Module: ActiveRecord::Batches

Included in:
Relation
Defined in:
lib/active_record/relation/batches.rb,
lib/active_record/relation/batches/batch_enumerator.rb

Overview

Active Record Batches

Defined Under Namespace

Classes: BatchEnumerator

Constant Summary collapse

ORDER_IGNORE_MESSAGE = 
"Scoped order is ignored, use :cursor with :order to configure custom order."
DEFAULT_ORDER = 
:asc

Instance Method Summary collapse

Instance Method Details

#find_each(start: nil, finish: nil, batch_size: 1000, error_on_ignore: nil, cursor: primary_key, order: DEFAULT_ORDER, &block) ⇒ Object

Looping through a collection of records from the database (using the Scoping::Named::ClassMethods.all method, for example) is very inefficient since it will try to instantiate all the objects at once.

In that case, batch processing methods allow you to work with the records in batches, thereby greatly reducing memory consumption.

The #find_each method uses #find_in_batches with a batch size of 1000 (or as specified by the :batch_size option).

Person.find_each do |person|
  person.do_awesome_stuff
end

Person.where("age > 21").find_each do |person|
  person.party_all_night!
end

If you do not provide a block to #find_each, it will return an Enumerator for chaining with other methods:

Person.find_each.with_index do |person, index|
  person.award_trophy(index + 1)
end

Options

  • :batch_size - Specifies the size of the batch. Defaults to 1000.

  • :start - Specifies the cursor column value to start from, inclusive of the value.

  • :finish - Specifies the cursor column value to end at, inclusive of the value.

  • :error_on_ignore - Overrides the application config to specify if an error should be raised when an order is present in the relation.

  • :cursor - Specifies the column to use for batching (can be a column name or an array of column names). Defaults to primary key.

  • :order - Specifies the cursor column order (can be :asc or :desc or an array consisting of :asc or :desc). Defaults to :asc.

    class Order < ActiveRecord::Base
  self.primary_key = [:id_1, :id_2]
end

Order.find_each(order: [:asc, :desc])

    In the above code, id_1 is sorted in ascending order and id_2 in descending order.

Limits are honored, and if present there is no requirement for the batch size: it can be less than, equal to, or greater than the limit.

The options start and finish are especially useful if you want multiple workers dealing with the same processing queue. You can make worker 1 handle all the records between id 1 and 9999 and worker 2 handle from 10000 and beyond by setting the :start and :finish option on each worker.

# In worker 1, let's process until 9999 records.
Person.find_each(finish: 9_999) do |person|
  person.party_all_night!
end

# In worker 2, let's process from record 10_000 and onwards.
Person.find_each(start: 10_000) do |person|
  person.party_all_night!
end

NOTE: Order can be ascending (:asc) or descending (:desc). It is automatically set to ascending on the primary key (“id ASC”). This also means that this method only works when the cursor column is orderable (e.g. an integer or string).

NOTE: When using custom columns for batching, they should include at least one unique column (e.g. primary key) as a tiebreaker. Also, to reduce the likelihood of race conditions, all columns should be static (unchangeable after it was set).

NOTE: By its nature, batch processing is subject to race conditions if other processes are modifying the database.



85
86
87
88
89
90
91
92
93
94
95
96
97
 
# File 'lib/active_record/relation/batches.rb', line 85

def find_each(start: nil, finish: nil, batch_size: 1000, error_on_ignore: nil, cursor: primary_key, order: DEFAULT_ORDER, &block)
  if block_given?
    find_in_batches(start: start, finish: finish, batch_size: batch_size, error_on_ignore: error_on_ignore, cursor: cursor, order: order) do |records|
      records.each(&block)
    end
  else
    enum_for(:find_each, start: start, finish: finish, batch_size: batch_size, error_on_ignore: error_on_ignore, cursor: cursor, order: order) do
      relation = self
      cursor = Array(cursor)
      apply_limits(relation, cursor, start, finish, build_batch_orders(cursor, order)).size
    end
  end
end

#find_in_batches(start: nil, finish: nil, batch_size: 1000, error_on_ignore: nil, cursor: primary_key, order: DEFAULT_ORDER) ⇒ Object

Yields each batch of records that was found by the find options as an array.

Person.where("age > 21").find_in_batches do |group|
  sleep(50) # Make sure it doesn't get too crowded in there!
  group.each { |person| person.party_all_night! }
end

If you do not provide a block to #find_in_batches, it will return an Enumerator for chaining with other methods:

Person.find_in_batches.with_index do |group, batch|
  puts "Processing group ##{batch}"
  group.each(&:recover_from_last_night!)
end

To be yielded each record one by one, use #find_each instead.

Options

  • :batch_size - Specifies the size of the batch. Defaults to 1000.

  • :start - Specifies the cursor column value to start from, inclusive of the value.

  • :finish - Specifies the cursor column value to end at, inclusive of the value.

  • :error_on_ignore - Overrides the application config to specify if an error should be raised when an order is present in the relation.

  • :cursor - Specifies the column to use for batching (can be a column name or an array of column names). Defaults to primary key.

  • :order - Specifies the cursor column order (can be :asc or :desc or an array consisting of :asc or :desc). Defaults to :asc.

    class Order < ActiveRecord::Base
  self.primary_key = [:id_1, :id_2]
end

Order.find_in_batches(order: [:asc, :desc])

    In the above code, id_1 is sorted in ascending order and id_2 in descending order.

Limits are honored, and if present there is no requirement for the batch size: it can be less than, equal to, or greater than the limit.

The options start and finish are especially useful if you want multiple workers dealing with the same processing queue. You can make worker 1 handle all the records between id 1 and 9999 and worker 2 handle from 10000 and beyond by setting the :start and :finish option on each worker.

# Let's process from record 10_000 on.
Person.find_in_batches(start: 10_000) do |group|
  group.each { |person| person.party_all_night! }
end

NOTE: Order can be ascending (:asc) or descending (:desc). It is automatically set to ascending on the primary key (“id ASC”). This also means that this method only works when the cursor column is orderable (e.g. an integer or string).

NOTE: When using custom columns for batching, they should include at least one unique column (e.g. primary key) as a tiebreaker. Also, to reduce the likelihood of race conditions, all columns should be static (unchangeable after it was set).

NOTE: By its nature, batch processing is subject to race conditions if other processes are modifying the database.



161
162
163
164
165
166
167
168
169
170
171
172
173
174
 
# File 'lib/active_record/relation/batches.rb', line 161

def find_in_batches(start: nil, finish: nil, batch_size: 1000, error_on_ignore: nil, cursor: primary_key, order: DEFAULT_ORDER)
  relation = self
  unless block_given?
    return to_enum(:find_in_batches, start: start, finish: finish, batch_size: batch_size, error_on_ignore: error_on_ignore, cursor: cursor, order: order) do
      cursor = Array(cursor)
      total = apply_limits(relation, cursor, start, finish, build_batch_orders(cursor, order)).size
      (total - 1).div(batch_size) + 1
    end
  end

  in_batches(of: batch_size, start: start, finish: finish, load: true, error_on_ignore: error_on_ignore, cursor: cursor, order: order) do |batch|
    yield batch.to_a
  end
end

#in_batches(of: 1000, start: nil, finish: nil, load: false, error_on_ignore: nil, cursor: primary_key, order: DEFAULT_ORDER, use_ranges: nil, &block) ⇒ Object

Yields ActiveRecord::Relation objects to work with a batch of records.

Person.where("age > 21").in_batches do |relation|
  relation.delete_all
  sleep(10) # Throttle the delete queries
end

If you do not provide a block to #in_batches, it will return a BatchEnumerator which is enumerable.

Person.in_batches.each_with_index do |relation, batch_index|
  puts "Processing relation ##{batch_index}"
  relation.delete_all
end

Examples of calling methods on the returned BatchEnumerator object:

Person.in_batches.delete_all
Person.in_batches.update_all(awesome: true)
Person.in_batches.each_record(&:party_all_night!)

Options

  • :of - Specifies the size of the batch. Defaults to 1000.

  • :load - Specifies if the relation should be loaded. Defaults to false.

  • :start - Specifies the cursor column value to start from, inclusive of the value.

  • :finish - Specifies the cursor column value to end at, inclusive of the value.

  • :error_on_ignore - Overrides the application config to specify if an error should be raised when an order is present in the relation.

  • :cursor - Specifies the column to use for batching (can be a column name or an array of column names). Defaults to primary key.

  • :order - Specifies the cursor column order (can be :asc or :desc or an array consisting of :asc or :desc). Defaults to :asc.

    class Order < ActiveRecord::Base
  self.primary_key = [:id_1, :id_2]
end

Order.in_batches(order: [:asc, :desc])

    In the above code, id_1 is sorted in ascending order and id_2 in descending order.

  • :use_ranges - Specifies whether to use range iteration (id >= x AND id <= y). It can make iterating over the whole or almost whole tables several times faster. Only whole table iterations use this style of iteration by default. You can disable this behavior by passing false. If you iterate over the table and the only condition is, e.g., archived_at: nil (and only a tiny fraction of the records are archived), it makes sense to opt in to this approach.

Limits are honored, and if present there is no requirement for the batch size, it can be less than, equal, or greater than the limit.

The options start and finish are especially useful if you want multiple workers dealing with the same processing queue. You can make worker 1 handle all the records between id 1 and 9999 and worker 2 handle from 10000 and beyond by setting the :start and :finish option on each worker.

# Let's process from record 10_000 on.
Person.in_batches(start: 10_000).update_all(awesome: true)

An example of calling where query method on the relation:

Person.in_batches.each do |relation|
  relation.update_all('age = age + 1')
  relation.where('age > 21').update_all(should_party: true)
  relation.where('age <= 21').delete_all
end

NOTE: If you are going to iterate through each record, you should call #each_record on the yielded BatchEnumerator:

Person.in_batches.each_record(&:party_all_night!)

NOTE: Order can be ascending (:asc) or descending (:desc). It is automatically set to ascending on the primary key (“id ASC”). This also means that this method only works when the cursor column is orderable (e.g. an integer or string).

NOTE: When using custom columns for batching, they should include at least one unique column (e.g. primary key) as a tiebreaker. Also, to reduce the likelihood of race conditions, all columns should be static (unchangeable after it was set).

NOTE: By its nature, batch processing is subject to race conditions if other processes are modifying the database.



259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
 
# File 'lib/active_record/relation/batches.rb', line 259

def in_batches(of: 1000, start: nil, finish: nil, load: false, error_on_ignore: nil, cursor: primary_key, order: DEFAULT_ORDER, use_ranges: nil, &block)
  cursor = Array(cursor).map(&:to_s)
  ensure_valid_options_for_batching!(cursor, start, finish, order)

  if arel.orders.present?
    act_on_ignored_order(error_on_ignore)
  end

  unless block
    return BatchEnumerator.new(of: of, start: start, finish: finish, relation: self, cursor: cursor, order: order, use_ranges: use_ranges)
  end

  batch_limit = of

  if limit_value
    remaining   = limit_value
    batch_limit = remaining if remaining < batch_limit
  end

  if self.loaded?
    batch_on_loaded_relation(
      relation: self,
      start: start,
      finish: finish,
      cursor: cursor,
      order: order,
      batch_limit: batch_limit,
      &block
    )
  else
    batch_on_unloaded_relation(
      relation: self,
      start: start,
      finish: finish,
      load: load,
      cursor: cursor,
      order: order,
      use_ranges: use_ranges,
      remaining: remaining,
      batch_limit: batch_limit,
      &block
    )
  end
end