2019-05-03 06:17:27 +08:00
|
|
|
# frozen_string_literal: true
|
|
|
|
|
2013-05-23 02:36:14 +08:00
|
|
|
class Search
|
2018-10-23 09:10:33 +08:00
|
|
|
DIACRITICS ||= /([\u0300-\u036f]|[\u1AB0-\u1AFF]|[\u1DC0-\u1DFF]|[\u20D0-\u20FF])/
|
2013-02-06 03:16:51 +08:00
|
|
|
|
2019-03-18 01:46:09 +08:00
|
|
|
cattr_accessor :preloaded_topic_custom_fields
|
|
|
|
self.preloaded_topic_custom_fields = Set.new
|
|
|
|
|
2013-02-06 03:16:51 +08:00
|
|
|
def self.per_facet
|
|
|
|
5
|
|
|
|
end
|
|
|
|
|
2018-10-23 09:10:33 +08:00
|
|
|
def self.strip_diacritics(str)
|
|
|
|
s = str.unicode_normalize(:nfkd)
|
|
|
|
s.gsub!(DIACRITICS, "")
|
|
|
|
s.strip!
|
|
|
|
s
|
|
|
|
end
|
|
|
|
|
2014-09-02 17:15:08 +08:00
|
|
|
def self.per_filter
|
|
|
|
50
|
|
|
|
end
|
|
|
|
|
2013-02-06 03:16:51 +08:00
|
|
|
def self.facets
|
2020-01-28 18:11:33 +08:00
|
|
|
%w(topic category user private_messages tags all_topics)
|
2013-02-06 03:16:51 +08:00
|
|
|
end
|
|
|
|
|
2017-08-01 03:28:48 +08:00
|
|
|
def self.ts_config(locale = SiteSetting.default_locale)
|
|
|
|
# if adding a text search configuration, you should check PG beforehand:
|
|
|
|
# SELECT cfgname FROM pg_ts_config;
|
|
|
|
# As an aside, dictionaries can be listed by `\dFd`, the
|
|
|
|
# physical locations are in /usr/share/postgresql/<version>/tsearch_data.
|
|
|
|
# But it may not appear there based on pg extension configuration.
|
2014-06-24 15:10:56 +08:00
|
|
|
# base docker config
|
|
|
|
#
|
2019-05-27 21:52:09 +08:00
|
|
|
case locale.split("_")[0].to_sym
|
|
|
|
when :da then 'danish'
|
|
|
|
when :nl then 'dutch'
|
|
|
|
when :en then 'english'
|
|
|
|
when :fi then 'finnish'
|
|
|
|
when :fr then 'french'
|
|
|
|
when :de then 'german'
|
|
|
|
when :hu then 'hungarian'
|
|
|
|
when :it then 'italian'
|
|
|
|
when :nb then 'norwegian'
|
|
|
|
when :pt then 'portuguese'
|
|
|
|
when :ro then 'romanian'
|
|
|
|
when :ru then 'russian'
|
|
|
|
when :es then 'spanish'
|
|
|
|
when :sv then 'swedish'
|
|
|
|
when :tr then 'turkish'
|
2017-08-01 03:28:48 +08:00
|
|
|
else 'simple' # use the 'simple' stemmer for other languages
|
2013-03-01 03:14:22 +08:00
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2017-08-01 03:28:48 +08:00
|
|
|
def self.prepare_data(search_data, purpose = :query)
|
2018-11-07 06:41:55 +08:00
|
|
|
purpose ||= :query
|
2014-06-24 15:10:56 +08:00
|
|
|
|
2018-11-07 06:41:55 +08:00
|
|
|
data = search_data.dup
|
2014-06-24 15:10:56 +08:00
|
|
|
data.force_encoding("UTF-8")
|
2018-11-07 06:41:55 +08:00
|
|
|
if purpose != :topic
|
|
|
|
# TODO cppjieba_rb is designed for chinese, we need something else for Japanese
|
|
|
|
# Korean appears to be safe cause words are already space seperated
|
|
|
|
# For Japanese we should investigate using kakasi
|
|
|
|
if ['zh_TW', 'zh_CN', 'ja'].include?(SiteSetting.default_locale) || SiteSetting.search_tokenize_chinese_japanese_korean
|
|
|
|
require 'cppjieba_rb' unless defined? CppjiebaRb
|
|
|
|
mode = (purpose == :query ? :query : :mix)
|
|
|
|
data = CppjiebaRb.segment(search_data, mode: mode)
|
2020-05-18 08:54:56 +08:00
|
|
|
|
|
|
|
# TODO: we still want to tokenize here but the current stopword list is too wide
|
|
|
|
# in cppjieba leading to words such as volume to be skipped. PG already has an English
|
|
|
|
# stopword list so use that vs relying on cppjieba
|
|
|
|
if ts_config != 'english'
|
|
|
|
data = CppjiebaRb.filter_stop_word(data)
|
|
|
|
else
|
|
|
|
data = data.filter { |s| s.present? }
|
|
|
|
end
|
|
|
|
|
|
|
|
data = data.join(' ')
|
|
|
|
|
2018-11-07 06:41:55 +08:00
|
|
|
else
|
|
|
|
data.squish!
|
|
|
|
end
|
|
|
|
|
|
|
|
if SiteSetting.search_ignore_accents
|
|
|
|
data = strip_diacritics(data)
|
|
|
|
end
|
2018-10-23 09:10:33 +08:00
|
|
|
end
|
2020-07-09 17:02:02 +08:00
|
|
|
|
|
|
|
data.gsub!(EmailCook.url_regexp) do |url|
|
|
|
|
uri = URI.parse(url)
|
|
|
|
uri.query = nil
|
|
|
|
uri.to_s
|
|
|
|
rescue URI::Error
|
|
|
|
# Don't fail even if URL turns out to be invalid
|
|
|
|
end
|
|
|
|
|
2014-06-24 15:10:56 +08:00
|
|
|
data
|
|
|
|
end
|
|
|
|
|
2016-03-14 20:27:02 +08:00
|
|
|
def self.word_to_date(str)
|
|
|
|
|
|
|
|
if str =~ /^[0-9]{1,3}$/
|
|
|
|
return Time.zone.now.beginning_of_day.days_ago(str.to_i)
|
|
|
|
end
|
|
|
|
|
|
|
|
if str =~ /^([12][0-9]{3})(-([0-1]?[0-9]))?(-([0-3]?[0-9]))?$/
|
|
|
|
year = $1.to_i
|
|
|
|
month = $2 ? $3.to_i : 1
|
|
|
|
day = $4 ? $5.to_i : 1
|
|
|
|
|
|
|
|
return if day == 0 || month == 0 || day > 31 || month > 12
|
|
|
|
|
2018-05-07 11:38:04 +08:00
|
|
|
return begin
|
|
|
|
Time.zone.parse("#{year}-#{month}-#{day}")
|
|
|
|
rescue ArgumentError
|
|
|
|
end
|
2016-03-14 20:27:02 +08:00
|
|
|
end
|
|
|
|
|
|
|
|
if str.downcase == "yesterday"
|
|
|
|
return Time.zone.now.beginning_of_day.yesterday
|
|
|
|
end
|
|
|
|
|
|
|
|
titlecase = str.downcase.titlecase
|
|
|
|
|
|
|
|
if Date::DAYNAMES.include?(titlecase)
|
|
|
|
return Time.zone.now.beginning_of_week(str.downcase.to_sym)
|
|
|
|
end
|
|
|
|
|
|
|
|
if idx = (Date::MONTHNAMES.find_index(titlecase) ||
|
|
|
|
Date::ABBR_MONTHNAMES.find_index(titlecase))
|
|
|
|
delta = Time.zone.now.month - idx
|
|
|
|
delta += 12 if delta < 0
|
|
|
|
Time.zone.now.beginning_of_month.months_ago(delta)
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2016-08-11 03:40:58 +08:00
|
|
|
def self.min_post_id_no_cache
|
|
|
|
return 0 unless SiteSetting.search_prefer_recent_posts?
|
|
|
|
|
2016-08-12 12:36:38 +08:00
|
|
|
offset, has_more = Post.unscoped
|
|
|
|
.order('id desc')
|
|
|
|
.offset(SiteSetting.search_recent_posts_size - 1)
|
|
|
|
.limit(2)
|
|
|
|
.pluck(:id)
|
|
|
|
|
|
|
|
has_more ? offset : 0
|
2016-08-11 03:40:58 +08:00
|
|
|
end
|
|
|
|
|
|
|
|
def self.min_post_id(opts = nil)
|
|
|
|
return 0 unless SiteSetting.search_prefer_recent_posts?
|
|
|
|
|
|
|
|
# It can be quite slow to count all the posts so let's cache it
|
2019-11-27 09:35:14 +08:00
|
|
|
Discourse.cache.fetch("search-min-post-id:#{SiteSetting.search_recent_posts_size}", expires_in: 1.week) do
|
2016-08-11 03:40:58 +08:00
|
|
|
min_post_id_no_cache
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2016-08-10 02:48:39 +08:00
|
|
|
attr_accessor :term
|
2017-06-22 03:51:15 +08:00
|
|
|
attr_reader :clean_term
|
2016-08-10 02:48:39 +08:00
|
|
|
|
2013-05-23 02:36:14 +08:00
|
|
|
def initialize(term, opts = nil)
|
|
|
|
@opts = opts || {}
|
|
|
|
@guardian = @opts[:guardian] || Guardian.new
|
2013-05-25 02:03:45 +08:00
|
|
|
@search_context = @opts[:search_context]
|
2015-07-27 14:13:11 +08:00
|
|
|
@blurb_length = @opts[:blurb_length]
|
2016-08-10 02:48:39 +08:00
|
|
|
@valid = true
|
2017-07-21 00:12:34 +08:00
|
|
|
@page = @opts[:page]
|
2014-12-05 07:22:39 +08:00
|
|
|
|
2018-08-24 08:00:51 +08:00
|
|
|
term = term.to_s.dup
|
|
|
|
|
2017-06-07 14:23:48 +08:00
|
|
|
# Removes any zero-width characters from search terms
|
2018-08-24 08:00:51 +08:00
|
|
|
term.gsub!(/[\u200B-\u200D\uFEFF]/, '')
|
2017-12-12 13:47:28 +08:00
|
|
|
# Replace curly quotes to regular quotes
|
2018-08-24 08:00:51 +08:00
|
|
|
term.gsub!(/[\u201c\u201d]/, '"')
|
|
|
|
|
|
|
|
@clean_term = term
|
2019-02-05 17:54:52 +08:00
|
|
|
@in_title = false
|
2017-06-22 03:51:15 +08:00
|
|
|
|
2015-06-26 15:36:28 +08:00
|
|
|
term = process_advanced_search!(term)
|
2015-09-15 15:21:46 +08:00
|
|
|
|
2015-06-26 15:36:28 +08:00
|
|
|
if term.present?
|
2018-11-07 06:41:55 +08:00
|
|
|
@term = Search.prepare_data(term, Topic === @search_context ? :topic : nil)
|
2015-06-26 15:36:28 +08:00
|
|
|
@original_term = PG::Connection.escape_string(@term)
|
|
|
|
end
|
|
|
|
|
2014-12-05 07:22:39 +08:00
|
|
|
if @search_pms && @guardian.user
|
|
|
|
@opts[:type_filter] = "private_messages"
|
|
|
|
@search_context = @guardian.user
|
|
|
|
end
|
|
|
|
|
2020-03-06 01:50:29 +08:00
|
|
|
if @search_all_topics && @guardian.user
|
2020-01-28 18:11:33 +08:00
|
|
|
@opts[:type_filter] = "all_topics"
|
|
|
|
end
|
|
|
|
|
2017-07-17 23:57:13 +08:00
|
|
|
@results = GroupedSearchResults.new(
|
2020-07-14 11:05:57 +08:00
|
|
|
type_filter: @opts[:type_filter],
|
|
|
|
term: clean_term,
|
|
|
|
blurb_term: term,
|
|
|
|
search_context: @search_context,
|
|
|
|
blurb_length: @blurb_length
|
2017-07-17 23:57:13 +08:00
|
|
|
)
|
2014-09-02 17:15:08 +08:00
|
|
|
end
|
|
|
|
|
2017-07-21 00:12:34 +08:00
|
|
|
def limit
|
|
|
|
if @opts[:type_filter].present?
|
|
|
|
Search.per_filter + 1
|
|
|
|
else
|
|
|
|
Search.per_facet + 1
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
def offset
|
|
|
|
if @page && @opts[:type_filter].present?
|
|
|
|
(@page - 1) * Search.per_filter
|
|
|
|
else
|
|
|
|
0
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2016-08-10 02:48:39 +08:00
|
|
|
def valid?
|
|
|
|
@valid
|
|
|
|
end
|
|
|
|
|
2014-09-02 17:15:08 +08:00
|
|
|
def self.execute(term, opts = nil)
|
|
|
|
self.new(term, opts).execute
|
2013-05-14 05:04:41 +08:00
|
|
|
end
|
|
|
|
|
|
|
|
# Query a term
|
2020-08-04 14:01:43 +08:00
|
|
|
def execute(readonly_mode: Discourse.readonly_mode?)
|
|
|
|
if SiteSetting.log_search_queries? && @opts[:search_type].present? && !readonly_mode
|
2017-07-17 23:57:13 +08:00
|
|
|
status, search_log_id = SearchLog.log(
|
2017-07-14 01:34:31 +08:00
|
|
|
term: @term,
|
|
|
|
search_type: @opts[:search_type],
|
|
|
|
ip_address: @opts[:ip_address],
|
|
|
|
user_id: @opts[:user_id]
|
|
|
|
)
|
2017-07-17 23:57:13 +08:00
|
|
|
@results.search_log_id = search_log_id unless status == :error
|
2017-07-14 01:34:31 +08:00
|
|
|
end
|
2016-08-10 02:48:39 +08:00
|
|
|
|
2017-07-26 08:51:44 +08:00
|
|
|
unless @filters.present? || @opts[:search_for_id]
|
2016-08-10 02:48:39 +08:00
|
|
|
min_length = @opts[:min_search_term_length] || SiteSetting.min_search_term_length
|
|
|
|
terms = (@term || '').split(/\s(?=(?:[^"]|"[^"]*")*$)/).reject { |t| t.length < min_length }
|
|
|
|
|
|
|
|
if terms.blank?
|
|
|
|
@term = ''
|
|
|
|
@valid = false
|
|
|
|
return
|
|
|
|
end
|
2015-06-23 10:14:06 +08:00
|
|
|
end
|
2013-05-14 05:04:41 +08:00
|
|
|
|
|
|
|
# If the term is a number or url to a topic, just include that topic
|
2020-01-28 18:11:33 +08:00
|
|
|
if @opts[:search_for_id] && ['topic', 'private_messages', 'all_topics'].include?(@results.type_filter)
|
2014-09-02 17:15:08 +08:00
|
|
|
if @term =~ /^\d+$/
|
|
|
|
single_topic(@term.to_i)
|
|
|
|
else
|
|
|
|
begin
|
2020-06-29 18:31:20 +08:00
|
|
|
if route = Discourse.route_for(@term)
|
|
|
|
if route[:controller] == "topics" && route[:action] == "show"
|
|
|
|
topic_id = (route[:id] || route[:topic_id]).to_i
|
|
|
|
single_topic(topic_id) if topic_id > 0
|
|
|
|
end
|
|
|
|
end
|
2014-09-02 17:15:08 +08:00
|
|
|
rescue ActionController::RoutingError
|
|
|
|
end
|
2013-05-14 05:04:41 +08:00
|
|
|
end
|
|
|
|
end
|
2013-03-02 01:45:25 +08:00
|
|
|
|
2019-03-18 15:25:45 +08:00
|
|
|
find_grouped_results if @results.posts.blank?
|
2014-09-02 17:15:08 +08:00
|
|
|
|
2019-03-18 01:46:09 +08:00
|
|
|
if preloaded_topic_custom_fields.present? && @results.posts.present?
|
|
|
|
topics = @results.posts.map(&:topic)
|
|
|
|
Topic.preload_custom_fields(topics, preloaded_topic_custom_fields)
|
|
|
|
end
|
|
|
|
|
2014-09-02 17:15:08 +08:00
|
|
|
@results
|
2013-05-14 05:04:41 +08:00
|
|
|
end
|
2013-02-06 03:16:51 +08:00
|
|
|
|
2015-06-23 10:14:06 +08:00
|
|
|
def self.advanced_filter(trigger, &block)
|
|
|
|
(@advanced_filters ||= {})[trigger] = block
|
|
|
|
end
|
|
|
|
|
|
|
|
def self.advanced_filters
|
|
|
|
@advanced_filters
|
|
|
|
end
|
|
|
|
|
2019-07-22 22:55:49 +08:00
|
|
|
advanced_filter(/^in:personal-direct$/) do |posts|
|
|
|
|
if @guardian.user
|
|
|
|
posts
|
|
|
|
.joins("LEFT JOIN topic_allowed_groups tg ON posts.topic_id = tg.topic_id")
|
|
|
|
.where(<<~SQL, user_id: @guardian.user.id)
|
|
|
|
tg.id IS NULL
|
|
|
|
AND posts.topic_id IN (
|
|
|
|
SELECT tau.topic_id
|
|
|
|
FROM topic_allowed_users tau
|
|
|
|
JOIN topic_allowed_users tau2
|
|
|
|
ON tau2.topic_id = tau.topic_id
|
|
|
|
AND tau2.id != tau.id
|
|
|
|
WHERE tau.user_id = :user_id
|
|
|
|
GROUP BY tau.topic_id
|
|
|
|
HAVING COUNT(*) = 1
|
|
|
|
)
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2019-06-28 16:19:57 +08:00
|
|
|
advanced_filter(/^in:tagged$/) do |posts|
|
|
|
|
posts
|
|
|
|
.where('EXISTS (SELECT 1 FROM topic_tags WHERE topic_tags.topic_id = posts.topic_id)')
|
|
|
|
end
|
|
|
|
|
|
|
|
advanced_filter(/^in:untagged$/) do |posts|
|
|
|
|
posts
|
|
|
|
.joins("LEFT JOIN topic_tags ON
|
|
|
|
topic_tags.topic_id = posts.topic_id")
|
|
|
|
.where("topic_tags.id IS NULL")
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^status:open$/) do |posts|
|
2015-06-23 10:14:06 +08:00
|
|
|
posts.where('NOT topics.closed AND NOT topics.archived')
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^status:closed$/) do |posts|
|
2015-06-23 10:14:06 +08:00
|
|
|
posts.where('topics.closed')
|
|
|
|
end
|
|
|
|
|
2019-12-17 00:41:34 +08:00
|
|
|
advanced_filter(/^status:public$/) do |posts|
|
|
|
|
category_ids = Category
|
|
|
|
.where(read_restricted: false)
|
|
|
|
.pluck(:id)
|
|
|
|
|
|
|
|
posts.where("topics.category_id in (?)", category_ids)
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^status:archived$/) do |posts|
|
2015-06-23 10:14:06 +08:00
|
|
|
posts.where('topics.archived')
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^status:noreplies$/) do |posts|
|
2015-06-23 10:14:06 +08:00
|
|
|
posts.where("topics.posts_count = 1")
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^status:single_user$/) do |posts|
|
2015-06-23 10:14:06 +08:00
|
|
|
posts.where("topics.participant_count = 1")
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^posts_count:(\d+)$/) do |posts, match|
|
2015-06-23 10:14:06 +08:00
|
|
|
posts.where("topics.posts_count = ?", match.to_i)
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^min_post_count:(\d+)$/) do |posts, match|
|
2016-11-28 22:57:18 +08:00
|
|
|
posts.where("topics.posts_count >= ?", match.to_i)
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^in:first|^f$/) do |posts|
|
2015-06-23 11:39:40 +08:00
|
|
|
posts.where("posts.post_number = 1")
|
|
|
|
end
|
2015-06-23 10:14:06 +08:00
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^in:pinned$/) do |posts|
|
2016-03-18 13:26:54 +08:00
|
|
|
posts.where("topics.pinned_at IS NOT NULL")
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^in:wiki$/) do |posts, match|
|
2016-05-13 16:26:53 +08:00
|
|
|
posts.where(wiki: true)
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^badge:(.*)$/) do |posts, match|
|
2019-10-21 18:32:27 +08:00
|
|
|
badge_id = Badge.where('name ilike ? OR id = ?', match, match.to_i).pluck_first(:id)
|
2015-09-15 15:21:46 +08:00
|
|
|
if badge_id
|
|
|
|
posts.where('posts.user_id IN (SELECT ub.user_id FROM user_badges ub WHERE ub.badge_id = ?)', badge_id)
|
|
|
|
else
|
|
|
|
posts.where("1 = 0")
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2020-04-16 09:32:21 +08:00
|
|
|
def post_action_type_filter(posts, post_action_type)
|
|
|
|
posts.where("posts.id IN (
|
|
|
|
SELECT pa.post_id FROM post_actions pa
|
|
|
|
WHERE pa.user_id = #{@guardian.user.id} AND
|
|
|
|
pa.post_action_type_id = #{post_action_type} AND
|
|
|
|
deleted_at IS NULL
|
|
|
|
)")
|
|
|
|
end
|
|
|
|
|
|
|
|
advanced_filter(/^in:(likes)$/) do |posts, match|
|
|
|
|
if @guardian.user
|
|
|
|
post_action_type_filter(posts, PostActionType.types[:like])
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
advanced_filter(/^in:(bookmarks)$/) do |posts, match|
|
2015-06-23 10:14:06 +08:00
|
|
|
if @guardian.user
|
2020-04-22 11:44:19 +08:00
|
|
|
posts.where("posts.id IN (SELECT post_id FROM bookmarks WHERE bookmarks.user_id = #{@guardian.user.id})")
|
2015-06-23 10:14:06 +08:00
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^in:posted$/) do |posts|
|
2015-06-23 10:14:06 +08:00
|
|
|
posts.where("posts.user_id = #{@guardian.user.id}") if @guardian.user
|
|
|
|
end
|
|
|
|
|
2019-12-17 00:41:34 +08:00
|
|
|
advanced_filter(/^in:created$/) do |posts|
|
|
|
|
posts.where(user_id: @guardian.user.id, post_number: 1) if @guardian.user
|
|
|
|
end
|
|
|
|
|
2020-01-23 04:16:17 +08:00
|
|
|
advanced_filter(/^created:@(.*)$/) do |posts, match|
|
|
|
|
user_id = User.where(username: match.downcase).pluck_first(:id)
|
|
|
|
posts.where(user_id: user_id, post_number: 1)
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^in:(watching|tracking)$/) do |posts, match|
|
2015-06-23 10:14:06 +08:00
|
|
|
if @guardian.user
|
|
|
|
level = TopicUser.notification_levels[match.to_sym]
|
|
|
|
posts.where("posts.topic_id IN (
|
|
|
|
SELECT tu.topic_id FROM topic_users tu
|
2017-03-08 20:37:29 +08:00
|
|
|
WHERE tu.user_id = :user_id AND
|
|
|
|
tu.notification_level >= :level
|
|
|
|
)", user_id: @guardian.user.id, level: level)
|
2015-06-23 10:14:06 +08:00
|
|
|
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^in:seen$/) do |posts|
|
2017-03-08 22:46:23 +08:00
|
|
|
if @guardian.user
|
|
|
|
posts
|
|
|
|
.joins("INNER JOIN post_timings ON
|
|
|
|
post_timings.topic_id = posts.topic_id
|
|
|
|
AND post_timings.post_number = posts.post_number
|
2017-08-31 12:06:56 +08:00
|
|
|
AND post_timings.user_id = #{ActiveRecord::Base.connection.quote(@guardian.user.id)}
|
2017-03-08 22:46:23 +08:00
|
|
|
")
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^in:unseen$/) do |posts|
|
2017-03-08 22:46:23 +08:00
|
|
|
if @guardian.user
|
|
|
|
posts
|
|
|
|
.joins("LEFT JOIN post_timings ON
|
|
|
|
post_timings.topic_id = posts.topic_id
|
|
|
|
AND post_timings.post_number = posts.post_number
|
2017-08-31 12:06:56 +08:00
|
|
|
AND post_timings.user_id = #{ActiveRecord::Base.connection.quote(@guardian.user.id)}
|
2017-03-08 22:46:23 +08:00
|
|
|
")
|
|
|
|
.where("post_timings.user_id IS NULL")
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^with:images$/) do |posts|
|
FEATURE: Include optimized thumbnails for topics (#9215)
This introduces new APIs for obtaining optimized thumbnails for topics. There are a few building blocks required for this:
- Introduces new `image_upload_id` columns on the `posts` and `topics` table. This replaces the old `image_url` column, which means that thumbnails are now restricted to uploads. Hotlinked thumbnails are no longer possible. In normal use (with pull_hotlinked_images enabled), this has no noticeable impact
- A migration attempts to match existing urls to upload records. If a match cannot be found then the posts will be queued for rebake
- Optimized thumbnails are generated during post_process_cooked. If thumbnails are missing when serializing a topic list, then a sidekiq job is queued
- Topic lists and topics now include a `thumbnails` key, which includes all the available images:
```
"thumbnails": [
{
"max_width": null,
"max_height": null,
"url": "//example.com/original-image.png",
"width": 1380,
"height": 1840
},
{
"max_width": 1024,
"max_height": 1024,
"url": "//example.com/optimized-image.png",
"width": 768,
"height": 1024
}
]
```
- Themes can request additional thumbnail sizes by using a modifier in their `about.json` file:
```
"modifiers": {
"topic_thumbnail_sizes": [
[200, 200],
[800, 800]
],
...
```
Remember that these are generated asynchronously, so your theme should include logic to fallback to other available thumbnails if your requested size has not yet been generated
- Two new raw plugin outlets are introduced, to improve the customisability of the topic list. `topic-list-before-columns` and `topic-list-before-link`
2020-05-05 16:07:50 +08:00
|
|
|
posts.where("posts.image_upload_id IS NOT NULL")
|
2017-06-08 02:13:36 +08:00
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^category:(.+)$/) do |posts, match|
|
2017-02-08 04:53:37 +08:00
|
|
|
exact = false
|
|
|
|
|
|
|
|
if match[0] == "="
|
|
|
|
exact = true
|
|
|
|
match = match[1..-1]
|
|
|
|
end
|
|
|
|
|
|
|
|
category_ids = Category.where('slug ilike ? OR name ilike ? OR id = ?',
|
|
|
|
match, match, match.to_i).pluck(:id)
|
2016-06-09 01:50:33 +08:00
|
|
|
if category_ids.present?
|
2017-02-08 04:53:37 +08:00
|
|
|
|
|
|
|
unless exact
|
|
|
|
category_ids +=
|
|
|
|
Category.where('parent_category_id = ?', category_ids.first).pluck(:id)
|
|
|
|
end
|
|
|
|
|
2019-03-19 11:23:14 +08:00
|
|
|
@category_filter_matched ||= true
|
2016-06-09 01:50:33 +08:00
|
|
|
posts.where("topics.category_id IN (?)", category_ids)
|
2015-08-14 09:53:16 +08:00
|
|
|
else
|
|
|
|
posts.where("1 = 0")
|
|
|
|
end
|
2015-06-23 10:14:06 +08:00
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^\#([\p{L}0-9\-:=]+)$/) do |posts, match|
|
2017-02-08 04:53:37 +08:00
|
|
|
|
|
|
|
exact = true
|
|
|
|
|
2019-03-13 15:23:01 +08:00
|
|
|
category_slug, subcategory_slug = match.to_s.split(":")
|
|
|
|
next unless category_slug
|
2018-05-28 13:28:02 +08:00
|
|
|
|
2019-03-13 15:23:01 +08:00
|
|
|
if subcategory_slug
|
2019-03-12 14:11:21 +08:00
|
|
|
|
2020-03-20 12:36:50 +08:00
|
|
|
category_id, _ = DB.query_single(<<~SQL, category_slug.downcase, subcategory_slug.downcase)
|
|
|
|
SELECT sub.id
|
|
|
|
FROM categories sub
|
|
|
|
JOIN categories c ON sub.parent_category_id = c.id
|
|
|
|
WHERE LOWER(c.slug) = ? AND LOWER(sub.slug) = ?
|
|
|
|
ORDER BY c.id
|
|
|
|
LIMIT 1
|
|
|
|
SQL
|
|
|
|
|
2016-05-11 17:53:54 +08:00
|
|
|
else
|
|
|
|
# main category
|
2019-03-13 15:23:01 +08:00
|
|
|
if category_slug[0] == "="
|
|
|
|
category_slug = category_slug[1..-1]
|
2017-02-08 04:53:37 +08:00
|
|
|
else
|
|
|
|
exact = false
|
|
|
|
end
|
|
|
|
|
2019-03-13 15:23:01 +08:00
|
|
|
category_id = Category.where("lower(slug) = ?", category_slug.downcase)
|
2017-02-08 04:53:37 +08:00
|
|
|
.order('case when parent_category_id is null then 0 else 1 end')
|
|
|
|
.pluck(:id)
|
|
|
|
.first
|
2016-05-11 17:53:54 +08:00
|
|
|
end
|
|
|
|
|
|
|
|
if category_id
|
2017-02-08 04:53:37 +08:00
|
|
|
category_ids = [category_id]
|
|
|
|
|
|
|
|
unless exact
|
|
|
|
category_ids +=
|
|
|
|
Category.where('parent_category_id = ?', category_id).pluck(:id)
|
|
|
|
end
|
2019-03-19 11:23:14 +08:00
|
|
|
|
|
|
|
@category_filter_matched ||= true
|
2017-02-08 04:53:37 +08:00
|
|
|
posts.where("topics.category_id IN (?)", category_ids)
|
2016-05-11 17:53:54 +08:00
|
|
|
else
|
2017-08-02 06:15:04 +08:00
|
|
|
# try a possible tag match
|
2019-10-21 18:32:27 +08:00
|
|
|
tag_id = Tag.where_name(category_slug).pluck_first(:id)
|
2017-08-02 06:15:04 +08:00
|
|
|
if (tag_id)
|
2019-06-27 15:53:26 +08:00
|
|
|
posts.where(<<~SQL, tag_id)
|
|
|
|
topics.id IN (
|
|
|
|
SELECT DISTINCT(tt.topic_id)
|
|
|
|
FROM topic_tags tt
|
|
|
|
WHERE tt.tag_id = ?
|
|
|
|
)
|
|
|
|
SQL
|
2017-08-02 06:15:04 +08:00
|
|
|
else
|
2019-06-27 15:53:26 +08:00
|
|
|
if tag_group_id = TagGroup.find_id_by_slug(category_slug)
|
|
|
|
posts.where(<<~SQL, tag_group_id)
|
|
|
|
topics.id IN (
|
|
|
|
SELECT DISTINCT(tt.topic_id)
|
|
|
|
FROM topic_tags tt
|
|
|
|
WHERE tt.tag_id in (
|
|
|
|
SELECT tag_id
|
|
|
|
FROM tag_group_memberships
|
|
|
|
WHERE tag_group_id = ?
|
|
|
|
)
|
|
|
|
)
|
|
|
|
SQL
|
|
|
|
|
2017-08-02 06:15:04 +08:00
|
|
|
# a bit yucky but we got to add the term back in
|
2019-06-27 15:53:26 +08:00
|
|
|
elsif match.to_s.length >= SiteSetting.min_search_term_length
|
|
|
|
posts.where <<~SQL
|
|
|
|
posts.id IN (
|
|
|
|
SELECT post_id FROM post_search_data pd1
|
|
|
|
WHERE pd1.search_data @@ #{Search.ts_query(term: "##{match}")})
|
|
|
|
SQL
|
2017-08-02 06:15:04 +08:00
|
|
|
end
|
|
|
|
end
|
2016-05-11 17:53:54 +08:00
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^group:(.+)$/) do |posts, match|
|
2019-10-21 18:32:27 +08:00
|
|
|
group_id = Group.where('name ilike ? OR (id = ? AND id > 0)', match, match.to_i).pluck_first(:id)
|
2015-09-15 15:39:14 +08:00
|
|
|
if group_id
|
|
|
|
posts.where("posts.user_id IN (select gu.user_id from group_users gu where gu.group_id = ?)", group_id)
|
|
|
|
else
|
|
|
|
posts.where("1 = 0")
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^user:(.+)$/) do |posts, match|
|
2019-10-21 18:32:27 +08:00
|
|
|
user_id = User.where(staged: false).where('username_lower = ? OR id = ?', match.downcase, match.to_i).pluck_first(:id)
|
2015-08-14 09:53:16 +08:00
|
|
|
if user_id
|
|
|
|
posts.where("posts.user_id = #{user_id}")
|
|
|
|
else
|
|
|
|
posts.where("1 = 0")
|
|
|
|
end
|
2015-06-23 10:14:06 +08:00
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^\@([a-zA-Z0-9_\-.]+)$/) do |posts, match|
|
2019-10-21 18:32:27 +08:00
|
|
|
user_id = User.where(staged: false).where(username_lower: match.downcase).pluck_first(:id)
|
2016-05-12 16:43:44 +08:00
|
|
|
if user_id
|
|
|
|
posts.where("posts.user_id = #{user_id}")
|
|
|
|
else
|
|
|
|
posts.where("1 = 0")
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^before:(.*)$/) do |posts, match|
|
2016-03-14 20:27:02 +08:00
|
|
|
if date = Search.word_to_date(match)
|
|
|
|
posts.where("posts.created_at < ?", date)
|
|
|
|
else
|
|
|
|
posts
|
|
|
|
end
|
2015-06-23 11:21:50 +08:00
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^after:(.*)$/) do |posts, match|
|
2016-03-14 20:27:02 +08:00
|
|
|
if date = Search.word_to_date(match)
|
|
|
|
posts.where("posts.created_at > ?", date)
|
|
|
|
else
|
|
|
|
posts
|
|
|
|
end
|
2015-06-23 11:21:50 +08:00
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^tags?:([\p{L}0-9,\-_+]+)$/) do |posts, match|
|
2018-01-04 05:45:21 +08:00
|
|
|
search_tags(posts, match, positive: true)
|
|
|
|
end
|
2017-05-31 22:36:15 +08:00
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^\-tags?:([\p{L}0-9,\-_+]+)$/) do |posts, match|
|
2018-01-04 05:45:21 +08:00
|
|
|
search_tags(posts, match, positive: false)
|
2016-04-26 03:55:15 +08:00
|
|
|
end
|
|
|
|
|
2019-02-25 07:55:24 +08:00
|
|
|
advanced_filter(/^filetypes?:([a-zA-Z0-9,\-_]+)$/) do |posts, match|
|
2017-07-07 01:11:32 +08:00
|
|
|
file_extensions = match.split(",").map(&:downcase)
|
2017-06-21 03:20:06 +08:00
|
|
|
posts.where("posts.id IN (
|
2018-08-04 07:56:26 +08:00
|
|
|
SELECT post_id
|
|
|
|
FROM topic_links
|
|
|
|
WHERE extension IN (:file_extensions)
|
|
|
|
|
2017-07-04 23:50:08 +08:00
|
|
|
UNION
|
2018-08-04 07:56:26 +08:00
|
|
|
|
|
|
|
SELECT post_uploads.post_id
|
|
|
|
FROM uploads
|
|
|
|
JOIN post_uploads ON post_uploads.upload_id = uploads.id
|
|
|
|
WHERE lower(uploads.extension) IN (:file_extensions)
|
|
|
|
)", file_extensions: file_extensions)
|
2017-06-21 03:20:06 +08:00
|
|
|
end
|
|
|
|
|
2013-05-23 02:36:14 +08:00
|
|
|
private
|
2013-02-06 03:16:51 +08:00
|
|
|
|
2018-01-04 05:45:21 +08:00
|
|
|
def search_tags(posts, match, positive:)
|
|
|
|
return if match.nil?
|
2018-10-05 17:23:52 +08:00
|
|
|
match.downcase!
|
2018-01-04 05:45:21 +08:00
|
|
|
modifier = positive ? "" : "NOT"
|
2014-10-18 11:54:11 +08:00
|
|
|
|
2018-01-04 05:45:21 +08:00
|
|
|
if match.include?('+')
|
|
|
|
tags = match.split('+')
|
|
|
|
|
|
|
|
posts.where("topics.id #{modifier} IN (
|
|
|
|
SELECT tt.topic_id
|
|
|
|
FROM topic_tags tt, tags
|
|
|
|
WHERE tt.tag_id = tags.id
|
|
|
|
GROUP BY tt.topic_id
|
2018-10-05 17:23:52 +08:00
|
|
|
HAVING to_tsvector(#{default_ts_config}, array_to_string(array_agg(lower(tags.name)), ' ')) @@ to_tsquery(#{default_ts_config}, ?)
|
2018-09-28 17:27:08 +08:00
|
|
|
)", tags.join('&'))
|
2018-01-04 05:45:21 +08:00
|
|
|
else
|
|
|
|
tags = match.split(",")
|
|
|
|
|
|
|
|
posts.where("topics.id #{modifier} IN (
|
|
|
|
SELECT DISTINCT(tt.topic_id)
|
|
|
|
FROM topic_tags tt, tags
|
2018-10-05 17:23:52 +08:00
|
|
|
WHERE tt.tag_id = tags.id AND lower(tags.name) IN (?)
|
2018-09-28 17:27:08 +08:00
|
|
|
)", tags)
|
2018-01-04 05:45:21 +08:00
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
def process_advanced_search!(term)
|
2015-09-15 15:21:46 +08:00
|
|
|
term.to_s.scan(/(([^" \t\n\x0B\f\r]+)?(("[^"]+")?))/).to_a.map do |(word, _)|
|
|
|
|
next if word.blank?
|
2015-06-23 10:14:06 +08:00
|
|
|
|
|
|
|
found = false
|
|
|
|
|
|
|
|
Search.advanced_filters.each do |matcher, block|
|
2015-09-15 15:21:46 +08:00
|
|
|
cleaned = word.gsub(/["']/, "")
|
|
|
|
if cleaned =~ matcher
|
2015-06-23 10:14:06 +08:00
|
|
|
(@filters ||= []) << [block, $1]
|
|
|
|
found = true
|
|
|
|
end
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2015-06-23 10:14:06 +08:00
|
|
|
|
2017-05-24 23:24:41 +08:00
|
|
|
if word == 'order:latest' || word == 'l'
|
2014-09-03 20:10:18 +08:00
|
|
|
@order = :latest
|
|
|
|
nil
|
2017-03-30 01:33:23 +08:00
|
|
|
elsif word == 'order:latest_topic'
|
|
|
|
@order = :latest_topic
|
|
|
|
nil
|
2019-02-25 07:55:24 +08:00
|
|
|
elsif word == 'in:title' || word == 't'
|
2018-02-20 11:41:00 +08:00
|
|
|
@in_title = true
|
|
|
|
nil
|
2015-06-26 15:36:28 +08:00
|
|
|
elsif word =~ /topic:(\d+)/
|
|
|
|
topic_id = $1.to_i
|
|
|
|
if topic_id > 1
|
|
|
|
topic = Topic.find_by(id: topic_id)
|
|
|
|
if @guardian.can_see?(topic)
|
|
|
|
@search_context = topic
|
|
|
|
end
|
2014-09-03 19:54:10 +08:00
|
|
|
end
|
2018-06-07 13:28:18 +08:00
|
|
|
nil
|
2014-10-18 12:38:58 +08:00
|
|
|
elsif word == 'order:views'
|
|
|
|
@order = :views
|
2018-06-07 13:28:18 +08:00
|
|
|
nil
|
2015-09-15 15:21:46 +08:00
|
|
|
elsif word == 'order:likes'
|
|
|
|
@order = :likes
|
2018-06-07 13:28:18 +08:00
|
|
|
nil
|
2020-01-28 18:11:33 +08:00
|
|
|
elsif word == 'in:all'
|
|
|
|
@search_all_topics = true
|
|
|
|
nil
|
2019-07-22 22:55:49 +08:00
|
|
|
elsif %w{in:private in:personal}.include?(word) # remove private after 2.4 release
|
|
|
|
@search_pms = true
|
|
|
|
nil
|
|
|
|
elsif word == "in:personal-direct"
|
2014-12-05 07:22:39 +08:00
|
|
|
@search_pms = true
|
2019-07-22 22:55:49 +08:00
|
|
|
@direct_pms_only = true
|
2018-06-07 13:28:18 +08:00
|
|
|
nil
|
2019-07-22 22:55:49 +08:00
|
|
|
elsif word =~ /^personal_messages:(.+)$/
|
2016-08-20 02:58:51 +08:00
|
|
|
@search_pms = true
|
2018-06-07 13:28:18 +08:00
|
|
|
nil
|
|
|
|
else
|
2015-06-23 10:14:06 +08:00
|
|
|
found ? nil : word
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2014-09-03 19:54:10 +08:00
|
|
|
end.compact.join(' ')
|
|
|
|
end
|
|
|
|
|
2013-05-24 02:26:51 +08:00
|
|
|
def find_grouped_results
|
|
|
|
if @results.type_filter.present?
|
|
|
|
raise Discourse::InvalidAccess.new("invalid type filter") unless Search.facets.include?(@results.type_filter)
|
2019-05-07 10:22:37 +08:00
|
|
|
# calling protected methods
|
2013-05-24 02:26:51 +08:00
|
|
|
send("#{@results.type_filter}_search")
|
2013-05-23 02:36:14 +08:00
|
|
|
else
|
2014-06-17 15:53:45 +08:00
|
|
|
unless @search_context
|
2015-06-23 10:14:06 +08:00
|
|
|
user_search if @term.present?
|
|
|
|
category_search if @term.present?
|
2017-08-25 23:52:18 +08:00
|
|
|
tags_search if @term.present?
|
2019-03-04 17:30:09 +08:00
|
|
|
groups_search if @term.present?
|
2013-05-23 02:36:14 +08:00
|
|
|
end
|
2013-05-24 02:26:51 +08:00
|
|
|
topic_search
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2013-02-06 03:16:51 +08:00
|
|
|
|
2013-05-24 02:26:51 +08:00
|
|
|
@results
|
2014-03-08 03:59:29 +08:00
|
|
|
rescue ActiveRecord::StatementInvalid
|
|
|
|
# In the event of a PG:Error return nothing, it is likely they used a foreign language whose
|
|
|
|
# locale is not supported by postgres
|
2013-05-24 02:26:51 +08:00
|
|
|
end
|
2018-06-07 13:28:18 +08:00
|
|
|
|
2013-05-23 02:36:14 +08:00
|
|
|
# If we're searching for a single topic
|
|
|
|
def single_topic(id)
|
2019-01-14 17:30:45 +08:00
|
|
|
if @opts[:restrict_to_archetype].present?
|
|
|
|
archetype = @opts[:restrict_to_archetype] == Archetype.default ? Archetype.default : Archetype.private_message
|
|
|
|
post = Post.joins(:topic)
|
|
|
|
.where("topics.id = :id AND topics.archetype = :archetype AND posts.post_number = 1", id: id, archetype: archetype)
|
|
|
|
.first
|
|
|
|
else
|
|
|
|
post = Post.find_by(topic_id: id, post_number: 1)
|
|
|
|
end
|
2014-09-02 17:15:08 +08:00
|
|
|
return nil unless @guardian.can_see?(post)
|
2013-05-14 05:04:41 +08:00
|
|
|
|
2014-09-02 17:15:08 +08:00
|
|
|
@results.add(post)
|
2013-05-24 02:26:51 +08:00
|
|
|
@results
|
2013-05-23 02:36:14 +08:00
|
|
|
end
|
|
|
|
|
2013-05-24 02:26:51 +08:00
|
|
|
def secure_category_ids
|
|
|
|
return @secure_category_ids unless @secure_category_ids.nil?
|
|
|
|
@secure_category_ids = @guardian.secure_category_ids
|
2013-05-23 02:36:14 +08:00
|
|
|
end
|
2013-02-06 03:16:51 +08:00
|
|
|
|
2013-05-24 02:26:51 +08:00
|
|
|
def category_search
|
2013-12-13 16:00:48 +08:00
|
|
|
# scope is leaking onto Category, this is not good and probably a bug in Rails
|
|
|
|
# the secure_category_ids will invoke the same method on User, it calls Category.where
|
|
|
|
# however the scope from the query below is leaking in to Category, this works around
|
|
|
|
# the issue while we figure out what is up in Rails
|
|
|
|
secure_category_ids
|
|
|
|
|
2013-05-24 02:26:51 +08:00
|
|
|
categories = Category.includes(:category_search_data)
|
|
|
|
.where("category_search_data.search_data @@ #{ts_query}")
|
2013-08-26 05:18:11 +08:00
|
|
|
.references(:category_search_data)
|
2013-05-24 02:26:51 +08:00
|
|
|
.order("topics_month DESC")
|
|
|
|
.secured(@guardian)
|
2017-08-01 08:28:16 +08:00
|
|
|
.limit(limit)
|
2013-05-14 05:04:41 +08:00
|
|
|
|
2014-09-02 17:15:08 +08:00
|
|
|
categories.each do |category|
|
|
|
|
@results.add(category)
|
2013-05-23 02:36:14 +08:00
|
|
|
end
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2013-05-23 02:36:14 +08:00
|
|
|
|
2013-05-24 02:26:51 +08:00
|
|
|
def user_search
|
2015-10-29 02:56:08 +08:00
|
|
|
return if SiteSetting.hide_user_profiles_from_public && !@guardian.user
|
|
|
|
|
2013-05-24 02:26:51 +08:00
|
|
|
users = User.includes(:user_search_data)
|
2015-11-05 06:04:37 +08:00
|
|
|
.references(:user_search_data)
|
2015-11-19 04:06:59 +08:00
|
|
|
.where(active: true)
|
|
|
|
.where(staged: false)
|
2015-11-05 06:04:37 +08:00
|
|
|
.where("user_search_data.search_data @@ #{ts_query("simple")}")
|
2013-05-24 02:26:51 +08:00
|
|
|
.order("CASE WHEN username_lower = '#{@original_term.downcase}' THEN 0 ELSE 1 END")
|
|
|
|
.order("last_posted_at DESC")
|
2017-08-01 08:28:16 +08:00
|
|
|
.limit(limit)
|
2013-05-23 02:36:14 +08:00
|
|
|
|
2014-09-02 17:15:08 +08:00
|
|
|
users.each do |user|
|
|
|
|
@results.add(user)
|
2013-05-24 02:26:51 +08:00
|
|
|
end
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2013-05-23 02:36:14 +08:00
|
|
|
|
2019-03-04 17:30:09 +08:00
|
|
|
def groups_search
|
|
|
|
groups = Group
|
|
|
|
.visible_groups(@guardian.user, "name ASC", include_everyone: false)
|
2019-03-08 16:23:44 +08:00
|
|
|
.where("name ILIKE :term OR full_name ILIKE :term", term: "%#{@term}%")
|
2019-03-04 17:30:09 +08:00
|
|
|
|
|
|
|
groups.each { |group| @results.add(group) }
|
|
|
|
end
|
|
|
|
|
2017-08-25 23:52:18 +08:00
|
|
|
def tags_search
|
|
|
|
return unless SiteSetting.tagging_enabled
|
|
|
|
|
|
|
|
tags = Tag.includes(:tag_search_data)
|
|
|
|
.where("tag_search_data.search_data @@ #{ts_query}")
|
|
|
|
.references(:tag_search_data)
|
|
|
|
.order("name asc")
|
|
|
|
.limit(limit)
|
|
|
|
|
|
|
|
tags.each do |tag|
|
|
|
|
@results.add(tag)
|
|
|
|
end
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2017-08-25 23:52:18 +08:00
|
|
|
|
2019-03-26 17:01:19 +08:00
|
|
|
PHRASE_MATCH_REGEXP_PATTERN = '"([^"]+)"'
|
|
|
|
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
def posts_query(limit, type_filter: nil, aggregate_search: false)
|
2015-09-25 23:43:04 +08:00
|
|
|
posts = Post.where(post_type: Topic.visible_post_types(@guardian.user))
|
2014-12-04 10:46:52 +08:00
|
|
|
.joins(:post_search_data, :topic)
|
|
|
|
.joins("LEFT JOIN categories ON categories.id = topics.category_id")
|
2015-02-19 09:56:49 +08:00
|
|
|
|
2018-05-07 17:43:55 +08:00
|
|
|
is_topic_search = @search_context.present? && @search_context.is_a?(Topic)
|
|
|
|
|
2019-05-01 23:36:20 +08:00
|
|
|
posts = posts.where("topics.visible") unless is_topic_search
|
2014-12-04 10:46:52 +08:00
|
|
|
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
if type_filter === "private_messages" || (is_topic_search && @search_context.private_message?)
|
2014-12-04 10:46:52 +08:00
|
|
|
posts = posts.where("topics.archetype = ?", Archetype.private_message)
|
2013-08-02 08:31:36 +08:00
|
|
|
|
2015-06-23 10:14:06 +08:00
|
|
|
unless @guardian.is_admin?
|
|
|
|
posts = posts.private_posts_for_user(@guardian.user)
|
|
|
|
end
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
elsif type_filter === "all_topics"
|
2020-01-28 18:11:33 +08:00
|
|
|
private_posts = posts.where("topics.archetype = ?", Archetype.private_message)
|
2020-01-28 19:26:42 +08:00
|
|
|
private_posts = private_posts.private_posts_for_user(@guardian.user)
|
2020-01-28 18:11:33 +08:00
|
|
|
|
|
|
|
posts = posts.where("topics.archetype <> ?", Archetype.private_message).or(private_posts)
|
2015-06-23 10:14:06 +08:00
|
|
|
else
|
|
|
|
posts = posts.where("topics.archetype <> ?", Archetype.private_message)
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2016-07-25 13:06:25 +08:00
|
|
|
|
|
|
|
if @term.present?
|
2015-02-19 09:56:49 +08:00
|
|
|
if is_topic_search
|
2016-07-25 13:06:25 +08:00
|
|
|
|
|
|
|
term_without_quote = @term
|
|
|
|
if @term =~ /"(.+)"/
|
|
|
|
term_without_quote = $1
|
|
|
|
end
|
|
|
|
|
|
|
|
if @term =~ /'(.+)'/
|
|
|
|
term_without_quote = $1
|
2014-10-18 12:19:08 +08:00
|
|
|
end
|
|
|
|
|
2015-06-23 10:14:06 +08:00
|
|
|
posts = posts.joins('JOIN users u ON u.id = posts.user_id')
|
|
|
|
posts = posts.where("posts.raw || ' ' || u.username || ' ' || COALESCE(u.name, '') ilike ?", "%#{term_without_quote}%")
|
2018-06-07 13:28:18 +08:00
|
|
|
else
|
2018-02-20 11:41:00 +08:00
|
|
|
# A is for title
|
2015-06-23 10:14:06 +08:00
|
|
|
# B is for category
|
2018-02-20 11:41:00 +08:00
|
|
|
# C is for tags
|
|
|
|
# D is for cooked
|
2015-06-23 10:14:06 +08:00
|
|
|
weights = @in_title ? 'A' : (SiteSetting.tagging_enabled ? 'ABCD' : 'ABD')
|
2020-07-16 12:21:19 +08:00
|
|
|
posts = posts.where(post_number: 1) if @in_title
|
2015-06-23 10:14:06 +08:00
|
|
|
posts = posts.where("post_search_data.search_data @@ #{ts_query(weight_filter: weights)}")
|
2019-03-26 17:01:19 +08:00
|
|
|
exact_terms = @term.scan(Regexp.new(PHRASE_MATCH_REGEXP_PATTERN)).flatten
|
2019-03-18 15:25:45 +08:00
|
|
|
|
2015-06-23 10:14:06 +08:00
|
|
|
exact_terms.each do |exact|
|
2018-05-08 13:59:03 +08:00
|
|
|
posts = posts.where("posts.raw ilike :exact OR topics.title ilike :exact", exact: "%#{exact}%")
|
2014-10-18 12:34:05 +08:00
|
|
|
end
|
2015-06-23 10:14:06 +08:00
|
|
|
end
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2014-10-18 12:19:08 +08:00
|
|
|
|
2013-05-25 04:17:09 +08:00
|
|
|
@filters.each do |block, match|
|
|
|
|
if block.arity == 1
|
2015-06-23 10:14:06 +08:00
|
|
|
posts = instance_exec(posts, &block) || posts
|
2013-05-25 04:17:09 +08:00
|
|
|
else
|
|
|
|
posts = instance_exec(posts, match, &block) || posts
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2013-05-25 02:03:45 +08:00
|
|
|
end if @filters
|
|
|
|
|
2013-05-25 04:17:09 +08:00
|
|
|
# If we have a search context, prioritize those posts first
|
2019-03-18 15:25:45 +08:00
|
|
|
posts =
|
|
|
|
if @search_context.present?
|
|
|
|
if @search_context.is_a?(User)
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
if type_filter === "private_messages"
|
2019-07-22 22:55:49 +08:00
|
|
|
@direct_pms_only ? posts : posts.private_posts_for_user(@search_context)
|
2019-03-18 15:25:45 +08:00
|
|
|
else
|
|
|
|
posts.where("posts.user_id = #{@search_context.id}")
|
|
|
|
end
|
|
|
|
elsif @search_context.is_a?(Category)
|
|
|
|
category_ids = Category
|
|
|
|
.where(parent_category_id: @search_context.id)
|
|
|
|
.pluck(:id)
|
|
|
|
.push(@search_context.id)
|
|
|
|
|
|
|
|
posts.where("topics.category_id in (?)", category_ids)
|
|
|
|
elsif @search_context.is_a?(Topic)
|
|
|
|
posts.where("topics.id = #{@search_context.id}")
|
|
|
|
.order("posts.post_number #{@order == :latest ? "DESC" : ""}")
|
2019-11-14 08:40:26 +08:00
|
|
|
elsif @search_context.is_a?(Tag)
|
|
|
|
posts = posts
|
|
|
|
.joins("LEFT JOIN topic_tags ON topic_tags.topic_id = topics.id")
|
|
|
|
.joins("LEFT JOIN tags ON tags.id = topic_tags.tag_id")
|
|
|
|
posts.where("tags.id = #{@search_context.id}")
|
2013-05-25 02:03:45 +08:00
|
|
|
end
|
2019-03-18 15:25:45 +08:00
|
|
|
else
|
2019-03-19 11:23:14 +08:00
|
|
|
posts = categories_ignored(posts) unless @category_filter_matched
|
|
|
|
posts
|
2013-05-25 02:03:45 +08:00
|
|
|
end
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
|
|
|
|
if aggregate_search
|
|
|
|
aggregate_relation = Post.unscoped
|
FIX: Search by relevance may return incorrect post number.
Follow up to d8c796bc4.
Note that his change increases query time by around 40% in the following
benchmark against `dev.discourse.org` but this is a tradeoff that has to be taken so that relevance
search is accurate.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number ORDER BY rank DESC, bumped_at DESC))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
18.040 (± 0.0%) i/s - 181.000 in 10.035241s
current aggregate search query with proper ranking
12.992 (± 0.0%) i/s - 130.000 in 10.007214s
Comparison:
current aggregate search query: 18.0 i/s
current aggregate search query with proper ranking: 13.0 i/s - 1.39x (± 0.00) slower
```
2020-07-15 11:41:45 +08:00
|
|
|
.select("subquery.topic_id id")
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
.group("subquery.topic_id")
|
|
|
|
|
|
|
|
posts = posts.select(posts.arel.projections)
|
|
|
|
end
|
|
|
|
|
2020-04-27 22:35:27 +08:00
|
|
|
if @order == :latest
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
posts = posts.reorder("posts.created_at DESC")
|
|
|
|
|
|
|
|
if aggregate_search
|
|
|
|
aggregate_relation = aggregate_relation
|
FIX: Search by relevance may return incorrect post number.
Follow up to d8c796bc4.
Note that his change increases query time by around 40% in the following
benchmark against `dev.discourse.org` but this is a tradeoff that has to be taken so that relevance
search is accurate.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number ORDER BY rank DESC, bumped_at DESC))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
18.040 (± 0.0%) i/s - 181.000 in 10.035241s
current aggregate search query with proper ranking
12.992 (± 0.0%) i/s - 130.000 in 10.007214s
Comparison:
current aggregate search query: 18.0 i/s
current aggregate search query with proper ranking: 13.0 i/s - 1.39x (± 0.00) slower
```
2020-07-15 11:41:45 +08:00
|
|
|
.select(
|
|
|
|
"(ARRAY_AGG(subquery.post_number ORDER BY subquery.created_at DESC))[1] post_number",
|
|
|
|
"MAX(subquery.created_at) created_at"
|
|
|
|
)
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
.order("created_at DESC")
|
2014-08-23 04:55:19 +08:00
|
|
|
end
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
elsif @order == :latest_topic
|
|
|
|
posts = posts.order("topic_created_at DESC")
|
|
|
|
|
|
|
|
if aggregate_search
|
|
|
|
posts = posts.select("topics.created_at topic_created_at")
|
|
|
|
|
|
|
|
aggregate_relation = aggregate_relation
|
FIX: Search by relevance may return incorrect post number.
Follow up to d8c796bc4.
Note that his change increases query time by around 40% in the following
benchmark against `dev.discourse.org` but this is a tradeoff that has to be taken so that relevance
search is accurate.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number ORDER BY rank DESC, bumped_at DESC))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
18.040 (± 0.0%) i/s - 181.000 in 10.035241s
current aggregate search query with proper ranking
12.992 (± 0.0%) i/s - 130.000 in 10.007214s
Comparison:
current aggregate search query: 18.0 i/s
current aggregate search query with proper ranking: 13.0 i/s - 1.39x (± 0.00) slower
```
2020-07-15 11:41:45 +08:00
|
|
|
.select(
|
|
|
|
"(ARRAY_AGG(subquery.post_number ORDER BY subquery.topic_created_at DESC))[1] post_number",
|
|
|
|
"MAX(subquery.topic_created_at) topic_created_at"
|
|
|
|
)
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
.order("topic_created_at DESC")
|
|
|
|
end
|
|
|
|
elsif @order == :views
|
|
|
|
posts = posts.order("topic_views DESC")
|
|
|
|
|
|
|
|
if aggregate_search
|
|
|
|
posts = posts.select("topics.views topic_views")
|
|
|
|
|
|
|
|
aggregate_relation = aggregate_relation
|
FIX: Search by relevance may return incorrect post number.
Follow up to d8c796bc4.
Note that his change increases query time by around 40% in the following
benchmark against `dev.discourse.org` but this is a tradeoff that has to be taken so that relevance
search is accurate.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number ORDER BY rank DESC, bumped_at DESC))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
18.040 (± 0.0%) i/s - 181.000 in 10.035241s
current aggregate search query with proper ranking
12.992 (± 0.0%) i/s - 130.000 in 10.007214s
Comparison:
current aggregate search query: 18.0 i/s
current aggregate search query with proper ranking: 13.0 i/s - 1.39x (± 0.00) slower
```
2020-07-15 11:41:45 +08:00
|
|
|
.select(
|
|
|
|
"(ARRAY_AGG(subquery.post_number ORDER BY subquery.topic_views DESC))[1] post_number",
|
|
|
|
"MAX(subquery.topic_views) topic_views"
|
|
|
|
)
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
.order("topic_views DESC")
|
|
|
|
end
|
|
|
|
elsif @order == :likes
|
|
|
|
posts = posts.order("posts.like_count DESC")
|
|
|
|
else
|
|
|
|
rank = <<~SQL
|
|
|
|
TS_RANK_CD(
|
|
|
|
post_search_data.search_data,
|
|
|
|
#{@term.blank? ? '' : ts_query(weight_filter: weights)},
|
|
|
|
#{SiteSetting.search_ranking_normalization}|32
|
|
|
|
)
|
|
|
|
SQL
|
|
|
|
|
|
|
|
category_priority_weights = <<~SQL
|
2020-04-27 22:35:27 +08:00
|
|
|
(
|
|
|
|
CASE categories.search_priority
|
|
|
|
WHEN #{Searchable::PRIORITIES[:very_low]}
|
|
|
|
THEN #{SiteSetting.category_search_priority_very_low_weight}
|
|
|
|
WHEN #{Searchable::PRIORITIES[:low]}
|
|
|
|
THEN #{SiteSetting.category_search_priority_low_weight}
|
|
|
|
WHEN #{Searchable::PRIORITIES[:high]}
|
|
|
|
THEN #{SiteSetting.category_search_priority_high_weight}
|
|
|
|
WHEN #{Searchable::PRIORITIES[:very_high]}
|
|
|
|
THEN #{SiteSetting.category_search_priority_very_high_weight}
|
|
|
|
ELSE
|
|
|
|
CASE WHEN topics.closed
|
|
|
|
THEN 0.9
|
|
|
|
ELSE 1
|
|
|
|
END
|
|
|
|
END
|
|
|
|
)
|
|
|
|
SQL
|
2019-03-20 15:24:23 +08:00
|
|
|
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
data_ranking =
|
|
|
|
if @term.blank?
|
|
|
|
"(#{category_priority_weights})"
|
|
|
|
else
|
|
|
|
"(#{rank} * #{category_priority_weights})"
|
|
|
|
end
|
|
|
|
|
|
|
|
if aggregate_search
|
|
|
|
posts = posts.select("#{data_ranking} rank", "topics.bumped_at topic_bumped_at")
|
|
|
|
.order("rank DESC", "topic_bumped_at DESC")
|
|
|
|
|
|
|
|
aggregate_relation = aggregate_relation
|
FIX: Search by relevance may return incorrect post number.
Follow up to d8c796bc4.
Note that his change increases query time by around 40% in the following
benchmark against `dev.discourse.org` but this is a tradeoff that has to be taken so that relevance
search is accurate.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number ORDER BY rank DESC, bumped_at DESC))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
18.040 (± 0.0%) i/s - 181.000 in 10.035241s
current aggregate search query with proper ranking
12.992 (± 0.0%) i/s - 130.000 in 10.007214s
Comparison:
current aggregate search query: 18.0 i/s
current aggregate search query with proper ranking: 13.0 i/s - 1.39x (± 0.00) slower
```
2020-07-15 11:41:45 +08:00
|
|
|
.select(
|
|
|
|
"(ARRAY_AGG(subquery.post_number ORDER BY subquery.rank DESC, subquery.topic_bumped_at DESC))[1] post_number",
|
|
|
|
"MAX(subquery.rank) rank", "MAX(subquery.topic_bumped_at) topic_bumped_at"
|
|
|
|
)
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
.order("rank DESC", "topic_bumped_at DESC")
|
2018-06-07 13:28:18 +08:00
|
|
|
else
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
posts = posts.order("#{data_ranking} DESC", "topics.bumped_at DESC")
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2013-05-23 02:36:14 +08:00
|
|
|
end
|
|
|
|
|
2017-08-01 03:28:48 +08:00
|
|
|
if secure_category_ids.present?
|
|
|
|
posts = posts.where("(categories.id IS NULL) OR (NOT categories.read_restricted) OR (categories.id IN (?))", secure_category_ids).references(:categories)
|
|
|
|
else
|
|
|
|
posts = posts.where("(categories.id IS NULL) OR (NOT categories.read_restricted)").references(:categories)
|
2013-05-23 23:13:23 +08:00
|
|
|
end
|
|
|
|
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
if aggregate_search
|
|
|
|
posts = yield(posts) if block_given?
|
|
|
|
posts = aggregate_relation.from(posts)
|
|
|
|
end
|
|
|
|
|
2017-08-01 03:28:48 +08:00
|
|
|
posts = posts.offset(offset)
|
2017-07-21 00:12:34 +08:00
|
|
|
posts.limit(limit)
|
2014-05-15 22:31:45 +08:00
|
|
|
end
|
|
|
|
|
2019-03-18 15:25:45 +08:00
|
|
|
def categories_ignored(posts)
|
|
|
|
posts.where(<<~SQL, Searchable::PRIORITIES[:ignore])
|
|
|
|
categories.id NOT IN (
|
|
|
|
SELECT categories.id WHERE categories.search_priority = ?
|
|
|
|
)
|
|
|
|
SQL
|
|
|
|
end
|
|
|
|
|
2018-02-20 11:41:00 +08:00
|
|
|
def self.default_ts_config
|
|
|
|
"'#{Search.ts_config}'"
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2015-08-10 15:41:14 +08:00
|
|
|
|
2017-08-01 03:28:48 +08:00
|
|
|
def default_ts_config
|
|
|
|
self.class.default_ts_config
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2015-08-10 15:41:14 +08:00
|
|
|
|
2020-07-08 15:29:02 +08:00
|
|
|
def self.ts_query(term: , ts_config: nil, joiner: nil, weight_filter: nil)
|
2020-07-28 11:53:25 +08:00
|
|
|
to_tsquery(
|
|
|
|
ts_config: ts_config,
|
|
|
|
term: set_tsquery_weight_filter(term, weight_filter),
|
|
|
|
joiner: joiner
|
|
|
|
)
|
|
|
|
end
|
|
|
|
|
|
|
|
def self.to_tsquery(ts_config: nil, term:, joiner: nil)
|
2017-08-01 03:28:48 +08:00
|
|
|
ts_config = ActiveRecord::Base.connection.quote(ts_config) if ts_config
|
2020-07-28 11:53:25 +08:00
|
|
|
tsquery = "TO_TSQUERY(#{ts_config || default_ts_config}, '#{term}')"
|
2020-07-08 15:29:02 +08:00
|
|
|
tsquery = "REPLACE(#{tsquery}::text, '&', '#{PG::Connection.escape_string(joiner)}')::tsquery" if joiner
|
|
|
|
tsquery
|
2016-08-13 01:04:46 +08:00
|
|
|
end
|
2014-08-28 15:15:53 +08:00
|
|
|
|
2020-07-28 11:53:25 +08:00
|
|
|
def self.set_tsquery_weight_filter(term, weight_filter)
|
|
|
|
term = term.gsub("'", "''")
|
|
|
|
"''#{PG::Connection.escape_string(term)}'':*#{weight_filter}"
|
|
|
|
end
|
|
|
|
|
2016-08-13 01:04:46 +08:00
|
|
|
def ts_query(ts_config = nil, weight_filter: nil)
|
2015-09-18 15:16:37 +08:00
|
|
|
@ts_query_cache ||= {}
|
|
|
|
@ts_query_cache["#{ts_config || default_ts_config} #{@term} #{weight_filter}"] ||=
|
2018-08-31 09:46:55 +08:00
|
|
|
Search.ts_query(term: @term, ts_config: ts_config, weight_filter: weight_filter)
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2015-09-18 15:16:37 +08:00
|
|
|
|
2017-07-21 00:12:34 +08:00
|
|
|
def wrap_rows(query)
|
2017-03-08 22:46:23 +08:00
|
|
|
"SELECT *, row_number() over() row_number FROM (#{query.to_sql}) xxx"
|
2015-09-18 15:49:57 +08:00
|
|
|
end
|
2014-09-01 15:04:40 +08:00
|
|
|
|
2016-08-13 01:04:46 +08:00
|
|
|
def aggregate_post_sql(opts)
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
default_opts = {
|
|
|
|
type_filter: opts[:type_filter]
|
|
|
|
}
|
|
|
|
|
|
|
|
min_id = Search.min_post_id
|
2016-08-13 01:04:46 +08:00
|
|
|
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
if @order == :likes
|
|
|
|
# likes are a pain to aggregate so skip
|
|
|
|
query = posts_query(limit, **default_opts).select('topics.id', 'posts.post_number')
|
|
|
|
|
|
|
|
if min_id > 0
|
|
|
|
low_set = query.dup.where("post_search_data.post_id < #{min_id}")
|
|
|
|
high_set = query.where("post_search_data.post_id >= #{min_id}")
|
|
|
|
|
|
|
|
{ default: wrap_rows(high_set), remaining: wrap_rows(low_set) }
|
2018-06-07 13:28:18 +08:00
|
|
|
else
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
{ default: wrap_rows(query) }
|
|
|
|
end
|
|
|
|
else
|
|
|
|
query = posts_query(limit, **default_opts, aggregate_search: true) do |posts|
|
|
|
|
if min_id > 0
|
|
|
|
posts.select("post_search_data.post_id post_search_data_post_id")
|
|
|
|
else
|
|
|
|
posts
|
|
|
|
end
|
2016-08-13 01:04:46 +08:00
|
|
|
end
|
|
|
|
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
if min_id > 0
|
|
|
|
low_set = query.dup.where("subquery.post_search_data_post_id < #{min_id}")
|
|
|
|
high_set = query.where("subquery.post_search_data_post_id >= #{min_id}")
|
2018-06-07 13:28:18 +08:00
|
|
|
|
FIX: Ensure that aggregating search shows the post with the higest rank.
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
2020-07-07 15:36:57 +08:00
|
|
|
{ default: wrap_rows(high_set), remaining: wrap_rows(low_set) }
|
|
|
|
else
|
|
|
|
{ default: wrap_rows(query) }
|
|
|
|
end
|
2016-08-13 01:04:46 +08:00
|
|
|
end
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2014-09-03 10:13:13 +08:00
|
|
|
|
2016-08-13 01:04:46 +08:00
|
|
|
def aggregate_posts(post_sql)
|
|
|
|
return [] unless post_sql
|
|
|
|
|
|
|
|
posts_eager_loads(Post)
|
|
|
|
.joins("JOIN (#{post_sql}) x ON x.id = posts.topic_id AND x.post_number = posts.post_number")
|
|
|
|
.order('row_number')
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2014-09-01 15:04:40 +08:00
|
|
|
|
2016-08-13 05:18:12 +08:00
|
|
|
def aggregate_search(opts = {})
|
2016-08-13 01:04:46 +08:00
|
|
|
post_sql = aggregate_post_sql(opts)
|
2017-03-08 22:46:23 +08:00
|
|
|
|
2016-08-13 05:18:12 +08:00
|
|
|
added = 0
|
|
|
|
|
2016-08-13 01:04:46 +08:00
|
|
|
aggregate_posts(post_sql[:default]).each do |p|
|
|
|
|
@results.add(p)
|
2014-08-23 04:55:19 +08:00
|
|
|
added += 1
|
|
|
|
end
|
|
|
|
|
2014-12-04 10:46:52 +08:00
|
|
|
if added < limit
|
|
|
|
aggregate_posts(post_sql[:remaining]).each { |p| @results.add(p) }
|
|
|
|
end
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
2014-12-04 10:46:52 +08:00
|
|
|
|
2013-05-24 02:26:51 +08:00
|
|
|
def private_messages_search
|
2017-05-31 07:14:09 +08:00
|
|
|
raise Discourse::InvalidAccess.new("anonymous can not search PMs") unless @guardian.user
|
|
|
|
|
2020-01-28 18:11:33 +08:00
|
|
|
aggregate_search(type_filter: "private_messages")
|
|
|
|
end
|
|
|
|
|
|
|
|
def all_topics_search
|
|
|
|
aggregate_search(type_filter: "all_topics")
|
2013-05-23 23:13:23 +08:00
|
|
|
end
|
2013-02-06 03:16:51 +08:00
|
|
|
|
2017-05-31 07:14:09 +08:00
|
|
|
def topic_search
|
2014-09-02 17:15:08 +08:00
|
|
|
if @search_context.is_a?(Topic)
|
2017-05-31 07:14:09 +08:00
|
|
|
posts = posts_eager_loads(posts_query(limit))
|
|
|
|
.where('posts.topic_id = ?', @search_context.id)
|
|
|
|
|
|
|
|
posts.each do |post|
|
|
|
|
@results.add(post)
|
|
|
|
end
|
2018-06-07 13:28:18 +08:00
|
|
|
else
|
2017-07-21 00:12:34 +08:00
|
|
|
aggregate_search
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
|
|
|
end
|
2017-05-31 07:14:09 +08:00
|
|
|
|
|
|
|
def posts_eager_loads(query)
|
2020-07-17 16:27:30 +08:00
|
|
|
query = query.includes(:user, :post_search_data)
|
2017-05-31 07:14:09 +08:00
|
|
|
topic_eager_loads = [:category]
|
2018-06-07 13:28:18 +08:00
|
|
|
|
2017-05-31 07:14:09 +08:00
|
|
|
if SiteSetting.tagging_enabled
|
|
|
|
topic_eager_loads << :tags
|
|
|
|
end
|
|
|
|
|
|
|
|
query.includes(topic: topic_eager_loads)
|
2018-06-07 13:28:18 +08:00
|
|
|
end
|
|
|
|
|
2013-02-06 03:16:51 +08:00
|
|
|
end
|