ncs_mdes_warehouse 0.1.1 → 0.2.0
Sign up to get free protection for your applications and to get access to all the features.
- data/.rspec +2 -0
- data/CHANGELOG.md +20 -0
- data/Rakefile +2 -3
- data/generated_models/ncs_navigator/warehouse/models/two_point_zero/spec_blood.rb +3 -3
- data/lib/ncs_navigator/warehouse.rb +1 -0
- data/lib/ncs_navigator/warehouse/cli.rb +46 -1
- data/lib/ncs_navigator/warehouse/comparator.rb +257 -0
- data/lib/ncs_navigator/warehouse/data_mapper.rb +20 -0
- data/lib/ncs_navigator/warehouse/database_initializer.rb +3 -4
- data/lib/ncs_navigator/warehouse/transform_status.rb +5 -1
- data/lib/ncs_navigator/warehouse/transformers/enum_transformer.rb +18 -3
- data/lib/ncs_navigator/warehouse/updating_shell.rb +1 -0
- data/lib/ncs_navigator/warehouse/version.rb +1 -1
- data/lib/ncs_navigator/warehouse/xml_emitter.rb +95 -13
- data/ncs_mdes_warehouse.gemspec +1 -1
- data/spec/ncs_navigator/warehouse/data_mapper_spec.rb +39 -0
- data/spec/ncs_navigator/warehouse/transform_load_spec.rb +3 -1
- data/spec/ncs_navigator/warehouse/transformers/database_spec.rb +3 -3
- data/spec/ncs_navigator/warehouse/transformers/enum_transformer_spec.rb +36 -5
- data/spec/ncs_navigator/warehouse/xml_emitter_spec.rb +176 -30
- data/spec/spec_helper.rb +54 -17
- metadata +39 -38
data/.rspec
ADDED
data/CHANGELOG.md
CHANGED
@@ -1,6 +1,26 @@
|
|
1
1
|
NCS Navigator MDES Warehouse History
|
2
2
|
====================================
|
3
3
|
|
4
|
+
0.2.0
|
5
|
+
-----
|
6
|
+
|
7
|
+
- Automatically set PSU ID and recruitment type in
|
8
|
+
`Transformers::EnumTransformer` if they are not already set. (#1630,
|
9
|
+
#1648)
|
10
|
+
|
11
|
+
- ETL: store invalid or failing record IDs as a separate column
|
12
|
+
instead of as part of TransformError#message. (#1636)
|
13
|
+
|
14
|
+
- When an MDES model string variable is set from a BigDecimal, coerce
|
15
|
+
it to a floating point string. The ruby default is scientific
|
16
|
+
notation, but that's not permitted in the MDES.
|
17
|
+
|
18
|
+
- New options for `emit-xml`: `--tables`, `--zip`,
|
19
|
+
`--include-pii`. (#1612, #1657, #1658)
|
20
|
+
|
21
|
+
- Add `compare` subcommand in `mdes-wh` for comparing the contents of
|
22
|
+
two warehouses. (#1667)
|
23
|
+
|
4
24
|
0.1.1
|
5
25
|
-----
|
6
26
|
|
data/Rakefile
CHANGED
@@ -15,12 +15,11 @@ task :spec => 'spec:all'
|
|
15
15
|
namespace :spec do
|
16
16
|
RSpec::Core::RakeTask.new(:fast) do |t|
|
17
17
|
t.pattern = "spec/**/*_spec.rb"
|
18
|
-
t.rspec_opts = %q(--
|
18
|
+
t.rspec_opts = %q(--tag ~slow)
|
19
19
|
end
|
20
20
|
|
21
21
|
RSpec::Core::RakeTask.new(:all) do |t|
|
22
22
|
t.pattern = "spec/**/*_spec.rb"
|
23
|
-
t.rspec_opts = %q(--format nested)
|
24
23
|
end
|
25
24
|
end
|
26
25
|
|
@@ -38,7 +37,7 @@ namespace :ci do
|
|
38
37
|
|
39
38
|
task :spec_setup do
|
40
39
|
ENV['CI_REPORTS'] = 'reports/spec-xml'
|
41
|
-
ENV['SPEC_OPTS'] = "#{ENV['SPEC_OPTS']} --format nested"
|
40
|
+
ENV['SPEC_OPTS'] = "#{ENV['SPEC_OPTS']} --format nested --no-color"
|
42
41
|
end
|
43
42
|
|
44
43
|
task :spec => [:spec_setup, 'ci:setup:rspecbase', 'rake:spec']
|
@@ -124,9 +124,9 @@ module NcsNavigator::Warehouse::Models::TwoPointZero
|
|
124
124
|
property :centrifuge_location,
|
125
125
|
NcsNavigator::Warehouse::DataMapper::NcsString,
|
126
126
|
{ :required => true, :length => 1..2, :set => ["1", "2", "-3", "-4"] }
|
127
|
-
|
128
|
-
|
129
|
-
:
|
127
|
+
property :equip_id,
|
128
|
+
NcsNavigator::Warehouse::DataMapper::NcsString,
|
129
|
+
{ :length => 0..36 }
|
130
130
|
property :centrifuge_time,
|
131
131
|
NcsNavigator::Warehouse::DataMapper::NcsString,
|
132
132
|
{ :length => 0..5, :format => /^([0-9][0-9]:[0-9][0-9])?$/ }
|
@@ -7,6 +7,7 @@ module NcsNavigator
|
|
7
7
|
DEFAULT_MDES_VERSION = '2.0'
|
8
8
|
|
9
9
|
autoload :CLI, 'ncs_navigator/warehouse/cli'
|
10
|
+
autoload :Comparator, 'ncs_navigator/warehouse/comparator'
|
10
11
|
autoload :Configuration, 'ncs_navigator/warehouse/configuration'
|
11
12
|
autoload :DataMapper, 'ncs_navigator/warehouse/data_mapper'
|
12
13
|
autoload :DatabaseInitializer, 'ncs_navigator/warehouse/database_initializer'
|
@@ -54,10 +54,20 @@ the contents of the current reporting database. The default name for
|
|
54
54
|
the XML file is the county name for the PSU plus the date; e.g.,
|
55
55
|
cook-20110728.xml.
|
56
56
|
DESC
|
57
|
+
method_option 'block-size', :type => :numeric, :aliases => %w(-b),
|
58
|
+
:desc => 'The maximum number of records to have in memory at once.',
|
59
|
+
:default => 5000
|
60
|
+
method_option 'include-pii', :type => :boolean, :default => false,
|
61
|
+
:desc => 'Include PII values in the emitted XML.'
|
62
|
+
method_option 'zip', :type => :boolean, :default => true,
|
63
|
+
:desc => 'Create a zip file alongside the XML. (Use --no-zip to disable.)'
|
64
|
+
method_option 'tables', :type => :string,
|
65
|
+
:desc => 'Emit XML for a subset of tables.', :banner => 'TABLE,TABLE,TABLE'
|
57
66
|
def emit_xml(filename=nil)
|
58
67
|
use_database
|
59
68
|
|
60
|
-
XmlEmitter.new(configuration, filename
|
69
|
+
XmlEmitter.new(configuration, filename,
|
70
|
+
options.merge(:tables => options[:tables].try(:split, /\s*,\s*/))).emit_xml
|
61
71
|
end
|
62
72
|
|
63
73
|
desc 'etl', 'Performs the full extract-transform-load process for this configuration'
|
@@ -83,6 +93,41 @@ DESC
|
|
83
93
|
exit 1
|
84
94
|
end
|
85
95
|
end
|
96
|
+
|
97
|
+
desc 'compare', 'Compares the contents of two warehouses, A & B.'
|
98
|
+
long_desc <<-DESC
|
99
|
+
Compares the contents of the MDES tables in two warehouses.
|
100
|
+
|
101
|
+
The comparison can be done at three levels:
|
102
|
+
|
103
|
+
1. Record counts.
|
104
|
+
|
105
|
+
2. IDs. (Which record IDs appear in one warehouse and not the other?)
|
106
|
+
|
107
|
+
3. Full contents. (Matching records based on ID, what variables have different values?)
|
108
|
+
|
109
|
+
Each level includes the one before it (i.e., doing a full content
|
110
|
+
comparison also does ID and count comparisons). Higher levels skip
|
111
|
+
tables which would add heat without light; i.e., level 2 only compares
|
112
|
+
IDs for a table when there are at least some records in each warehouse
|
113
|
+
and level 3 only compares content for tables where there are some
|
114
|
+
overlapping IDs.
|
115
|
+
DESC
|
116
|
+
method_option 'warehouse-a', :type => :string, :aliases => %w(-a),
|
117
|
+
:desc => 'The configuration file for warehouse A. The environment default will be used if not specified.'
|
118
|
+
method_option 'warehouse-b', :type => :string, :required => true, :aliases => %w(-b),
|
119
|
+
:desc => 'The configuration file for warehouse B.'
|
120
|
+
method_option 'level', :type => :numeric, :default => 1,
|
121
|
+
:desc => 'The level of detail for the comparison.'
|
122
|
+
def compare
|
123
|
+
if options['warehouse-a']
|
124
|
+
options['config'] = options['warehouse-a']
|
125
|
+
end
|
126
|
+
config_a = configuration
|
127
|
+
config_b = Configuration.from_file(options['warehouse-b'])
|
128
|
+
|
129
|
+
Comparator.new(config_a, config_b, options).compare
|
130
|
+
end
|
86
131
|
end
|
87
132
|
end
|
88
133
|
|
@@ -0,0 +1,257 @@
|
|
1
|
+
require 'ncs_navigator/warehouse'
|
2
|
+
|
3
|
+
require 'forwardable'
|
4
|
+
|
5
|
+
module NcsNavigator::Warehouse
|
6
|
+
class Comparator
|
7
|
+
extend Forwardable
|
8
|
+
|
9
|
+
attr_reader :level
|
10
|
+
attr_reader :a, :b
|
11
|
+
|
12
|
+
def_delegators :a, :shell
|
13
|
+
|
14
|
+
def initialize(config_a, config_b, options={})
|
15
|
+
@a = config_a; @b = config_b
|
16
|
+
@level = options[:level] ||= 1
|
17
|
+
|
18
|
+
shell.clear_line_and_say "Connecting to warehouse A..."
|
19
|
+
DatabaseInitializer.new(@a).set_up_repository(:reporting, 'a')
|
20
|
+
shell.clear_line_and_say "Connecting to warehouse B..."
|
21
|
+
DatabaseInitializer.new(@b).set_up_repository(:reporting, 'b')
|
22
|
+
shell.clear_line_and_say ''
|
23
|
+
end
|
24
|
+
|
25
|
+
def compare
|
26
|
+
shell.say_line "Running level #{level} compare..."
|
27
|
+
count_diff if level >= 1
|
28
|
+
id_diff if level >= 2
|
29
|
+
content_diff if level >= 3
|
30
|
+
end
|
31
|
+
|
32
|
+
def count_diff
|
33
|
+
@count_diffs = collect_differences(CountDiff, models.collect { |m| [m] })
|
34
|
+
shell_summarize_differences(CountDiff, @count_diffs)
|
35
|
+
end
|
36
|
+
|
37
|
+
def id_diff
|
38
|
+
@id_diffs = collect_differences(
|
39
|
+
IdDiff, models.select { |m| no_zero_counts?(m) }.collect { |m| [m] })
|
40
|
+
shell_summarize_differences(IdDiff, @id_diffs)
|
41
|
+
end
|
42
|
+
|
43
|
+
def content_diff
|
44
|
+
@content_diffs = collect_differences(
|
45
|
+
ContentDiff, models.collect { |m| [m, list_shared_ids(m)] }.select { |m, shared| shared })
|
46
|
+
shell_summarize_differences(ContentDiff, @content_diffs)
|
47
|
+
end
|
48
|
+
|
49
|
+
private
|
50
|
+
|
51
|
+
def models
|
52
|
+
@models ||= begin
|
53
|
+
shell.clear_line_and_say "Loading MDES models..."
|
54
|
+
a.models_module.mdes_order
|
55
|
+
end
|
56
|
+
end
|
57
|
+
|
58
|
+
def collect_differences(differ, arg_sets)
|
59
|
+
start = Time.now
|
60
|
+
arg_sets.collect do |arg_set|
|
61
|
+
model = arg_set.first
|
62
|
+
shell.clear_line_and_say "comparing #{differ.thing_plural} for #{model.mdes_table_name}..."
|
63
|
+
differ.new(*arg_set).compare
|
64
|
+
end.compact.tap {
|
65
|
+
shell.clear_line_and_say "#{differ.thing} comparisons complete in %.1fs.\n" % (Time.now - start)
|
66
|
+
}
|
67
|
+
end
|
68
|
+
|
69
|
+
def shell_summarize_differences(differ, diffs)
|
70
|
+
if diffs.empty?
|
71
|
+
puts "There are no #{differ.thing} differences."
|
72
|
+
else
|
73
|
+
puts "#{differ.thing_plural} differ in #{diffs.size} table#{'s' if diffs.size != 1}:"
|
74
|
+
diffs.each do |diff|
|
75
|
+
puts diff.shell_summary
|
76
|
+
end
|
77
|
+
end
|
78
|
+
puts
|
79
|
+
end
|
80
|
+
|
81
|
+
def no_zero_counts?(model)
|
82
|
+
if diff = @count_diffs.find { |d| d.table_name == model.mdes_table_name }
|
83
|
+
diff.a_count != 0 && diff.b_count != 0
|
84
|
+
else
|
85
|
+
true
|
86
|
+
end
|
87
|
+
end
|
88
|
+
|
89
|
+
def list_shared_ids(model)
|
90
|
+
if diff = @id_diffs.find { |d| d.model == model }
|
91
|
+
diff.shared
|
92
|
+
else
|
93
|
+
nil
|
94
|
+
end
|
95
|
+
end
|
96
|
+
|
97
|
+
# @private
|
98
|
+
class CountDiff
|
99
|
+
attr_reader :table_name, :a_count, :b_count
|
100
|
+
|
101
|
+
def self.thing; 'count'; end
|
102
|
+
def self.thing_plural; 'counts'; end
|
103
|
+
|
104
|
+
def initialize(model)
|
105
|
+
@table_name = model.mdes_table_name
|
106
|
+
@q = "SELECT COUNT(*) FROM #{table_name}"
|
107
|
+
end
|
108
|
+
|
109
|
+
def compare
|
110
|
+
@a_count = count('a_reporting')
|
111
|
+
@b_count = count('b_reporting')
|
112
|
+
if difference != 0
|
113
|
+
self
|
114
|
+
end
|
115
|
+
end
|
116
|
+
|
117
|
+
def difference
|
118
|
+
a_count - b_count
|
119
|
+
end
|
120
|
+
|
121
|
+
def count(repo)
|
122
|
+
::DataMapper.repository(repo).adapter.select(@q).first
|
123
|
+
end
|
124
|
+
|
125
|
+
def shell_summary
|
126
|
+
" %30s: A = %6d %s %-6d = B | %6d" % [
|
127
|
+
table_name,
|
128
|
+
a_count, a_count > b_count ? '>' : '<', b_count,
|
129
|
+
difference.abs
|
130
|
+
]
|
131
|
+
end
|
132
|
+
end
|
133
|
+
|
134
|
+
# @private
|
135
|
+
class IdDiff
|
136
|
+
attr_reader :model, :a_ids, :b_ids
|
137
|
+
|
138
|
+
def self.thing; 'ID'; end
|
139
|
+
def self.thing_plural; 'IDs'; end
|
140
|
+
|
141
|
+
def initialize(model)
|
142
|
+
@model = model
|
143
|
+
@q = "SELECT #{model.key.first.name} FROM #{model.mdes_table_name}"
|
144
|
+
end
|
145
|
+
|
146
|
+
def compare
|
147
|
+
@a_ids = ids('a_reporting')
|
148
|
+
@b_ids = ids('b_reporting')
|
149
|
+
unless a_only.empty? && b_only.empty?
|
150
|
+
self
|
151
|
+
end
|
152
|
+
end
|
153
|
+
|
154
|
+
def a_only
|
155
|
+
@a_only ||= a_ids - b_ids
|
156
|
+
end
|
157
|
+
|
158
|
+
def b_only
|
159
|
+
@b_only ||= b_ids - a_ids
|
160
|
+
end
|
161
|
+
|
162
|
+
def shared
|
163
|
+
@shared ||= a_ids & b_ids
|
164
|
+
end
|
165
|
+
|
166
|
+
def ids(repo)
|
167
|
+
::DataMapper.repository(repo).adapter.select(@q)
|
168
|
+
end
|
169
|
+
|
170
|
+
def shell_summary
|
171
|
+
[
|
172
|
+
" #{model.mdes_table_name}",
|
173
|
+
(" #{a_only.size} in A only: #{a_only.join(', ')}" unless a_only.empty?),
|
174
|
+
(" #{b_only.size} in B only: #{b_only.join(', ')}" unless b_only.empty?),
|
175
|
+
" #{shared.size} the same in both A and B."
|
176
|
+
].compact.join("\n")
|
177
|
+
end
|
178
|
+
end
|
179
|
+
|
180
|
+
# @private
|
181
|
+
class ContentDiff
|
182
|
+
BLOCK_SIZE = 5000
|
183
|
+
|
184
|
+
attr_reader :model, :shared_ids
|
185
|
+
|
186
|
+
def self.thing; 'content'; end
|
187
|
+
def self.thing_plural; 'contents'; end
|
188
|
+
|
189
|
+
def initialize(model, shared_ids)
|
190
|
+
@model = model
|
191
|
+
@shared_ids = shared_ids
|
192
|
+
end
|
193
|
+
|
194
|
+
def compare
|
195
|
+
id_blocks.each do |ids|
|
196
|
+
key = model.key.first.name
|
197
|
+
ids_param = "'#{ids.join("', '")}'"
|
198
|
+
find_all = lambda { |repo_name|
|
199
|
+
q = "SELECT * FROM #{model.mdes_table_name} WHERE #{key} IN (#{ids_param}) ORDER BY #{key}"
|
200
|
+
::DataMapper.repository(repo_name).adapter.select(q)
|
201
|
+
}
|
202
|
+
a_records = find_all['a_reporting']
|
203
|
+
b_records = find_all['b_reporting']
|
204
|
+
|
205
|
+
a_records.zip(b_records).each do |a_rec, b_rec|
|
206
|
+
compare_records(a_rec[key], a_rec, b_rec)
|
207
|
+
end
|
208
|
+
end
|
209
|
+
|
210
|
+
unless differences.empty?
|
211
|
+
self
|
212
|
+
end
|
213
|
+
end
|
214
|
+
|
215
|
+
def differences
|
216
|
+
@differences ||= {}
|
217
|
+
end
|
218
|
+
|
219
|
+
def id_blocks
|
220
|
+
@id_blocks ||=
|
221
|
+
begin
|
222
|
+
allocated = 0
|
223
|
+
blocks = []
|
224
|
+
while allocated < shared_ids.size
|
225
|
+
blocks << shared_ids[allocated...(allocated + BLOCK_SIZE)]
|
226
|
+
allocated += BLOCK_SIZE
|
227
|
+
end
|
228
|
+
blocks
|
229
|
+
end
|
230
|
+
end
|
231
|
+
|
232
|
+
def compare_records(id, a_record, b_record)
|
233
|
+
rec_diff = a_record.members.inject({}) { |h, prop|
|
234
|
+
a_value = a_record.send(prop)
|
235
|
+
b_value = b_record.send(prop)
|
236
|
+
h[prop] = [a_value, b_value] if a_value != b_value
|
237
|
+
h
|
238
|
+
}
|
239
|
+
unless rec_diff.empty?
|
240
|
+
differences[id] = rec_diff
|
241
|
+
end
|
242
|
+
end
|
243
|
+
|
244
|
+
def shell_summary
|
245
|
+
[
|
246
|
+
" #{model.mdes_table_name}:",
|
247
|
+
differences.each_pair.collect do |id, rec_diff|
|
248
|
+
[" #{id}:"] +
|
249
|
+
rec_diff.each_pair.collect { |var, vals|
|
250
|
+
" #{var}: #{vals.collect(&:inspect).join(' != ')}"
|
251
|
+
}
|
252
|
+
end
|
253
|
+
].flatten.join("\n")
|
254
|
+
end
|
255
|
+
end
|
256
|
+
end
|
257
|
+
end
|
@@ -23,16 +23,36 @@ module NcsNavigator::Warehouse
|
|
23
23
|
end
|
24
24
|
end
|
25
25
|
|
26
|
+
module NcsStringType
|
27
|
+
protected
|
28
|
+
|
29
|
+
def self.included(into)
|
30
|
+
into.class_eval do
|
31
|
+
alias :original_typecast_to_primitive :typecast_to_primitive
|
32
|
+
|
33
|
+
def typecast_to_primitive(value)
|
34
|
+
if value.is_a?(BigDecimal)
|
35
|
+
value.to_s('F')
|
36
|
+
else
|
37
|
+
original_typecast_to_primitive(value)
|
38
|
+
end
|
39
|
+
end
|
40
|
+
end
|
41
|
+
end
|
42
|
+
end
|
43
|
+
|
26
44
|
##
|
27
45
|
# DataMapper `:string` that's an {NcsType}.
|
28
46
|
class NcsString < ::DataMapper::Property::String
|
29
47
|
include NcsType
|
48
|
+
include NcsStringType
|
30
49
|
end
|
31
50
|
|
32
51
|
##
|
33
52
|
# DataMapper `:text` that's an {NcsType}.
|
34
53
|
class NcsText < ::DataMapper::Property::Text
|
35
54
|
include NcsType
|
55
|
+
include NcsStringType
|
36
56
|
end
|
37
57
|
|
38
58
|
##
|
@@ -44,18 +44,17 @@ module NcsNavigator::Warehouse
|
|
44
44
|
# be the reporting repo unless only the working repo is
|
45
45
|
# configured in.
|
46
46
|
# @return [void]
|
47
|
-
def set_up_repository(mode=:reporting)
|
47
|
+
def set_up_repository(mode=:reporting, prefix="mdes_warehouse")
|
48
48
|
fail "Invalid mode #{mode.inspect}" unless [:reporting, :working, :both].include?(mode)
|
49
49
|
modes = case mode
|
50
50
|
when :both then [:reporting, :working]
|
51
51
|
else [mode]
|
52
52
|
end
|
53
53
|
connect_one(modes.first, :default)
|
54
|
-
modes.each { |m| connect_one(m) }
|
54
|
+
modes.each { |m| connect_one(m, [prefix, m].join('_').to_sym) }
|
55
55
|
end
|
56
56
|
|
57
|
-
def connect_one(which_one, dm_name
|
58
|
-
dm_name ||= :"mdes_warehouse_#{which_one}"
|
57
|
+
def connect_one(which_one, dm_name)
|
59
58
|
log.info "Connecting DataMapper repository #{dm_name.inspect}"
|
60
59
|
p = params(which_one)
|
61
60
|
log.debug " using #{p.merge('password' => 'SUPPRESSED').inspect}"
|