partitioned 0.8.0 → 1.0.1
Sign up to get free protection for your applications and to get access to all the features.
- data/README +85 -36
- data/Rakefile +3 -0
- data/examples/README +46 -18
- data/lib/monkey_patch_activerecord.rb +14 -8
- data/lib/monkey_patch_postgres.rb +46 -13
- data/lib/partitioned/active_record_overrides.rb +13 -5
- data/lib/partitioned/bulk_methods_mixin.rb +91 -146
- data/lib/partitioned/by_created_at.rb +3 -1
- data/lib/partitioned/by_foreign_key.rb +5 -0
- data/lib/partitioned/by_id.rb +10 -4
- data/lib/partitioned/by_integer_field.rb +9 -0
- data/lib/partitioned/by_monthly_time_field.rb +8 -1
- data/lib/partitioned/by_time_field.rb +16 -8
- data/lib/partitioned/by_weekly_time_field.rb +6 -3
- data/lib/partitioned/multi_level/configurator/data.rb +1 -0
- data/lib/partitioned/multi_level/configurator/dsl.rb +11 -0
- data/lib/partitioned/multi_level/configurator/reader.rb +18 -0
- data/lib/partitioned/multi_level/partition_manager.rb +13 -4
- data/lib/partitioned/multi_level.rb +3 -1
- data/lib/partitioned/partitioned_base/configurator/data.rb +10 -1
- data/lib/partitioned/partitioned_base/configurator/dsl.rb +20 -15
- data/lib/partitioned/partitioned_base/configurator/reader.rb +3 -0
- data/lib/partitioned/partitioned_base/configurator.rb +4 -0
- data/lib/partitioned/partitioned_base/partition_manager.rb +17 -15
- data/lib/partitioned/partitioned_base/sql_adapter.rb +25 -23
- data/lib/partitioned/partitioned_base.rb +112 -41
- data/lib/partitioned/version.rb +2 -1
- data/partitioned.gemspec +3 -2
- metadata +68 -73
@@ -1,77 +1,56 @@
|
|
1
1
|
module Partitioned
|
2
|
+
#
|
3
|
+
# MixIn used to extend ActiveRecord::Base classes implementing bulk insert and update operations
|
4
|
+
# through {#create_many} and {#update_many}. {Partitioned::PartitionedBase} classes are extended
|
5
|
+
# with these methods by default.
|
6
|
+
#
|
7
|
+
# @example to use in non-{Partitioned::PartitionedBase} class:
|
8
|
+
# class Company < ActiveRecord::Base
|
9
|
+
# extend Partitioned::BulkMethodsMixin
|
10
|
+
# end
|
11
|
+
#
|
2
12
|
module BulkMethodsMixin
|
13
|
+
# exception thrown when row data structures are inconsistent between rows in single call to {#create_many} or {#update_many}
|
3
14
|
class BulkUploadDataInconsistent < StandardError
|
4
15
|
def initialize(model, table_name, expected_columns, found_columns, while_doing)
|
5
16
|
super("#{model.name}: for table: #{table_name}; #{expected_columns} != #{found_columns}; #{while_doing}")
|
6
17
|
end
|
7
18
|
end
|
8
|
-
|
19
|
+
|
9
20
|
# BULK creation of many rows
|
10
21
|
#
|
11
|
-
#
|
12
|
-
#
|
13
|
-
#
|
14
|
-
#
|
15
|
-
#
|
16
|
-
#
|
17
|
-
#
|
18
|
-
#
|
19
|
-
#
|
20
|
-
#
|
21
|
-
#
|
22
|
-
#
|
23
|
-
#
|
24
|
-
#
|
25
|
-
# :
|
26
|
-
#
|
27
|
-
#
|
28
|
-
#
|
29
|
-
#
|
30
|
-
# }
|
31
|
-
#
|
32
|
-
#
|
33
|
-
#
|
34
|
-
#
|
35
|
-
#
|
36
|
-
#
|
37
|
-
#
|
38
|
-
#
|
39
|
-
#
|
40
|
-
#
|
41
|
-
#
|
42
|
-
#
|
43
|
-
# :salary => 2000,
|
44
|
-
# }]
|
45
|
-
#
|
46
|
-
# options = {
|
47
|
-
# :returning => [:id]
|
48
|
-
# }
|
49
|
-
#
|
50
|
-
# Employee.create_many(rows, options) returns [#<Employee id: 1>, #<Employee id: 2>]
|
51
|
-
#
|
52
|
-
# third example uses :slice_size option.
|
53
|
-
# Slice_size - is an integer that specifies how many
|
54
|
-
# records will be created in a single SQL query.
|
55
|
-
#
|
56
|
-
# rows = [{
|
57
|
-
# :name => 'Keith',
|
58
|
-
# :salary => 1000,
|
59
|
-
# },
|
60
|
-
# {
|
61
|
-
# :name => 'Alex',
|
62
|
-
# :salary => 2000,
|
63
|
-
# },
|
64
|
-
# {
|
65
|
-
# :name => 'Mark',
|
66
|
-
# :salary => 3000,
|
67
|
-
# }]
|
68
|
-
#
|
69
|
-
# options = {
|
70
|
-
# :slice_size => 2
|
71
|
-
# }
|
72
|
-
#
|
73
|
-
# Employee.create_many(rows, options) will generate two insert queries
|
74
|
-
#
|
22
|
+
# @example no options used
|
23
|
+
# rows = [
|
24
|
+
# { :name => 'Keith', :salary => 1000 },
|
25
|
+
# { :name => 'Alex', :salary => 2000 }
|
26
|
+
# ]
|
27
|
+
# Employee.create_many(rows)
|
28
|
+
#
|
29
|
+
# @example with :returning option to returns key value
|
30
|
+
# rows = [
|
31
|
+
# { :name => 'Keith', :salary => 1000 },
|
32
|
+
# { :name => 'Alex', :salary => 2000 }
|
33
|
+
# ]
|
34
|
+
# options = { :returning => [:id] }
|
35
|
+
# Employee.create_many(rows, options)
|
36
|
+
# [#<Employee id: 1>, #<Employee id: 2>]
|
37
|
+
#
|
38
|
+
# @example with :slice_size option (will generate two insert queries)
|
39
|
+
# rows = [
|
40
|
+
# { :name => 'Keith', :salary => 1000 },
|
41
|
+
# { :name => 'Alex', :salary => 2000 },
|
42
|
+
# { :name => 'Mark', :salary => 3000 }
|
43
|
+
# ]
|
44
|
+
# options = { :slice_size => 2 }
|
45
|
+
# Employee.create_many(rows, options)
|
46
|
+
#
|
47
|
+
# @param [Array<Hash>] rows ([]) data to be inserted into database
|
48
|
+
# @param [Hash] options ({}) options for bulk inserts
|
49
|
+
# @option options [Integer] :slice_size (1000) how many records will be created in a single SQL query
|
50
|
+
# @option options [Boolean] :check_consitency (true) ensure some modicum of sanity on the incoming dataset, specifically: does each row define the same set of key/value pairs
|
51
|
+
# @option options [Array or String] :returning (nil) list of fields to return.
|
52
|
+
# @return [Array<Hash>] rows returned from DB as option[:returning] requests
|
53
|
+
# @raise [BulkUploadDataInconsistent] raised when key/value pairs between rows are inconsistent (check disabled with option :check_consistency)
|
75
54
|
def create_many(rows, options = {})
|
76
55
|
return [] if rows.blank?
|
77
56
|
options[:slice_size] = 1000 unless options.has_key?(:slice_size)
|
@@ -127,90 +106,56 @@ module Partitioned
|
|
127
106
|
#
|
128
107
|
# BULK updates of many rows
|
129
108
|
#
|
130
|
-
#
|
131
|
-
#
|
132
|
-
#
|
133
|
-
#
|
134
|
-
# options:
|
135
|
-
#
|
136
|
-
#
|
137
|
-
#
|
138
|
-
#
|
139
|
-
#
|
140
|
-
#
|
141
|
-
#
|
142
|
-
#
|
143
|
-
#
|
144
|
-
#
|
145
|
-
# rows
|
146
|
-
# :
|
147
|
-
#
|
148
|
-
#
|
149
|
-
#
|
150
|
-
#
|
151
|
-
#
|
152
|
-
# }
|
153
|
-
#
|
154
|
-
# :
|
155
|
-
#
|
156
|
-
#
|
157
|
-
#
|
158
|
-
#
|
159
|
-
#
|
160
|
-
# }
|
161
|
-
#
|
162
|
-
#
|
163
|
-
#
|
164
|
-
#
|
165
|
-
#
|
166
|
-
# rows = [{
|
167
|
-
# :id => 1,
|
168
|
-
# :salary => 1000,
|
169
|
-
# :company_id => 10
|
170
|
-
# },
|
171
|
-
# {
|
172
|
-
# :id => 10,
|
173
|
-
# :salary => 2000,
|
174
|
-
# :company_id => 12
|
175
|
-
# },
|
176
|
-
# {
|
177
|
-
# :id => 23,
|
178
|
-
# :salary => 2500,
|
179
|
-
# :company_id => 5
|
180
|
-
# }]
|
181
|
-
#
|
182
|
-
# options = {
|
183
|
-
# :set_array => '"company_id = datatable.company_id"',
|
184
|
-
# :where => '"#{table_name}.salary = datatable.salary"'
|
185
|
-
# }
|
186
|
-
#
|
187
|
-
# Employee.update_many(rows, options)
|
188
|
-
#
|
189
|
-
#
|
190
|
-
# this version sets the where clause to the KEY of the hash passed in
|
191
|
-
# and the set_array is generated from the VALUES
|
192
|
-
#
|
193
|
-
# rows = {
|
194
|
-
# { :id => 1 } => {
|
195
|
-
# :salary => 100000,
|
196
|
-
# :company_id => 10
|
197
|
-
# },
|
198
|
-
# { :id => 10 } => {
|
199
|
-
# :salary => 110000,
|
200
|
-
# :company_id => 12
|
201
|
-
# },
|
202
|
-
# { :id => 23 } => {
|
203
|
-
# :salary => 90000,
|
204
|
-
# :company_id => 5
|
109
|
+
# @return [Array<Hash>] rows returned from DB as option[:returning] requests
|
110
|
+
# @raise [BulkUploadDataInconsistent] raised when key/value pairs between rows are inconsistent (check disabled with option :check_consistency)
|
111
|
+
# @param [Hash] options ({}) options for bulk inserts
|
112
|
+
# @option options [Integer] :slice_size (1000) how many records will be created in a single SQL query
|
113
|
+
# @option options [Boolean] :check_consitency (true) ensure some modicum of sanity on the incoming dataset, specifically: does each row define the same set of key/value pairs
|
114
|
+
# @option options [Array] :returning (nil) list of fields to return.
|
115
|
+
# @option options [String] :returning (nil) single field to return.
|
116
|
+
#
|
117
|
+
# @overload update_many(rows = [], options = {})
|
118
|
+
# @param [Array<Hash>] rows ([]) data to be updated
|
119
|
+
# @option options [String] :set_array (built from first row passed in) the set clause
|
120
|
+
# @option options [String] :where ('"#{table_name}.id = datatable.id"') the where clause
|
121
|
+
#
|
122
|
+
# @overload update_many(rows = {}, options = {})
|
123
|
+
# @param [Hash<Hash, Hash>] rows ({}) data to be updated
|
124
|
+
# @option options [String] :set_array (built from the values in the first key/value pair of `rows`) the set clause
|
125
|
+
# @option options [String] :where (built from the keys in the first key/value pair of `rows`) the where clause
|
126
|
+
#
|
127
|
+
# @example using "set_array" to add the value of "salary" to the specific employee's salary the default where clause matches IDs so, it works here.
|
128
|
+
# rows = [
|
129
|
+
# { :id => 1, :salary => 1000 },
|
130
|
+
# { :id => 10, :salary => 2000 },
|
131
|
+
# { :id => 23, :salary => 2500 }
|
132
|
+
# ]
|
133
|
+
# options = { :set_array => '"salary = datatable.salary"' }
|
134
|
+
# Employee.update_many(rows, options)
|
135
|
+
#
|
136
|
+
# @example using where clause to match salary.
|
137
|
+
# rows = [
|
138
|
+
# { :id => 1, :salary => 1000, :company_id => 10 },
|
139
|
+
# { :id => 10, :salary => 2000, :company_id => 12 },
|
140
|
+
# { :id => 23, :salary => 2500, :company_id => 5 }
|
141
|
+
# ]
|
142
|
+
# options = {
|
143
|
+
# :set_array => '"company_id = datatable.company_id"',
|
144
|
+
# :where => '"#{table_name}.salary = datatable.salary"'
|
205
145
|
# }
|
206
|
-
#
|
146
|
+
# Employee.update_many(rows, options)
|
207
147
|
#
|
208
|
-
#
|
148
|
+
# @example setting where clause to the KEY of the hash passed in and the set_array is generated from the VALUES
|
149
|
+
# rows = {
|
150
|
+
# { :id => 1 } => { :salary => 100000, :company_id => 10 },
|
151
|
+
# { :id => 10 } => { :salary => 110000, :company_id => 12 },
|
152
|
+
# { :id => 23 } => { :salary => 90000, :company_id => 5 }
|
153
|
+
# }
|
154
|
+
# Employee.update_many(rows)
|
209
155
|
#
|
210
|
-
# Remember that you should probably set updated_at using "updated = datatable.updated_at"
|
211
|
-
#
|
212
|
-
#
|
213
|
-
|
156
|
+
# @note Remember that you should probably set updated_at using "updated = datatable.updated_at"
|
157
|
+
# or "updated_at = now()" in the set_array if you want to follow
|
158
|
+
# the standard active record model for time columns (and you have an updated_at column)
|
214
159
|
def update_many(rows, options = {})
|
215
160
|
return [] if rows.blank?
|
216
161
|
if rows.is_a?(Hash)
|
@@ -1,11 +1,13 @@
|
|
1
1
|
module Partitioned
|
2
2
|
#
|
3
|
-
#
|
3
|
+
# Partition tables by created_at grouping them by week, with
|
4
4
|
# a week defined as seven days starting on Monday.
|
5
5
|
#
|
6
6
|
class ByCreatedAt < ByWeeklyTimeField
|
7
7
|
self.abstract_class = true
|
8
8
|
|
9
|
+
# the field to partition on, `created_at`
|
10
|
+
# @return [Symbol] the partition field: `created_at`
|
9
11
|
def self.partition_time_field
|
10
12
|
return :created_at
|
11
13
|
end
|
@@ -1,11 +1,16 @@
|
|
1
1
|
module Partitioned
|
2
|
+
# Partitioned abstract class for all partitioned models based as a single integer field value that is used as a foreign key
|
2
3
|
class ByForeignKey < ByIntegerField
|
3
4
|
self.abstract_class = true
|
4
5
|
|
6
|
+
# the field to partition on
|
7
|
+
# @return [Integer] re-routed to {#self.partition_foreign_key}
|
5
8
|
def self.partition_integer_field
|
6
9
|
return partition_foreign_key
|
7
10
|
end
|
8
11
|
|
12
|
+
# the field to partition on
|
13
|
+
# @return [String] the name of the foreign key field
|
9
14
|
def self.partition_foreign_key
|
10
15
|
raise MethodNotImplemented.new(self, :partition_foreign_key)
|
11
16
|
end
|
data/lib/partitioned/by_id.rb
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
module Partitioned
|
2
2
|
#
|
3
|
-
#
|
4
|
-
# the value of its primary key.
|
3
|
+
# Table partitioning by id. this partitioning breaks up data by
|
4
|
+
# the value of its primary key. A specific record's child table
|
5
5
|
# is determined by the number resulting from the integer math:
|
6
6
|
# ID / ById::partition_table_size * ById::partition_table_size
|
7
7
|
#
|
@@ -9,21 +9,27 @@ module Partitioned
|
|
9
9
|
self.abstract_class = true
|
10
10
|
|
11
11
|
#
|
12
|
-
#
|
12
|
+
# Specific to this partitioning, we need to prefetch the primary key (id)
|
13
13
|
# before we attempt to do the insert because the insert wants to know the
|
14
14
|
# name of the specific child table to access.
|
15
15
|
#
|
16
|
+
# @return [Boolean] true
|
16
17
|
def self.prefetch_primary_key?
|
17
18
|
return true
|
18
19
|
end
|
19
20
|
|
20
21
|
#
|
21
|
-
#
|
22
|
+
# The number of records in each child table.
|
22
23
|
#
|
24
|
+
# @return [Integer] the number of rows in a partition
|
23
25
|
def self.partition_table_size
|
24
26
|
return 10000000
|
25
27
|
end
|
26
28
|
|
29
|
+
#
|
30
|
+
# The name of the field to partition on
|
31
|
+
#
|
32
|
+
# @return [String] the name of the field to partition on
|
27
33
|
def self.partition_integer_field
|
28
34
|
return :id
|
29
35
|
end
|
@@ -1,15 +1,24 @@
|
|
1
1
|
module Partitioned
|
2
|
+
#
|
3
|
+
# Partitioned abstract class for all partitioned models based as a single integer field value.
|
4
|
+
#
|
2
5
|
class ByIntegerField < PartitionedBase
|
3
6
|
self.abstract_class = true
|
4
7
|
|
8
|
+
# the size of each table
|
9
|
+
# @return [Integer] how many different values are in each partition
|
5
10
|
def self.partition_table_size
|
6
11
|
return 1
|
7
12
|
end
|
8
13
|
|
14
|
+
# the name of the partition key field
|
15
|
+
# @return [String] the name of the field
|
9
16
|
def self.partition_integer_field
|
10
17
|
raise MethodNotImplemented.new(self, :partition_integer_field)
|
11
18
|
end
|
12
19
|
|
20
|
+
# the normalized key value for a given key value
|
21
|
+
# @return [Integer] the normalized value
|
13
22
|
def self.partition_normalize_key_value(integer_field_value)
|
14
23
|
return integer_field_value / partition_table_size * partition_table_size
|
15
24
|
end
|
@@ -1,15 +1,22 @@
|
|
1
1
|
module Partitioned
|
2
2
|
#
|
3
|
-
#
|
3
|
+
# Partition tables by a time field grouping them by week, with
|
4
4
|
# a week defined as seven days starting on Monday.
|
5
5
|
#
|
6
6
|
class ByMonthlyTimeField < ByTimeField
|
7
7
|
self.abstract_class = true
|
8
8
|
|
9
|
+
# Normalize a partition key value by month.
|
10
|
+
#
|
11
|
+
# @param [Time] time_value the time value to normalize
|
12
|
+
# @return [Time] the value normalized
|
9
13
|
def self.partition_normalize_key_value(time_value)
|
10
14
|
return time_value.at_beginning_of_month
|
11
15
|
end
|
12
16
|
|
17
|
+
# The size of the partition table, a month
|
18
|
+
#
|
19
|
+
# @return [Integer] the size of this partition
|
13
20
|
def self.partition_table_size
|
14
21
|
return 1.month
|
15
22
|
end
|
@@ -1,16 +1,20 @@
|
|
1
1
|
module Partitioned
|
2
2
|
#
|
3
|
-
#
|
3
|
+
# Partition tables by a time field grouping them by day.
|
4
4
|
#
|
5
5
|
class ByTimeField < PartitionedBase
|
6
6
|
self.abstract_class = true
|
7
7
|
|
8
8
|
#
|
9
|
-
#
|
10
|
-
# start_date and end_date skipping step
|
9
|
+
# Generate an enumerable that represents all the dates between
|
10
|
+
# start_date and end_date skipping step.
|
11
11
|
#
|
12
|
-
#
|
12
|
+
# This can be used to calls that take an enumerable like create_infrastructure.
|
13
13
|
#
|
14
|
+
# @param [Date] start_date the first date to generate the range from
|
15
|
+
# @param [Date] end_date the last date to generate the range from
|
16
|
+
# @param [Object] step (:default) number of values to advance (:default means use {#self.partition_table_size}).
|
17
|
+
# @return [Enumerable] the range generated
|
14
18
|
def self.partition_generate_range(start_date, end_date, step = :default)
|
15
19
|
step = partition_table_size if step == :default
|
16
20
|
current_date = partition_normalize_key_value(start_date)
|
@@ -23,23 +27,27 @@ module Partitioned
|
|
23
27
|
end
|
24
28
|
|
25
29
|
#
|
26
|
-
#
|
30
|
+
# Normalize the value to the current day.
|
27
31
|
#
|
32
|
+
# @param [Time] time_value the partitioned key value
|
33
|
+
# @return [Time] time_value normalized
|
28
34
|
def self.partition_normalize_key_value(time_value)
|
29
35
|
return time_value.to_date
|
30
36
|
end
|
31
37
|
|
32
38
|
#
|
33
|
-
#
|
39
|
+
# The size of the partition, 1.day
|
34
40
|
#
|
41
|
+
# @return [Integer] the size of the partition
|
35
42
|
def self.partition_table_size
|
36
43
|
return 1.day
|
37
44
|
end
|
38
45
|
|
39
46
|
#
|
40
|
-
#
|
41
|
-
#
|
47
|
+
# Abstract -- implement in a derived class.
|
48
|
+
# The name of the time-related field we will use to partition child tables.
|
42
49
|
#
|
50
|
+
# @raise MethodNotImplemented
|
43
51
|
def self.partition_time_field
|
44
52
|
raise MethodNotImplemented.new(self, :partition_time_field)
|
45
53
|
end
|
@@ -1,15 +1,17 @@
|
|
1
1
|
module Partitioned
|
2
2
|
#
|
3
|
-
#
|
3
|
+
# Partition tables by a time field grouping them by week, with
|
4
4
|
# a week defined as seven days starting on Monday.
|
5
5
|
#
|
6
6
|
class ByWeeklyTimeField < ByTimeField
|
7
7
|
self.abstract_class = true
|
8
8
|
|
9
9
|
#
|
10
|
-
#
|
11
|
-
# the
|
10
|
+
# Normalize a partition key value by week. We've picked
|
11
|
+
# the beginning of the week to key on, which is Monday.
|
12
12
|
#
|
13
|
+
# @param [Time] time_value the time value to normalize
|
14
|
+
# @return [Time] the value normalized
|
13
15
|
def self.partition_normalize_key_value(time_value)
|
14
16
|
return time_value.at_beginning_of_week
|
15
17
|
end
|
@@ -17,6 +19,7 @@ module Partitioned
|
|
17
19
|
#
|
18
20
|
# The size of the partition table, 7 days (1.week)
|
19
21
|
#
|
22
|
+
# @return [Integer] the size of this partition
|
20
23
|
def self.partition_table_size
|
21
24
|
return 1.week
|
22
25
|
end
|
@@ -1,7 +1,15 @@
|
|
1
1
|
module Partitioned
|
2
2
|
class MultiLevel
|
3
|
+
#
|
4
|
+
# Configuration manager for multi-level partitioning
|
5
|
+
# it supports, the front-end UI (a Domain Specific Language) using {Dsl}
|
6
|
+
# state using {Data}
|
7
|
+
# and a parser using {Reader}
|
8
|
+
#
|
3
9
|
module Configurator
|
10
|
+
# The Domain Specific Language UI manager for multi level partitioning classes
|
4
11
|
class Dsl < Partitioned::PartitionedBase::Configurator::Dsl
|
12
|
+
# used to prevent use of invalid directives
|
5
13
|
class InvalidForMultiLevelPartitioning < StandardError
|
6
14
|
def initialize(model, dsl_key, remedy)
|
7
15
|
super("#{model.name}: '#{dsl_key}' is not valid for multi-level partitioning. #{remedy}")
|
@@ -18,10 +26,13 @@ module Partitioned
|
|
18
26
|
#
|
19
27
|
# Definition of classes which will be used at multi level partitioning.
|
20
28
|
#
|
29
|
+
# @param [*Array<Class>] classes the classes, in order, used to partition this model
|
30
|
+
# @return [optional]
|
21
31
|
def using_classes(*classes)
|
22
32
|
data.using_classes += [*classes]
|
23
33
|
end
|
24
34
|
|
35
|
+
# @raise [InvalidForMultiLevelPartitioning] always raised. `on` is not a valid DSL directive for multi-level partitioning
|
25
36
|
def on(*ignored)
|
26
37
|
raise InvalidForMultiLevelPartitioning.new(model, :on, "the partitioned keyword 'using' is used to define multi-level partitioned tables.")
|
27
38
|
end
|
@@ -1,8 +1,13 @@
|
|
1
1
|
module Partitioned
|
2
2
|
class MultiLevel
|
3
3
|
module Configurator
|
4
|
+
# coalesces and parses all {Data} objects allowing the
|
5
|
+
# {PartitionManager} to request partitioning information froma
|
6
|
+
# centralized source from multi level partitioned models
|
4
7
|
class Reader < Partitioned::PartitionedBase::Configurator::Reader
|
8
|
+
# configurator for a specific class level
|
5
9
|
UsingConfigurator = Struct.new(:model, :sliced_class, :dsl)
|
10
|
+
|
6
11
|
def initialize(most_derived_activerecord_class)
|
7
12
|
super
|
8
13
|
@using_classes = nil
|
@@ -12,6 +17,7 @@ module Partitioned
|
|
12
17
|
#
|
13
18
|
# The field used to partition child tables.
|
14
19
|
#
|
20
|
+
# @return [Array<Symbol>] fields used to partition this model
|
15
21
|
def on_fields
|
16
22
|
unless @on_fields
|
17
23
|
@on_fields = using_collect(&:on_field).map(&:to_sym)
|
@@ -72,10 +78,22 @@ module Partitioned
|
|
72
78
|
return parts.join('_')
|
73
79
|
end
|
74
80
|
|
81
|
+
# retrieve a specific configurator from an ordered list. for multi-level partitioning
|
82
|
+
# we need to find the specific configurator for the partitioning level we are interested
|
83
|
+
# in managing.
|
84
|
+
#
|
85
|
+
# @param [Integer] index the partitioning level to query
|
86
|
+
# @return [Configurator] the configurator for the specific level queried
|
75
87
|
def using_configurator(index)
|
76
88
|
return using_class(index).configurator
|
77
89
|
end
|
78
90
|
|
91
|
+
# retrieve a specific partitioning class from an ordered list. for multi-level partitioning
|
92
|
+
# we need to find the specific {Partitioned::PartitionedBase} class for the partitioning level we are interested
|
93
|
+
# in managing.
|
94
|
+
#
|
95
|
+
# @param [Integer] index the partitioning level to query
|
96
|
+
# @return [{Partitioned::PartitionedBase}] the class for the specific level queried
|
79
97
|
def using_class(index)
|
80
98
|
return using_classes[index]
|
81
99
|
end
|
@@ -1,10 +1,15 @@
|
|
1
1
|
module Partitioned
|
2
2
|
class MultiLevel
|
3
|
+
#
|
4
|
+
# the manger of partitioned requests for models partitioned multiple times
|
5
|
+
#
|
3
6
|
class PartitionManager < Partitioned::PartitionedBase::PartitionManager
|
4
7
|
#
|
5
|
-
#
|
8
|
+
# The once called function to prepare a parent table for partitioning as well
|
6
9
|
# as create the schema that the child tables will be placed in.
|
7
10
|
#
|
11
|
+
# @param [Enumerable] enumerable (Array<Array>) the key values that should be used to create the parent partition tables.
|
12
|
+
# @return [optional]
|
8
13
|
def create_infrastructure(enumerable = [[]])
|
9
14
|
super()
|
10
15
|
enumerable.each do |*partition_key_values|
|
@@ -15,11 +20,13 @@ module Partitioned
|
|
15
20
|
protected
|
16
21
|
|
17
22
|
#
|
18
|
-
#
|
23
|
+
# Create a specific child table that does not currently
|
19
24
|
# exist and whose schema (the schema that the table exists in)
|
20
25
|
# also already exists (#create_infrastructure is designed to
|
21
26
|
# create this).
|
22
27
|
#
|
28
|
+
# @param [*Array<Object>] partition_key_values all key values needed to create a partition
|
29
|
+
# @return [optional]
|
23
30
|
def create_new_partition(*partition_key_values)
|
24
31
|
create_partition_table(*partition_key_values)
|
25
32
|
if is_leaf_partition?(*partition_key_values)
|
@@ -31,14 +38,16 @@ module Partitioned
|
|
31
38
|
end
|
32
39
|
|
33
40
|
#
|
34
|
-
#
|
41
|
+
# Is the table a child table without itself having any children.
|
35
42
|
# generally leaf tables are where all indexes and foreign key
|
36
43
|
# constraints will be placed because that is where the data will be.
|
37
44
|
#
|
38
45
|
# Non leaf tables will typically have a rule placed on them
|
39
|
-
# (via add_parent_table_rules) that prevents any inserts from
|
46
|
+
# (via add_parent_table_rules) that prevents any inserts from occurring
|
40
47
|
# on them.
|
41
48
|
#
|
49
|
+
# @param [*Array<Object>] partition_key_values all key values specifying a given child table
|
50
|
+
# @return [Boolean] true if this partition should contain records
|
42
51
|
def is_leaf_partition?(*partition_key_values)
|
43
52
|
return partition_key_values.length == parent_table_class.configurator.on_fields.length
|
44
53
|
end
|
@@ -1,6 +1,6 @@
|
|
1
1
|
module Partitioned
|
2
2
|
#
|
3
|
-
#
|
3
|
+
# Table partitioning by a referenced id column which itself is partitioned
|
4
4
|
# further weekly by a date column.
|
5
5
|
#
|
6
6
|
class MultiLevel < PartitionedBase
|
@@ -9,6 +9,8 @@ module Partitioned
|
|
9
9
|
#
|
10
10
|
# Normalize the values for the each of using class.
|
11
11
|
#
|
12
|
+
# @param [Array<Object>] value the partition key values
|
13
|
+
# @return [Array<Object>] the normalized values for the key values passed in
|
12
14
|
def self.partition_normalize_key_value(values)
|
13
15
|
normalized_values = []
|
14
16
|
[*values].each_with_index do |value,index|
|
@@ -2,9 +2,10 @@ module Partitioned
|
|
2
2
|
class PartitionedBase
|
3
3
|
module Configurator
|
4
4
|
#
|
5
|
-
#
|
5
|
+
# The state configured by the Dsl and read by Reader.
|
6
6
|
#
|
7
7
|
class Data
|
8
|
+
# represents a SQL index
|
8
9
|
class Index
|
9
10
|
attr_accessor :field, :options
|
10
11
|
def initialize(field, options = {})
|
@@ -12,6 +13,7 @@ module Partitioned
|
|
12
13
|
@options = options
|
13
14
|
end
|
14
15
|
end
|
16
|
+
# represents a SQL foreign key reference
|
15
17
|
class ForeignKey
|
16
18
|
attr_accessor :referencing_field, :referenced_table, :referenced_field
|
17
19
|
def initialize(referencing_field, referenced_table = nil, referenced_field = :id)
|
@@ -24,6 +26,13 @@ module Partitioned
|
|
24
26
|
@referenced_field = referenced_field
|
25
27
|
end
|
26
28
|
|
29
|
+
#
|
30
|
+
# Produce a table name from the name of the foreign key. in rails, this really
|
31
|
+
# means "foo_id" should be mapped to "foos", and "company_id" should be mapped to
|
32
|
+
# "companies"
|
33
|
+
#
|
34
|
+
# @param [String] foreign_key_field the name of the foreign key field
|
35
|
+
# @return [String] the name of the table associated with the foreign key
|
27
36
|
def self.foreign_key_to_foreign_table_name(foreign_key_field)
|
28
37
|
return ActiveSupport::Inflector::pluralize(foreign_key_field.to_s.sub(/_id$/,''))
|
29
38
|
end
|