fedi/mastodon
1
0
Fork 1
mirror of https://github.com/mastodon/mastodon.git synced 2024-11-25 07:05:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

Bump chewy from 5.2.0 to 7.2.3 (supports Elasticsearch 7.x) (#16915)

Browse source 
* Bump chewy from 5.2.0 to 7.2.2

* fix style (codeclimate)

* fix style

* fix style

* Bump chewy from 7.2.2 to 7.2.3
This commit is contained in:
Takeshi Umeda 2021-11-19 06:02:08 +09:00 committed by GitHub
parent 2b6a25c609
commit 3419d3ec84
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
15 changed files with 136 additions and 145 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
Gemfile
View file

@ -30,7 +30,7 @@ gem 'bootsnap', '~> 1.9.1', require: false
gem 'browser'
gem 'charlock_holmes', '~> 0.7.7'
gem 'iso-639'
gem 'chewy', '~> 5.2'
gem 'chewy', '~> 7.2'
gem 'cld3', '~> 3.4.2'
gem 'devise', '~> 4.8'
gem 'devise-two-factor', '~> 4.0'

38
Gemfile.lock
View file

@ -147,9 +147,9 @@ GEM
activesupport
cbor (0.5.9.6)
charlock_holmes (0.7.7)
chewy (5.2.0)
chewy (7.2.3)
activesupport (>= 5.2)
elasticsearch (>= 2.0.0)
elasticsearch (>= 7.12.0, < 7.14.0)
elasticsearch-dsl
chunky_png (1.4.0)
cld3 (3.4.2)
@ -197,13 +197,13 @@ GEM
railties (>= 3.2)
e2mmap (0.1.0)
ed25519 (1.2.4)
elasticsearch (7.10.1)
elasticsearch-api (= 7.10.1)
elasticsearch-transport (= 7.10.1)
elasticsearch-api (7.10.1)
elasticsearch (7.13.3)
elasticsearch-api (= 7.13.3)
elasticsearch-transport (= 7.13.3)
elasticsearch-api (7.13.3)
multi_json
elasticsearch-dsl (0.1.9)
elasticsearch-transport (7.10.1)
elasticsearch-dsl (0.1.10)
elasticsearch-transport (7.13.3)
faraday (~> 1)
multi_json
encryptor (3.0.0)
@ -214,11 +214,25 @@ GEM
fabrication (2.22.0)
faker (2.19.0)
i18n (>= 1.6, < 2)
faraday (1.3.0)
faraday (1.8.0)
faraday-em_http (~> 1.0)
faraday-em_synchrony (~> 1.0)
faraday-excon (~> 1.1)
faraday-httpclient (~> 1.0.1)
faraday-net_http (~> 1.0)
faraday-net_http_persistent (~> 1.1)
faraday-patron (~> 1.0)
faraday-rack (~> 1.0)
multipart-post (>= 1.2, < 3)
ruby2_keywords
ruby2_keywords (>= 0.0.4)
faraday-em_http (1.0.0)
faraday-em_synchrony (1.0.0)
faraday-excon (1.1.0)
faraday-httpclient (1.0.1)
faraday-net_http (1.0.1)
faraday-net_http_persistent (1.2.0)
faraday-patron (1.0.0)
faraday-rack (1.0.0)
fast_blank (1.0.1)
fastimage (2.2.5)
ffi (1.15.4)
@ -539,7 +553,7 @@ GEM
ruby-saml (1.13.0)
nokogiri (>= 1.10.5)
rexml
ruby2_keywords (0.0.4)
ruby2_keywords (0.0.5)
rufus-scheduler (3.7.0)
fugit (~> 1.1, >= 1.1.6)
safety_net_attestation (0.4.0)
@ -686,7 +700,7 @@ DEPENDENCIES
capistrano-yarn (~> 2.0)
capybara (~> 3.36)
charlock_holmes (~> 0.7.7)
chewy (~> 5.2)
chewy (~> 7.2)
cld3 (~> 3.4.2)
climate_control (~> 0.2)
color_diff (~> 0.1)

26
app/chewy/accounts_index.rb
View file

@ -23,21 +23,21 @@ class AccountsIndex < Chewy::Index
},
}
define_type ::Account.searchable.includes(:account_stat), delete_if: ->(account) { account.destroyed? || !account.searchable? } do
root date_detection: false do
field :id, type: 'long'
index_scope ::Account.searchable.includes(:account_stat), delete_if: ->(account) { account.destroyed? || !account.searchable? }
field :display_name, type: 'text', analyzer: 'content' do
field :edge_ngram, type: 'text', analyzer: 'edge_ngram', search_analyzer: 'content'
end
root date_detection: false do
field :id, type: 'long'
field :acct, type: 'text', analyzer: 'content', value: ->(account) { [account.username, account.domain].compact.join('@') } do
field :edge_ngram, type: 'text', analyzer: 'edge_ngram', search_analyzer: 'content'
end
field :following_count, type: 'long', value: ->(account) { account.following.local.count }
field :followers_count, type: 'long', value: ->(account) { account.followers.local.count }
field :last_status_at, type: 'date', value: ->(account) { account.last_status_at || account.created_at }
field :display_name, type: 'text', analyzer: 'content' do
field :edge_ngram, type: 'text', analyzer: 'edge_ngram', search_analyzer: 'content'
end
field :acct, type: 'text', analyzer: 'content', value: ->(account) { [account.username, account.domain].compact.join('@') } do
field :edge_ngram, type: 'text', analyzer: 'edge_ngram', search_analyzer: 'content'
end
field :following_count, type: 'long', value: ->(account) { account.following.local.count }
field :followers_count, type: 'long', value: ->(account) { account.followers.local.count }
field :last_status_at, type: 'date', value: ->(account) { account.last_status_at || account.created_at }
end
end

58
app/chewy/statuses_index.rb
View file

@ -31,36 +31,36 @@ class StatusesIndex < Chewy::Index
},
}
define_type ::Status.unscoped.kept.without_reblogs.includes(:media_attachments, :preloadable_poll) do
crutch :mentions do |collection|
data = ::Mention.where(status_id: collection.map(&:id)).where(account: Account.local, silent: false).pluck(:status_id, :account_id)
data.each.with_object({}) { |(id, name), result| (result[id] ||= []).push(name) }
index_scope ::Status.unscoped.kept.without_reblogs.includes(:media_attachments, :preloadable_poll)
crutch :mentions do |collection|
data = ::Mention.where(status_id: collection.map(&:id)).where(account: Account.local, silent: false).pluck(:status_id, :account_id)
data.each.with_object({}) { |(id, name), result| (result[id] ||= []).push(name) }
end
crutch :favourites do |collection|
data = ::Favourite.where(status_id: collection.map(&:id)).where(account: Account.local).pluck(:status_id, :account_id)
data.each.with_object({}) { |(id, name), result| (result[id] ||= []).push(name) }
end
crutch :reblogs do |collection|
data = ::Status.where(reblog_of_id: collection.map(&:id)).where(account: Account.local).pluck(:reblog_of_id, :account_id)
data.each.with_object({}) { |(id, name), result| (result[id] ||= []).push(name) }
end
crutch :bookmarks do |collection|
data = ::Bookmark.where(status_id: collection.map(&:id)).where(account: Account.local).pluck(:status_id, :account_id)
data.each.with_object({}) { |(id, name), result| (result[id] ||= []).push(name) }
end
root date_detection: false do
field :id, type: 'long'
field :account_id, type: 'long'
field :text, type: 'text', value: ->(status) { [status.spoiler_text, Formatter.instance.plaintext(status)].concat(status.media_attachments.map(&:description)).concat(status.preloadable_poll ? status.preloadable_poll.options : []).join("\n\n") } do
field :stemmed, type: 'text', analyzer: 'content'
end
crutch :favourites do |collection|
data = ::Favourite.where(status_id: collection.map(&:id)).where(account: Account.local).pluck(:status_id, :account_id)
data.each.with_object({}) { |(id, name), result| (result[id] ||= []).push(name) }
end
crutch :reblogs do |collection|
data = ::Status.where(reblog_of_id: collection.map(&:id)).where(account: Account.local).pluck(:reblog_of_id, :account_id)
data.each.with_object({}) { |(id, name), result| (result[id] ||= []).push(name) }
end
crutch :bookmarks do |collection|
data = ::Bookmark.where(status_id: collection.map(&:id)).where(account: Account.local).pluck(:status_id, :account_id)
data.each.with_object({}) { |(id, name), result| (result[id] ||= []).push(name) }
end
root date_detection: false do
field :id, type: 'long'
field :account_id, type: 'long'
field :text, type: 'text', value: ->(status) { [status.spoiler_text, Formatter.instance.plaintext(status)].concat(status.media_attachments.map(&:description)).concat(status.preloadable_poll ? status.preloadable_poll.options : []).join("\n\n") } do
field :stemmed, type: 'text', analyzer: 'content'
end
field :searchable_by, type: 'long', value: ->(status, crutches) { status.searchable_by(crutches) }
end
field :searchable_by, type: 'long', value: ->(status, crutches) { status.searchable_by(crutches) }
end
end

16
app/chewy/tags_index.rb
View file

@ -23,15 +23,15 @@ class TagsIndex < Chewy::Index
},
}
define_type ::Tag.listable, delete_if: ->(tag) { tag.destroyed? || !tag.listable? } do
root date_detection: false do
field :name, type: 'text', analyzer: 'content' do
field :edge_ngram, type: 'text', analyzer: 'edge_ngram', search_analyzer: 'content'
end
index_scope ::Tag.listable, delete_if: ->(tag) { tag.destroyed? || !tag.listable? }
field :reviewed, type: 'boolean', value: ->(tag) { tag.reviewed? }
field :usage, type: 'long', value: ->(tag) { tag.history.reduce(0) { |total, day| total + day[:accounts].to_i } }
field :last_status_at, type: 'date', value: ->(tag) { tag.last_status_at || tag.created_at }
root date_detection: false do
field :name, type: 'text', analyzer: 'content' do
field :edge_ngram, type: 'text', analyzer: 'edge_ngram', search_analyzer: 'content'
end
field :reviewed, type: 'boolean', value: ->(tag) { tag.reviewed? }
field :usage, type: 'long', value: ->(tag) { tag.history.reduce(0) { |total, day| total + day[:accounts].to_i } }
field :last_status_at, type: 'date', value: ->(tag) { tag.last_status_at || tag.created_at }
end
end

2
app/models/account.rb
View file

@ -143,7 +143,7 @@ class Account < ApplicationRecord
delegate :chosen_languages, to: :user, prefix: false, allow_nil: true
update_index('accounts#account', :self)
update_index('accounts', :self)
def local?
domain.nil?

2
app/models/account_stat.rb
View file

@ -19,5 +19,5 @@ class AccountStat < ApplicationRecord
belongs_to :account, inverse_of: :account_stat
update_index('accounts#account', :account)
update_index('accounts', :account)
end

2
app/models/bookmark.rb
View file

@ -13,7 +13,7 @@
class Bookmark < ApplicationRecord
include Paginable
update_index('statuses#status', :status) if Chewy.enabled?
update_index('statuses', :status) if Chewy.enabled?
belongs_to :account, inverse_of: :bookmarks
belongs_to :status, inverse_of: :bookmarks

2
app/models/favourite.rb
View file

@ -13,7 +13,7 @@
class Favourite < ApplicationRecord
include Paginable
update_index('statuses#status', :status)
update_index('statuses', :status)
belongs_to :account, inverse_of: :favourites
belongs_to :status, inverse_of: :favourites

2
app/models/status.rb
View file

@ -42,7 +42,7 @@ class Status < ApplicationRecord
# will be based on current time instead of `created_at`
attr_accessor :override_timestamps
update_index('statuses#status', :proper)
update_index('statuses', :proper)
enum visibility: [:public, :unlisted, :private, :direct, :limited], _suffix: :visibility

2
app/models/tag.rb
View file

@ -39,7 +39,7 @@ class Tag < ApplicationRecord
scope :recently_used, ->(account) { joins(:statuses).where(statuses: { id: account.statuses.select(:id).limit(1000) }).group(:id).order(Arel.sql('count(*) desc')) }
scope :matches_name, ->(term) { where(arel_table[:name].lower.matches(arel_table.lower("#{sanitize_sql_like(Tag.normalize(term))}%"), nil, true)) } # Search with case-sensitive to use B-tree index
update_index('tags#tag', :self)
update_index('tags', :self)
def to_param
name

2
app/services/batched_remove_status_service.rb
View file

@ -31,7 +31,7 @@ class BatchedRemoveStatusService < BaseService
# Since we skipped all callbacks, we also need to manually
# deindex the statuses
Chewy.strategy.current.update(StatusesIndex::Status, statuses_and_reblogs) if Chewy.enabled?
Chewy.strategy.current.update(StatusesIndex, statuses_and_reblogs) if Chewy.enabled?
return if options[:skip_side_effects]

4
app/services/delete_account_service.rb
View file

@ -189,7 +189,7 @@ class DeleteAccountService < BaseService
@account.favourites.in_batches do |favourites|
ids = favourites.pluck(:status_id)
StatusStat.where(status_id: ids).update_all('favourites_count = GREATEST(0, favourites_count - 1)')
Chewy.strategy.current.update(StatusesIndex::Status, ids) if Chewy.enabled?
Chewy.strategy.current.update(StatusesIndex, ids) if Chewy.enabled?
Rails.cache.delete_multi(ids.map { |id| "statuses/#{id}" })
favourites.delete_all
end
@ -197,7 +197,7 @@ class DeleteAccountService < BaseService
def purge_bookmarks!
@account.bookmarks.in_batches do |bookmarks|
Chewy.strategy.current.update(StatusesIndex::Status, bookmarks.pluck(:status_id)) if Chewy.enabled?
Chewy.strategy.current.update(StatusesIndex, bookmarks.pluck(:status_id)) if Chewy.enabled?
bookmarks.delete_all
end
end

20
config/initializers/chewy.rb
View file

@ -37,23 +37,3 @@ end
# Mastodon is run with hidden services enabled, because
# ElasticSearch is *not* supposed to be accessed through a proxy
Faraday.ignore_env_proxy = true
# Elasticsearch 7.x workaround
Elasticsearch::Transport::Client.prepend Module.new {
def search(arguments = {})
arguments[:rest_total_hits_as_int] = true
super arguments
end
}
Elasticsearch::API::Indices::IndicesClient.prepend Module.new {
def create(arguments = {})
arguments[:include_type_name] = true
super arguments
end
def put_mapping(arguments = {})
arguments[:include_type_name] = true
super arguments
end
}

103
lib/mastodon/search_cli.rb
View file

@ -64,11 +64,7 @@ module Mastodon
progress.title = 'Estimating workload '
# Estimate the amount of data that has to be imported first
indices.each do |index|
index.types.each do |type|
progress.total = (progress.total || 0) + type.adapter.default_scope.count
end
end
progress.total = indices.sum { |index| index.adapter.default_scope.count }
# Now import all the actual data. Mind that unlike chewy:sync, we don't
# fetch and compare all record IDs from the database and the index to
@ -80,67 +76,68 @@ module Mastodon
batch_size = 1_000
slice_size = (batch_size / options[:concurrency]).ceil
index.types.each do |type|
type.adapter.default_scope.reorder(nil).find_in_batches(batch_size: batch_size) do |batch|
futures = []
index.adapter.default_scope.reorder(nil).find_in_batches(batch_size: batch_size) do |batch|
futures = []
batch.each_slice(slice_size) do |records|
futures << Concurrent::Future.execute(executor: pool) do
begin
if !progress.total.nil? && progress.progress + records.size > progress.total
# The number of items has changed between start and now,
# since there is no good way to predict the final count from
# here, just change the progress bar to an indeterminate one
batch.each_slice(slice_size) do |records|
futures << Concurrent::Future.execute(executor: pool) do
begin
if !progress.total.nil? && progress.progress + records.size > progress.total
# The number of items has changed between start and now,
# since there is no good way to predict the final count from
# here, just change the progress bar to an indeterminate one
progress.total = nil
progress.total = nil
end
grouped_records = nil
bulk_body = nil
index_count = 0
delete_count = 0
ActiveRecord::Base.connection_pool.with_connection do
grouped_records = records.to_a.group_by do |record|
index.adapter.send(:delete_from_index?, record) ? :delete : :to_index
end
grouped_records = nil
bulk_body = nil
index_count = 0
delete_count = 0
bulk_body = Chewy::Index::Import::BulkBuilder.new(index, **grouped_records).bulk_body
end
ActiveRecord::Base.connection_pool.with_connection do
grouped_records = type.adapter.send(:grouped_objects, records)
bulk_body = Chewy::Type::Import::BulkBuilder.new(type, **grouped_records).bulk_body
end
index_count = grouped_records[:to_index].size if grouped_records.key?(:to_index)
delete_count = grouped_records[:delete].size if grouped_records.key?(:delete)
index_count = grouped_records[:index].size if grouped_records.key?(:index)
delete_count = grouped_records[:delete].size if grouped_records.key?(:delete)
# The following is an optimization for statuses specifically, since
# we want to de-index statuses that cannot be searched by anybody,
# but can't use Chewy's delete_if logic because it doesn't use
# crutches and our searchable_by logic depends on them
if index == StatusesIndex
bulk_body.map! do |entry|
if entry[:to_index] && entry.dig(:to_index, :data, 'searchable_by').blank?
index_count -= 1
delete_count += 1
# The following is an optimization for statuses specifically, since
# we want to de-index statuses that cannot be searched by anybody,
# but can't use Chewy's delete_if logic because it doesn't use
# crutches and our searchable_by logic depends on them
if type == StatusesIndex::Status
bulk_body.map! do |entry|
if entry[:index] && entry.dig(:index, :data, 'searchable_by').blank?
index_count -= 1
delete_count += 1
{ delete: entry[:index].except(:data) }
else
entry
end
{ delete: entry[:to_index].except(:data) }
else
entry
end
end
Chewy::Type::Import::BulkRequest.new(type).perform(bulk_body)
progress.progress += records.size
added.increment(index_count)
removed.increment(delete_count)
sleep 1
rescue => e
progress.log pastel.red("Error importing #{index}: #{e}")
end
Chewy::Index::Import::BulkRequest.new(index).perform(bulk_body)
progress.progress += records.size
added.increment(index_count)
removed.increment(delete_count)
sleep 1
rescue => e
progress.log pastel.red("Error importing #{index}: #{e}")
end
end
futures.map(&:value)
end
futures.map(&:value)
end
end