Add search deploy date options

This commit is contained in:
KMY 2023-09-16 10:53:09 +09:00
parent 173b9f4fd7
commit 0ffd8acacd
4 changed files with 12 additions and 3 deletions

View file

@ -3,10 +3,12 @@
class Importer::BaseImporter
# @param [Integer] batch_size
# @param [Concurrent::ThreadPoolExecutor] executor
def initialize(batch_size:, executor:, full: true)
def initialize(batch_size:, executor:, full: true, from: nil, to: nil)
@batch_size = batch_size
@executor = executor
@full = full
@from = from.to_date if from.present?
@to = to.to_date if to.present?
@wait_for = Concurrent::Set.new
end

View file

@ -27,6 +27,9 @@ class Importer::PublicStatusesIndexImporter < Importer::BaseImporter
end
def scope
Status.indexable.reorder(nil)
to_index = Status.indexable.reorder(nil)
to_index = to_index.where('statuses.created_at >= ?', @from) if @from.present?
to_index = to_index.where('statuses.created_at < ?', @to) if @to.present?
to_index
end
end

View file

@ -17,6 +17,8 @@ class Importer::StatusesIndexImporter < Importer::BaseImporter
bulk = ActiveRecord::Base.connection_pool.with_connection do
to_index = index.adapter.default_scope.where(id: status_ids)
to_index = to_index.where('created_at >= ?', @from) if @from.present?
to_index = to_index.where('created_at < ?', @to) if @to.present?
crutches = Chewy::Index::Crutch::Crutches.new index, to_index
to_index.map do |object|
# This is unlikely to happen, but the post may have been

View file

@ -21,6 +21,8 @@ module Mastodon::CLI
option :clean, type: :boolean, default: true, desc: 'Remove outdated documents from the index'
option :reset_chewy, type: :boolean, default: false, desc: "Reset Chewy's internal index"
option :full, type: :boolean, default: false, desc: 'Import full data over Mastodon default importer'
option :from, type: :string, default: nil, desc: 'Statuses start date'
option :to, type: :string, default: nil, desc: 'Statuses end date'
desc 'deploy', 'Create or upgrade Elasticsearch indices and populate them'
long_desc <<~LONG_DESC
If Elasticsearch is empty, this command will create the necessary indices
@ -42,7 +44,7 @@ module Mastodon::CLI
end
pool = Concurrent::FixedThreadPool.new(options[:concurrency], max_queue: options[:concurrency] * 10)
importers = indices.index_with { |index| "Importer::#{index.name}Importer".constantize.new(batch_size: options[:batch_size], executor: pool, full: options[:full]) }
importers = indices.index_with { |index| "Importer::#{index.name}Importer".constantize.new(batch_size: options[:batch_size], executor: pool, full: options[:full], from: options[:from], to: options[:to]) }
progress = ProgressBar.create(total: nil, format: '%t%c/%u |%b%i| %e (%r docs/s)', autofinish: false)
Chewy::Stash::Specification.reset! if options[:reset_chewy]