beetle_etl 0.0.13 → 0.0.16
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/lib/beetle_etl/steps/assign_ids.rb +9 -18
- data/lib/beetle_etl/steps/load.rb +3 -3
- data/lib/beetle_etl/steps/table_diff.rb +4 -25
- data/lib/beetle_etl/version.rb +1 -1
- data/spec/feature/feature_spec.rb +2 -2
- data/spec/steps/assign_ids_spec.rb +31 -60
- data/spec/steps/load_spec.rb +4 -4
- data/spec/steps/table_diff_spec.rb +7 -33
- metadata +2 -2
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA1:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 07d1e92398037e916f19762ae9a373de8cd3edfa
|
4
|
+
data.tar.gz: c064dfbfd3dfd4103864c51d8742f537963abf8c
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: b984f80cdba06d018d49fa9414cf5443b8098b0edf34f1eb149df32f5c718f1af08676ca05743e064bf7e19475ad32a31459feedb68e0a45b091378c8c869665
|
7
|
+
data.tar.gz: 3b80f7d9ca88ed49130a6ef6719e9d65fb28d0dca101a2ee23e191bcb413dcf93ae213fefe194e736169cca2b2ab3e0a67444f975a741a7bf323944c4328fb1d
|
@@ -6,25 +6,16 @@ module BeetleETL
|
|
6
6
|
end
|
7
7
|
|
8
8
|
def run
|
9
|
-
assign_new_ids
|
10
|
-
map_existing_ids
|
11
|
-
end
|
12
|
-
|
13
|
-
def assign_new_ids
|
14
|
-
database.execute <<-SQL
|
15
|
-
UPDATE #{stage_table_name_sql}
|
16
|
-
SET id = nextval('#{table_name}_id_seq')
|
17
|
-
WHERE transition = 'CREATE'
|
18
|
-
SQL
|
19
|
-
end
|
20
|
-
|
21
|
-
def map_existing_ids
|
22
9
|
database.execute <<-SQL
|
23
|
-
UPDATE #{stage_table_name_sql}
|
24
|
-
SET id = public.id
|
25
|
-
FROM #{
|
26
|
-
|
27
|
-
|
10
|
+
UPDATE #{stage_table_name_sql} stage_update
|
11
|
+
SET id = COALESCE(public.id, nextval('#{table_name}_id_seq'))
|
12
|
+
FROM #{stage_table_name_sql} stage
|
13
|
+
LEFT OUTER JOIN #{public_table_name_sql} public
|
14
|
+
on (
|
15
|
+
stage.external_id = public.external_id
|
16
|
+
AND public.external_source = '#{external_source}'
|
17
|
+
)
|
18
|
+
WHERE stage_update.external_id = stage.external_id
|
28
19
|
SQL
|
29
20
|
end
|
30
21
|
|
@@ -12,7 +12,7 @@ module BeetleETL
|
|
12
12
|
end
|
13
13
|
|
14
14
|
def run
|
15
|
-
%w(create update delete
|
15
|
+
%w(create update delete reinstate).each do |transition|
|
16
16
|
public_send(:"load_#{transition}")
|
17
17
|
end
|
18
18
|
end
|
@@ -63,7 +63,7 @@ module BeetleETL
|
|
63
63
|
SQL
|
64
64
|
end
|
65
65
|
|
66
|
-
def
|
66
|
+
def load_reinstate
|
67
67
|
database.execute <<-SQL
|
68
68
|
UPDATE #{public_table_name_sql} public
|
69
69
|
SET
|
@@ -72,7 +72,7 @@ module BeetleETL
|
|
72
72
|
deleted_at = NULL
|
73
73
|
FROM #{stage_table_name_sql} stage
|
74
74
|
WHERE stage.id = public.id
|
75
|
-
AND stage.transition = '
|
75
|
+
AND stage.transition = 'REINSTATE'
|
76
76
|
SQL
|
77
77
|
end
|
78
78
|
|
@@ -11,7 +11,7 @@ module BeetleETL
|
|
11
11
|
end
|
12
12
|
|
13
13
|
def run
|
14
|
-
%w(create
|
14
|
+
%w(create update delete reinstate).each do |transition|
|
15
15
|
public_send(:"transition_#{transition}")
|
16
16
|
end
|
17
17
|
end
|
@@ -29,24 +29,6 @@ module BeetleETL
|
|
29
29
|
SQL
|
30
30
|
end
|
31
31
|
|
32
|
-
def transition_keep
|
33
|
-
database.execute <<-SQL
|
34
|
-
UPDATE #{stage_table_name_sql} stage
|
35
|
-
SET transition = 'KEEP'
|
36
|
-
WHERE EXISTS (
|
37
|
-
SELECT 1
|
38
|
-
FROM #{public_table_name} public
|
39
|
-
WHERE public.external_id = stage.external_id
|
40
|
-
AND public.external_source = '#{external_source}'
|
41
|
-
AND public.deleted_at IS NULL
|
42
|
-
AND
|
43
|
-
(#{public_record_columns.join(', ')})
|
44
|
-
IS NOT DISTINCT FROM
|
45
|
-
(#{stage_record_columns.join(', ')})
|
46
|
-
)
|
47
|
-
SQL
|
48
|
-
end
|
49
|
-
|
50
32
|
def transition_update
|
51
33
|
database.execute <<-SQL
|
52
34
|
UPDATE #{stage_table_name_sql} stage
|
@@ -73,10 +55,7 @@ module BeetleETL
|
|
73
55
|
public.external_id,
|
74
56
|
'DELETE'
|
75
57
|
FROM #{public_table_name_sql} public
|
76
|
-
LEFT OUTER JOIN
|
77
|
-
SELECT *
|
78
|
-
FROM #{stage_table_name_sql}
|
79
|
-
) stage
|
58
|
+
LEFT OUTER JOIN #{stage_table_name_sql} stage
|
80
59
|
ON (stage.external_id = public.external_id)
|
81
60
|
WHERE stage.external_id IS NULL
|
82
61
|
AND public.external_source = '#{external_source}'
|
@@ -84,10 +63,10 @@ module BeetleETL
|
|
84
63
|
SQL
|
85
64
|
end
|
86
65
|
|
87
|
-
def
|
66
|
+
def transition_reinstate
|
88
67
|
database.execute <<-SQL
|
89
68
|
UPDATE #{stage_table_name_sql} stage
|
90
|
-
SET transition = '
|
69
|
+
SET transition = 'REINSTATE'
|
91
70
|
WHERE EXISTS (
|
92
71
|
SELECT 1
|
93
72
|
FROM #{public_table_name_sql} public
|
data/lib/beetle_etl/version.rb
CHANGED
@@ -36,7 +36,7 @@ describe BeetleETL do
|
|
36
36
|
end
|
37
37
|
|
38
38
|
it 'performs all possible transitions', :feature do
|
39
|
-
# create, keep, update, delete,
|
39
|
+
# create, keep, update, delete, reinstate
|
40
40
|
|
41
41
|
import1
|
42
42
|
import2
|
@@ -108,7 +108,7 @@ describe BeetleETL do
|
|
108
108
|
end
|
109
109
|
|
110
110
|
def import3
|
111
|
-
#
|
111
|
+
# reinstate with update
|
112
112
|
insert_into(:source__Organisation).values(
|
113
113
|
[ :pkOrgId , :Name , :Adresse , :Abteilung ] ,
|
114
114
|
[ 1 , 'Apple' , 'Apple Street' , 'iPhone' ] ,
|
@@ -4,6 +4,8 @@ module BeetleETL
|
|
4
4
|
describe AssignIds do
|
5
5
|
|
6
6
|
let(:external_source) { 'my_source' }
|
7
|
+
let(:another_source) { 'another_source' }
|
8
|
+
|
7
9
|
subject { AssignIds.new(:example_table) }
|
8
10
|
|
9
11
|
before do
|
@@ -12,19 +14,6 @@ module BeetleETL
|
|
12
14
|
config.external_source = external_source
|
13
15
|
config.database = test_database
|
14
16
|
end
|
15
|
-
|
16
|
-
test_database.create_table(subject.stage_table_name.to_sym) do
|
17
|
-
Integer :id
|
18
|
-
String :external_id, size: 255
|
19
|
-
String :transition, size: 255
|
20
|
-
end
|
21
|
-
|
22
|
-
test_database.create_table(:example_table) do
|
23
|
-
primary_key :id
|
24
|
-
String :external_id, size: 255
|
25
|
-
String :external_source, size: 255
|
26
|
-
end
|
27
|
-
|
28
17
|
end
|
29
18
|
|
30
19
|
describe '#dependencies' do
|
@@ -34,66 +23,48 @@ module BeetleETL
|
|
34
23
|
end
|
35
24
|
|
36
25
|
describe '#run' do
|
37
|
-
|
38
|
-
|
39
|
-
|
26
|
+
before do
|
27
|
+
test_database.create_table(subject.stage_table_name.to_sym) do
|
28
|
+
Integer :id
|
29
|
+
String :external_id, size: 255
|
30
|
+
String :transition, size: 255
|
40
31
|
end
|
41
32
|
|
42
|
-
|
33
|
+
test_database.create_table(:example_table) do
|
34
|
+
primary_key :id
|
35
|
+
String :external_id, size: 255
|
36
|
+
String :external_source, size: 255
|
37
|
+
end
|
43
38
|
end
|
44
|
-
end
|
45
39
|
|
46
|
-
|
47
|
-
|
48
|
-
|
49
|
-
[ :external_id , :external_source ] ,
|
50
|
-
[ 'keep_id' , external_source ] ,
|
51
|
-
)
|
40
|
+
it 'assigns ids for' do
|
41
|
+
# - generated ones for new records
|
42
|
+
# - mapped ones by external_id for existing records
|
52
43
|
|
53
|
-
insert_into(subject.stage_table_name.to_sym).values(
|
54
|
-
[ :external_id , :transition ] ,
|
55
|
-
[ 'create_id' , 'CREATE' ] ,
|
56
|
-
[ 'keep_id' , 'KEEP' ] ,
|
57
|
-
)
|
58
|
-
|
59
|
-
subject.assign_new_ids
|
60
|
-
|
61
|
-
expect(subject.stage_table_name.to_sym).to have_values(
|
62
|
-
[ :id , :external_id , :transition ] ,
|
63
|
-
[ 2 , 'create_id' , 'CREATE' ] ,
|
64
|
-
[ nil , 'keep_id' , 'KEEP' ] ,
|
65
|
-
)
|
66
|
-
end
|
67
|
-
end
|
68
|
-
|
69
|
-
describe '#map_existing_ids' do
|
70
|
-
it 'assigns ids for existing records by their external id' do
|
71
44
|
insert_into(:example_table).values(
|
72
|
-
[ :external_id
|
73
|
-
[ '
|
74
|
-
[ '
|
75
|
-
[ '
|
76
|
-
[ '
|
45
|
+
[ :external_id , :external_source ] ,
|
46
|
+
[ 'a' , external_source ] ,
|
47
|
+
[ 'b' , external_source ] ,
|
48
|
+
[ 'c' , external_source ] ,
|
49
|
+
[ 'd' , another_source ] ,
|
77
50
|
)
|
78
51
|
|
79
52
|
insert_into(subject.stage_table_name.to_sym).values(
|
80
|
-
[ :external_id
|
81
|
-
[ '
|
82
|
-
[ '
|
83
|
-
[ '
|
84
|
-
[ '
|
85
|
-
[ 'undelete_id' , 'UNDELETE' ] ,
|
53
|
+
[ :external_id ] ,
|
54
|
+
[ 'new value' ] ,
|
55
|
+
[ 'a' ] ,
|
56
|
+
[ 'b' ] ,
|
57
|
+
[ 'c' ] ,
|
86
58
|
)
|
87
59
|
|
88
|
-
subject.
|
60
|
+
subject.run
|
89
61
|
|
90
62
|
expect(subject.stage_table_name.to_sym).to have_values(
|
91
|
-
[ :id , :external_id
|
92
|
-
[
|
93
|
-
[
|
94
|
-
[
|
95
|
-
[
|
96
|
-
[ 4 , 'undelete_id' , 'UNDELETE' ] ,
|
63
|
+
[ :id , :external_id ] ,
|
64
|
+
[ 1 , 'a' ] ,
|
65
|
+
[ 2 , 'b' ] ,
|
66
|
+
[ 3 , 'c' ] ,
|
67
|
+
[ 5 , 'new value' ] ,
|
97
68
|
)
|
98
69
|
end
|
99
70
|
end
|
data/spec/steps/load_spec.rb
CHANGED
@@ -65,7 +65,7 @@ module BeetleETL
|
|
65
65
|
|
66
66
|
describe '#run' do
|
67
67
|
it 'runs all load steps' do
|
68
|
-
%w(create update delete
|
68
|
+
%w(create update delete reinstate).each do |transition|
|
69
69
|
expect(subject).to receive(:"load_#{transition}")
|
70
70
|
end
|
71
71
|
|
@@ -131,7 +131,7 @@ module BeetleETL
|
|
131
131
|
end
|
132
132
|
end
|
133
133
|
|
134
|
-
describe '#
|
134
|
+
describe '#load_reinstate' do
|
135
135
|
it 'restores deleted records' do
|
136
136
|
insert_into(:example_table).values(
|
137
137
|
[ :id , :external_id , :external_source , :foo_id , :created_at , :updated_at , :deleted_at , :payload ] ,
|
@@ -140,10 +140,10 @@ module BeetleETL
|
|
140
140
|
|
141
141
|
insert_into(subject.stage_table_name.to_sym).values(
|
142
142
|
[ :id , :external_id , :transition , :external_foo_id , :foo_id , :payload ] ,
|
143
|
-
[ 1 , 'external_id' , '
|
143
|
+
[ 1 , 'external_id' , 'REINSTATE' , 'foo_id' , 33 , 'updated content' ] ,
|
144
144
|
)
|
145
145
|
|
146
|
-
subject.
|
146
|
+
subject.load_reinstate
|
147
147
|
|
148
148
|
expect(:example_table).to have_values(
|
149
149
|
[ :id , :external_id , :external_source , :foo_id , :created_at , :updated_at , :deleted_at , :payload ] ,
|
@@ -46,7 +46,7 @@ module BeetleETL
|
|
46
46
|
|
47
47
|
describe '#run' do
|
48
48
|
it 'runs all transitions' do
|
49
|
-
%w(create
|
49
|
+
%w(create update delete reinstate).each do |transition|
|
50
50
|
expect(subject).to receive(:"transition_#{transition}")
|
51
51
|
end
|
52
52
|
|
@@ -79,32 +79,6 @@ module BeetleETL
|
|
79
79
|
end
|
80
80
|
end
|
81
81
|
|
82
|
-
describe 'transition_keep' do
|
83
|
-
it 'assigns KEEP if the record already exists and is not deleted, comparing all columns
|
84
|
-
except externald_*_id columns and columns not contained in the stage table' do
|
85
|
-
|
86
|
-
insert_into(:example_table).values(
|
87
|
-
[ :external_id , :external_source , :payload , :ignored_attribute , :foo_id , :deleted_at ] ,
|
88
|
-
[ 'existing' , external_source , 'existing content' , 'ignored content' , 1 , nil ] ,
|
89
|
-
[ 'deleted' , external_source , 'deleted content' , 'ignored content' , 2 , 1.day.ago ] ,
|
90
|
-
)
|
91
|
-
|
92
|
-
insert_into(subject.stage_table_name.to_sym).values(
|
93
|
-
[ :external_id , :payload , :foo_id , :external_foo_id ] ,
|
94
|
-
[ 'existing' , 'existing content' , 1 , 'ignored column' ] ,
|
95
|
-
[ 'deleted' , 'deleted content' , 2 , 'ignored column' ] ,
|
96
|
-
)
|
97
|
-
|
98
|
-
subject.transition_keep
|
99
|
-
|
100
|
-
expect(subject.stage_table_name.to_sym).to have_values(
|
101
|
-
[ :external_id , :transition ] ,
|
102
|
-
[ 'existing' , 'KEEP' ] ,
|
103
|
-
[ 'deleted' , nil ] ,
|
104
|
-
)
|
105
|
-
end
|
106
|
-
end
|
107
|
-
|
108
82
|
describe '#transition_update' do
|
109
83
|
it 'assigns UPDATE to non-deleted records with changed values comparing all columns
|
110
84
|
except externald_*_id columns and columns not contained in the stage table' do
|
@@ -151,8 +125,8 @@ module BeetleETL
|
|
151
125
|
end
|
152
126
|
end
|
153
127
|
|
154
|
-
describe '
|
155
|
-
it 'assigns
|
128
|
+
describe 'transition_reinstate' do
|
129
|
+
it 'assigns REINSTATE to previously deleted records' do
|
156
130
|
insert_into(:example_table).values(
|
157
131
|
[ :external_id , :external_source , :payload , :ignored_attribute , :foo_id , :deleted_at ] ,
|
158
132
|
[ 'existing' , external_source , 'existing content' , 'ignored content' , 1 , nil ] ,
|
@@ -165,12 +139,12 @@ module BeetleETL
|
|
165
139
|
[ 'deleted' , 'updated content' , 2 , 'ignored_column' ] ,
|
166
140
|
)
|
167
141
|
|
168
|
-
subject.
|
142
|
+
subject.transition_reinstate
|
169
143
|
|
170
144
|
expect(subject.stage_table_name.to_sym).to have_values(
|
171
|
-
[ :external_id , :transition
|
172
|
-
[ 'existing' , nil
|
173
|
-
[ 'deleted' , '
|
145
|
+
[ :external_id , :transition ] ,
|
146
|
+
[ 'existing' , nil ] ,
|
147
|
+
[ 'deleted' , 'REINSTATE' ] ,
|
174
148
|
)
|
175
149
|
end
|
176
150
|
end
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: beetle_etl
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.0.
|
4
|
+
version: 0.0.16
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Luciano Maiwald
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2015-04-
|
11
|
+
date: 2015-04-20 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: sequel
|