol
|
21
|
+
|
22
|
+
# The enclosure character
|
23
|
+
attr_accessor :enclose
|
24
|
+
|
25
|
+
# Initialize the object.
|
26
|
+
# * <tt>control</tt>: The Control object
|
27
|
+
# * <tt>configuration</tt>: The configuration map
|
28
|
+
# * <tt>mapping</tt>: The output mapping
|
29
|
+
#
|
30
|
+
# Configuration options:
|
31
|
+
# * <tt>:file<tt>: The file to write to (REQUIRED)
|
32
|
+
# * <tt>:append</tt>: Set to true to append to the file (default is to overwrite)
|
33
|
+
# * <tt>:separator</tt>: Record separator (default is a comma)
|
34
|
+
# * <tt>:eol</tt>: End of line marker (default is \n)
|
35
|
+
# * <tt>:enclose</tt>: Enclosure character (default is none)
|
36
|
+
# * <tt>:unique</tt>: Set to true to only write unique records
|
37
|
+
# * <tt>:append_rows</tt>: Array of rows to append
|
38
|
+
#
|
39
|
+
# Mapping options:
|
40
|
+
# * <tt>:order</tt>: The order array
|
41
|
+
def initialize(control, configuration, mapping={})
|
42
|
+
super
|
43
|
+
path = Pathname.new(configuration[:file])
|
44
|
+
@file = path.absolute? ? path : Pathname.new(File.dirname(File.expand_path(control.file))) + path
|
45
|
+
@append = configuration[:append] ||= false
|
46
|
+
@separator = configuration[:separator] ||= ','
|
47
|
+
@eol = configuration[:eol] ||= "\n"
|
48
|
+
@enclose = configuration[:enclose]
|
49
|
+
@unique = configuration[:unique] ? configuration[:unique] + scd_required_fields : configuration[:unique]
|
50
|
+
@unique.uniq! unless @unique.nil?
|
51
|
+
@order = mapping[:order] ? mapping[:order] + scd_required_fields : order_from_source
|
52
|
+
@order.uniq! unless @order.nil?
|
53
|
+
raise ControlError, "Order required in mapping" unless @order
|
54
|
+
end
|
55
|
+
|
56
|
+
# Close the destination. This will flush the buffer and close the underlying stream or connection.
|
57
|
+
def close
|
58
|
+
buffer << append_rows if append_rows
|
59
|
+
flush
|
60
|
+
f.close
|
61
|
+
end
|
62
|
+
|
63
|
+
# Flush the destination buffer
|
64
|
+
def flush
|
65
|
+
#puts "Flushing buffer (#{file}) with #{buffer.length} rows"
|
66
|
+
buffer.flatten.each do |row|
|
67
|
+
#puts "row change type: #{row.change_type}"
|
68
|
+
# check to see if this row's compound key constraint already exists
|
69
|
+
# note that the compound key constraint may not utilize virtual fields
|
70
|
+
next unless row_allowed?(row)
|
71
|
+
|
72
|
+
# add any virtual fields
|
73
|
+
add_virtuals!(row)
|
74
|
+
|
75
|
+
# collect all of the values using the order designated in the configuration
|
76
|
+
values = order.collect do |name|
|
77
|
+
value = row[name]
|
78
|
+
case value
|
79
|
+
when Date, Time, DateTime
|
80
|
+
value.to_s(:db)
|
81
|
+
else
|
82
|
+
value.to_s
|
83
|
+
end
|
84
|
+
end
|
85
|
+
|
86
|
+
values.collect! { |v| v.gsub(/\\/, '\\\\\\\\')}
|
87
|
+
values.collect! { |v| v.gsub(separator, "\\#{separator}")}
|
88
|
+
values.collect! { |v| v.gsub(/\n|\r/, '')}
|
89
|
+
|
90
|
+
# enclose the value if required
|
91
|
+
if !enclose.nil?
|
92
|
+
values.collect! { |v| enclose + v.gsub(/(#{enclose})/, '\\\\\1') + enclose }
|
93
|
+
end
|
94
|
+
|
95
|
+
# write the values joined by the separator defined in the configuration
|
96
|
+
f.write(values.join(separator))
|
97
|
+
|
98
|
+
# write the end-of-line
|
99
|
+
f.write(eol)
|
100
|
+
end
|
101
|
+
f.flush
|
102
|
+
buffer.clear
|
103
|
+
#puts "After flush there are #{buffer.length} rows"
|
104
|
+
end
|
105
|
+
|
106
|
+
private
|
107
|
+
# Get the open file stream
|
108
|
+
def f
|
109
|
+
@f ||= open(file, mode)
|
110
|
+
end
|
111
|
+
|
112
|
+
def options
|
113
|
+
@options ||= {
|
114
|
+
:col_sep => separator,
|
115
|
+
:row_sep => eol,
|
116
|
+
:force_quotes => !enclose.nil?
|
117
|
+
}
|
118
|
+
end
|
119
|
+
|
120
|
+
# Get the appropriate mode to open the file stream
|
121
|
+
def mode
|
122
|
+
append ? 'a' : 'w'
|
123
|
+
end
|
124
|
+
end
|
125
|
+
end
|
126
|
+
end
|
@@ -0,0 +1,136 @@
|
|
1
|
+
module ETL #:nodoc:
|
2
|
+
module Control #:nodoc:
|
3
|
+
# Destination which writes directly to a database. This is useful when you are dealing with
|
4
|
+
# a small amount of data. For larger amounts of data you should probably use the bulk
|
5
|
+
# loader if it is supported with your target database as it will use a much faster load
|
6
|
+
# method.
|
7
|
+
class InsertUpdateDatabaseDestination < Destination
|
8
|
+
# The target connection
|
9
|
+
attr_reader :target
|
10
|
+
|
11
|
+
# The table
|
12
|
+
attr_reader :table
|
13
|
+
|
14
|
+
# Specify the order from the source
|
15
|
+
attr_reader :order
|
16
|
+
|
17
|
+
# Specify the primarykey from the source
|
18
|
+
attr_reader :primarykey
|
19
|
+
|
20
|
+
# Set to true to truncate the destination table first
|
21
|
+
attr_reader :truncate
|
22
|
+
|
23
|
+
# Initialize the database destination
|
24
|
+
#
|
25
|
+
# * <tt>control</tt>: The ETL::Control::Control instance
|
26
|
+
# * <tt>configuration</tt>: The configuration Hash
|
27
|
+
# * <tt>mapping</tt>: The mapping
|
28
|
+
#
|
29
|
+
# Configuration options:
|
30
|
+
# * <tt>:database</tt>: The database name (REQUIRED)
|
31
|
+
# * <tt>:target</tt>: The target connection (REQUIRED)
|
32
|
+
# * <tt>:table</tt>: The table to write to (REQUIRED)
|
33
|
+
# * <tt>:truncate</tt>: Set to true to truncate before writing (defaults to false)
|
34
|
+
# * <tt>:unique</tt>: Set to true to only insert unique records (defaults to false)
|
35
|
+
# * <tt>:append_rows</tt>: Array of rows to append
|
36
|
+
#
|
37
|
+
# Mapping options:
|
38
|
+
# * <tt>:order</tt>: The order of fields to write (REQUIRED)
|
39
|
+
# * <tt>:primarykey</tt>: The primary key of fields to select insert or update (REQUIRED)
|
40
|
+
def initialize(control, configuration, mapping={})
|
41
|
+
super
|
42
|
+
@target = configuration[:target]
|
43
|
+
@table = configuration[:table]
|
44
|
+
@truncate = configuration[:truncate] ||= false
|
45
|
+
@unique = configuration[:unique] ? configuration[:unique] + [scd_effective_date_field] : configuration[:unique]
|
46
|
+
@unique.uniq! unless @unique.nil?
|
47
|
+
@order = mapping[:order] ? mapping[:order] + scd_required_fields : order_from_source
|
48
|
+
@order.uniq! unless @order.nil?
|
49
|
+
@primarykey = mapping[:primarykey] ? mapping[:primarykey] + scd_required_fields : nil
|
50
|
+
@primarykey.uniq! unless @primarykey.nil?
|
51
|
+
raise ControlError, "Primarykey required in mapping" unless @primarykey
|
52
|
+
raise ControlError, "Order required in mapping" unless @order
|
53
|
+
raise ControlError, "Table required" unless @table
|
54
|
+
raise ControlError, "Target required" unless @target
|
55
|
+
end
|
56
|
+
|
57
|
+
# Flush the currently buffered data
|
58
|
+
def flush
|
59
|
+
conn.transaction do
|
60
|
+
buffer.flatten.each do |row|
|
61
|
+
# check to see if this row's compound key constraint already exists
|
62
|
+
# note that the compound key constraint may not utilize virtual fields
|
63
|
+
next unless row_allowed?(row)
|
64
|
+
|
65
|
+
# add any virtual fields
|
66
|
+
add_virtuals!(row)
|
67
|
+
|
68
|
+
primarykeyfilter = []
|
69
|
+
primarykey.each do |name|
|
70
|
+
primarykeyfilter << "#{conn.quote_column_name(name)} = #{conn.quote(row[name])}"
|
71
|
+
end
|
72
|
+
q = "SELECT * FROM #{conn.quote_table_name(table_name)} WHERE #{primarykeyfilter.join(' AND ')}"
|
73
|
+
ETL::Engine.logger.debug("Executing select: #{q}")
|
74
|
+
res = conn.execute(q, "Select row #{current_row}")
|
75
|
+
none = true
|
76
|
+
|
77
|
+
case conn
|
78
|
+
when ActiveRecord::ConnectionAdapters::PostgreSQLAdapter
|
79
|
+
res.each { none = false }
|
80
|
+
when ActiveRecord::ConnectionAdapters::MysqlAdapter
|
81
|
+
res.each_hash { none = false }
|
82
|
+
res.free
|
83
|
+
when ActiveRecord::ConnectionAdapters::Mysql2Adapter
|
84
|
+
res.each { none = false }
|
85
|
+
else raise "Unsupported adapter #{conn.class} for this destination"
|
86
|
+
end
|
87
|
+
|
88
|
+
if none
|
89
|
+
names = []
|
90
|
+
values = []
|
91
|
+
order.each do |name|
|
92
|
+
names << conn.quote_column_name(name)
|
93
|
+
values << conn.quote(row[name])
|
94
|
+
end
|
95
|
+
q = "INSERT INTO #{conn.quote_table_name(table_name)} (#{names.join(',')}) VALUES (#{values.join(',')})"
|
96
|
+
ETL::Engine.logger.debug("Executing insert: #{q}")
|
97
|
+
conn.insert(q, "Insert row #{current_row}")
|
98
|
+
else
|
99
|
+
updatevalues = []
|
100
|
+
order.each do |name|
|
101
|
+
updatevalues << "#{conn.quote_column_name(name)} = #{conn.quote(row[name])}"
|
102
|
+
end
|
103
|
+
q = "UPDATE #{conn.quote_table_name(table_name)} SET #{updatevalues.join(',')} WHERE #{primarykeyfilter.join(' AND ')}"
|
104
|
+
ETL::Engine.logger.debug("Executing update: #{q}")
|
105
|
+
conn.update(q, "Update row #{current_row}")
|
106
|
+
end
|
107
|
+
@current_row += 1
|
108
|
+
end
|
109
|
+
buffer.clear
|
110
|
+
end
|
111
|
+
end
|
112
|
+
|
113
|
+
# Close the connection
|
114
|
+
def close
|
115
|
+
buffer << append_rows if append_rows
|
116
|
+
flush
|
117
|
+
end
|
118
|
+
|
119
|
+
private
|
120
|
+
def conn
|
121
|
+
@conn ||= begin
|
122
|
+
conn = ETL::Engine.connection(target)
|
123
|
+
conn.truncate(table_name) if truncate
|
124
|
+
conn
|
125
|
+
rescue
|
126
|
+
raise RuntimeError, "Problem to connect to db"
|
127
|
+
end
|
128
|
+
end
|
129
|
+
|
130
|
+
def table_name
|
131
|
+
ETL::Engine.table(table, ETL::Engine.connection(target))
|
132
|
+
end
|
133
|
+
|
134
|
+
end
|
135
|
+
end
|
136
|
+
end
|