discourse/lib/search.rb

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1573 lines
45 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
class Search
DIACRITICS ||= /([\u0300-\u036f]|[\u1AB0-\u1AFF]|[\u1DC0-\u1DFF]|[\u20D0-\u20FF])/
HIGHLIGHT_CSS_CLASS = "search-highlight"
2013-02-06 03:16:51 +08:00
cattr_accessor :preloaded_topic_custom_fields
self.preloaded_topic_custom_fields = Set.new
def self.on_preload(&blk)
(@preload ||= Set.new) << blk
end
def self.preload(results, object)
@preload.each { |preload| preload.call(results, object) } if @preload
end
2013-02-06 03:16:51 +08:00
def self.per_facet
5
end
def self.per_filter
SiteSetting.search_page_size
end
2013-02-06 03:16:51 +08:00
def self.facets
%w[topic category user private_messages tags all_topics exclude_topics]
2013-02-06 03:16:51 +08:00
end
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.
# base docker config
#
case locale.split("_")[0].to_sym
when :da
"danish"
when :nl
"dutch"
when :en
"english"
when :fi
"finnish"
when :fr
"french"
when :de
"german"
when :hu
"hungarian"
when :it
"italian"
when :nb
"norwegian"
when :pt
"portuguese"
when :ro
"romanian"
when :ru
"russian"
when :es
"spanish"
when :sv
"swedish"
when :tr
"turkish"
else
"simple" # use the 'simple' stemmer for other languages
end
end
def self.unaccent(str)
if SiteSetting.search_ignore_accents
DB.query("SELECT unaccent(:str)", str: str)[0].unaccent
else
str
end
end
def self.wrap_unaccent(expr)
SiteSetting.search_ignore_accents ? "unaccent(#{expr})" : expr
end
def self.segment_chinese?
%w[zh_TW zh_CN].include?(SiteSetting.default_locale) || SiteSetting.search_tokenize_chinese
end
def self.segment_japanese?
SiteSetting.default_locale == "ja" || SiteSetting.search_tokenize_japanese
end
def self.japanese_punctuation_regexp
# Regexp adapted from https://github.com/6/tiny_segmenter/blob/15a5b825993dfd2c662df3766f232051716bef5b/lib/tiny_segmenter.rb#L7
@japanese_punctuation_regexp ||=
Regexp.compile("[-–—―.。・()(){}【】⟨⟩、、,,،…‥〽「」『』〜~!:?\"'|__“”;/⁄/«»]")
end
def self.clean_term(term)
term = term.to_s.dup
# Removes any zero-width characters from search terms
term.gsub!(/[\u200B-\u200D\uFEFF]/, "")
# Replace curly quotes to regular quotes
term.gsub!(/[\u201c\u201d]/, '"')
# Replace fancy apostophes to regular apostophes
term.gsub!(/[\u02b9\u02bb\u02bc\u02bd\u02c8\u2018\u2019\u201b\u2032\uff07]/, "'")
term
end
def self.prepare_data(search_data, purpose = nil)
data = search_data.dup
data.force_encoding("UTF-8")
data = clean_term(data)
if purpose != :topic && need_segmenting?(data)
if segment_chinese?
require "cppjieba_rb" unless defined?(CppjiebaRb)
segmented_data = []
# We need to split up the string here because Cppjieba has a bug where text starting with numeric chars will
# be split into two segments. For example, '123abc' becomes '123' and 'abc' after segmentation.
data.scan(/(?<chinese>[\p{Han}。,、“”《》…\.:?!;()]+)|([^\p{Han}]+)/) do
match_data = $LAST_MATCH_INFO
if match_data[:chinese]
segments = CppjiebaRb.segment(match_data.to_s, mode: :mix)
segments = CppjiebaRb.filter_stop_word(segments) if ts_config != "english"
segments = segments.filter { |s| s.present? }
segmented_data << segments.join(" ")
else
segmented_data << match_data.to_s.squish
end
end
data = segmented_data.join(" ")
elsif segment_japanese?
data.gsub!(japanese_punctuation_regexp, " ")
data = TinyJapaneseSegmenter.segment(data)
data = data.filter { |s| s.present? }
data = data.join(" ")
else
data.squish!
end
end
data.gsub!(/\S+/) do |str|
if str =~ %r{\A["]?((https?://)[\S]+)["]?\z}
begin
uri = URI.parse(Regexp.last_match[1])
uri.query = nil
str = uri.to_s
rescue URI::Error
# don't fail if uri does not parse
end
end
str
end
data
end
def self.word_to_date(str)
return Time.zone.now.beginning_of_day.days_ago(str.to_i) if str =~ /\A[0-9]{1,3}\z/
if str =~ /\A([12][0-9]{3})(-([0-1]?[0-9]))?(-([0-3]?[0-9]))?\z/
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
)
end
return Time.zone.now.beginning_of_day.yesterday if str.downcase == "yesterday"
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
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
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
Discourse
.cache
.fetch("search-min-post-id:#{SiteSetting.search_recent_posts_size}", expires_in: 1.week) do
min_post_id_no_cache
end
end
def self.need_segmenting?(data)
return false if data.match?(/\A\d+\z/)
!URI.parse(data).path.start_with?("/")
rescue URI::InvalidURIError
true
end
attr_accessor :term
attr_reader :clean_term, :guardian
def initialize(term, opts = nil)
@opts = opts || {}
@guardian = @opts[:guardian] || Guardian.new
@search_context = @opts[:search_context]
@blurb_length = @opts[:blurb_length]
@valid = true
@page = @opts[:page]
@search_all_pms = false
term = Search.clean_term(term)
@clean_term = term
@in_title = false
term = process_advanced_search!(term)
if !@order &&
SiteSetting.search_default_sort_order !=
SearchSortOrderSiteSetting.value_from_id(:relevance)
@order = SearchSortOrderSiteSetting.id_from_value(SiteSetting.search_default_sort_order)
end
if term.present?
@term = Search.prepare_data(term, Topic === @search_context ? :topic : nil)
@original_term = Search.escape_string(@term)
end
if @search_pms || @search_all_pms || @opts[:type_filter] == "private_messages"
@opts[:type_filter] = "private_messages"
@search_context ||= @guardian.user
unless @search_context.present? && @guardian.can_see_private_messages?(@search_context.id)
raise Discourse::InvalidAccess.new
end
end
@opts[:type_filter] = "all_topics" if @search_all_topics && @guardian.user
@results =
GroupedSearchResults.new(
type_filter: @opts[:type_filter],
term: clean_term,
blurb_term: term,
search_context: @search_context,
blurb_length: @blurb_length,
is_header_search: !use_full_page_limit,
can_lazy_load_categories: @guardian.can_lazy_load_categories?,
)
end
def limit
if use_full_page_limit
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
def valid?
@valid
end
def use_full_page_limit
@opts[:search_type] == :full_page || Topic === @search_context
end
def self.execute(term, opts = nil)
self.new(term, opts).execute
2013-05-14 05:04:41 +08:00
end
# Query a term
def execute(readonly_mode: Discourse.readonly_mode?)
if log_query?(readonly_mode)
status, search_log_id =
SearchLog.log(
term: @clean_term,
2017-07-14 01:34:31 +08:00
search_type: @opts[:search_type],
ip_address: @opts[:ip_address],
user_agent: @opts[:user_agent],
2017-07-14 01:34:31 +08:00
user_id: @opts[:user_id],
)
@results.search_log_id = search_log_id unless status == :error
2017-07-14 01:34:31 +08:00
end
unless @filters.present? || @opts[:search_for_id]
min_length = min_search_term_length
terms = (@term || "").split(/\s(?=(?:[^"]|"[^"]*")*$)/).reject { |t| t.length < min_length }
if terms.blank?
@term = ""
@valid = false
return
end
end
2013-05-14 05:04:41 +08:00
# If the term is a number or url to a topic, just include that topic
if @opts[:search_for_id] && %w[topic private_messages all_topics].include?(@results.type_filter)
if @term =~ /\A\d+\z/
single_topic(@term.to_i)
else
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
2013-05-14 05:04:41 +08:00
end
end
find_grouped_results if @results.posts.blank?
if preloaded_topic_custom_fields.present? && @results.posts.present?
topics = @results.posts.map(&:topic)
Topic.preload_custom_fields(topics, preloaded_topic_custom_fields)
end
Search.preload(@results, self)
@results
2013-05-14 05:04:41 +08:00
end
2013-02-06 03:16:51 +08:00
def self.advanced_order(trigger, &block)
advanced_orders[trigger] = block
end
def self.advanced_orders
@advanced_orders ||= {}
end
def self.advanced_filter(trigger, &block)
advanced_filters[trigger] = block
end
def self.advanced_filters
@advanced_filters ||= {}
end
def self.custom_topic_eager_load(tables = nil, &block)
(@custom_topic_eager_loads ||= []) << (tables || block)
end
def self.custom_topic_eager_loads
Array.wrap(@custom_topic_eager_loads)
end
advanced_filter(/\Ain:personal-direct\z/i) do |posts|
if @guardian.user
posts.joins("LEFT JOIN topic_allowed_groups tg ON posts.topic_id = tg.topic_id").where(
<<~SQL,
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
user_id: @guardian.user.id,
)
end
end
advanced_filter(/\Ain:all-pms\z/i) { |posts| posts.private_posts if @guardian.is_admin? }
advanced_filter(/\Ain:tagged\z/i) do |posts|
posts.where("EXISTS (SELECT 1 FROM topic_tags WHERE topic_tags.topic_id = posts.topic_id)")
end
advanced_filter(/\Ain:untagged\z/i) do |posts|
posts.joins(
"LEFT JOIN topic_tags ON
topic_tags.topic_id = posts.topic_id",
).where("topic_tags.id IS NULL")
end
advanced_filter(/\Astatus:open\z/i) do |posts|
posts.where("NOT topics.closed AND NOT topics.archived")
end
advanced_filter(/\Astatus:closed\z/i) { |posts| posts.where("topics.closed") }
advanced_filter(/\Astatus:public\z/i) do |posts|
category_ids = Category.where(read_restricted: false).pluck(:id)
posts.where("topics.category_id in (?)", category_ids)
end
advanced_filter(/\Astatus:archived\z/i) { |posts| posts.where("topics.archived") }
advanced_filter(/\Astatus:noreplies\z/i) { |posts| posts.where("topics.posts_count = 1") }
advanced_filter(/\Astatus:single_user\z/i) { |posts| posts.where("topics.participant_count = 1") }
advanced_filter(/\Aposts_count:(\d+)\z/i) do |posts, match|
posts.where("topics.posts_count = ?", match.to_i)
end
advanced_filter(/\Amin_post_count:(\d+)\z/i) do |posts, match|
posts.where("topics.posts_count >= ?", match.to_i)
end
advanced_filter(/\Amin_posts:(\d+)\z/i) do |posts, match|
posts.where("topics.posts_count >= ?", match.to_i)
end
advanced_filter(/\Amax_posts:(\d+)\z/i) do |posts, match|
posts.where("topics.posts_count <= ?", match.to_i)
end
advanced_filter(/\Ain:first|^f\z/i) { |posts| posts.where("posts.post_number = 1") }
advanced_filter(/\Ain:pinned\z/i) { |posts| posts.where("topics.pinned_at IS NOT NULL") }
advanced_filter(/\Ain:wiki\z/i) { |posts, match| posts.where(wiki: true) }
2016-05-13 16:26:53 +08:00
advanced_filter(/\Abadge:(.*)\z/i) do |posts, match|
badge_id = Badge.where("name ilike ? OR id = ?", match, match.to_i).pick(:id)
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
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 = ? AND
pa.post_action_type_id = ? AND
deleted_at IS NULL
)",
@guardian.user.id,
post_action_type,
)
end
advanced_filter(/\Ain:(likes)\z/i) do |posts, match|
post_action_type_filter(posts, PostActionType.types[:like]) if @guardian.user
end
# NOTE: With polymorphic bookmarks it may make sense to possibly expand
# this at some point, as it only acts on posts at the moment. On the other
# hand, this may not be necessary, as the user bookmark list has advanced
# search based on a RegisteredBookmarkable's #search_query method.
advanced_filter(/\Ain:(bookmarks)\z/i) do |posts, match|
posts.where(<<~SQL, @guardian.user.id) if @guardian.user
posts.id IN (
SELECT bookmarkable_id FROM bookmarks
WHERE bookmarks.user_id = ? AND bookmarks.bookmarkable_type = 'Post'
)
SQL
end
advanced_filter(/\Ain:posted\z/i) do |posts|
posts.where("posts.user_id = ?", @guardian.user.id) if @guardian.user
end
advanced_filter(/\Ain:(created|mine)\z/i) do |posts|
posts.where(user_id: @guardian.user.id, post_number: 1) if @guardian.user
end
advanced_filter(/\Acreated:@(.*)\z/i) do |posts, match|
user_id = User.where(username_lower: match.downcase).pick(:id)
posts.where(user_id: user_id, post_number: 1)
end
advanced_filter(/\Ain:(watching|tracking)\z/i) do |posts, match|
if @guardian.user
level = TopicUser.notification_levels[match.downcase.to_sym]
posts.where(
"posts.topic_id IN (
SELECT tu.topic_id FROM topic_users tu
WHERE tu.user_id = :user_id AND
tu.notification_level >= :level
)",
user_id: @guardian.user.id,
level: level,
)
end
end
advanced_filter(/\Ain:seen\z/i) do |posts|
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
AND post_timings.user_id = #{ActiveRecord::Base.connection.quote(@guardian.user.id)}
",
)
end
end
advanced_filter(/\Ain:unseen\z/i) do |posts|
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
AND post_timings.user_id = #{ActiveRecord::Base.connection.quote(@guardian.user.id)}
",
).where("post_timings.user_id IS NULL")
end
end
advanced_filter(/\Awith:images\z/i) { |posts| posts.where("posts.image_upload_id IS NOT NULL") }
advanced_filter(/\Acategor(?:y|ies):(.+)\z/i) do |posts, terms|
category_ids = []
matches =
terms
.split(",")
.map do |term|
if term[0] == "="
[term[1..-1], true]
else
[term, false]
end
end
.to_h
if matches.present?
sql = <<~SQL
SELECT c.id, term
FROM
categories c
JOIN
unnest(ARRAY[:matches]) AS term ON
c.slug ILIKE term OR
c.name ILIKE term OR
(term ~ '^[0-9]{1,10}$' AND c.id = term::int)
SQL
found = DB.query(sql, matches: matches.keys)
if found.present?
found.each do |row|
category_ids << row.id
@category_filter_matched ||= true
category_ids += Category.subcategory_ids(row.id) if !matches[row.term]
end
end
end
if category_ids.present?
posts.where("topics.category_id IN (?)", category_ids.uniq)
else
posts.where("1 = 0")
end
end
advanced_filter(/\A\#([\p{L}\p{M}0-9\-:=]+)\z/i) do |posts, match|
2019-03-13 15:23:01 +08:00
category_slug, subcategory_slug = match.to_s.split(":")
next unless category_slug
exact = true
if category_slug[0] == "="
category_slug = category_slug[1..-1]
else
exact = false
end
category_id =
if subcategory_slug
Category
.where("lower(slug) = ?", subcategory_slug.downcase)
.where(
parent_category_id:
Category.where("lower(slug) = ?", category_slug.downcase).select(:id),
)
.pick(:id)
else
Category
.where("lower(slug) = ?", category_slug.downcase)
.order("case when parent_category_id is null then 0 else 1 end")
.pick(:id)
end
if category_id
category_ids = [category_id]
category_ids += Category.subcategory_ids(category_id) if !exact
@category_filter_matched ||= true
posts.where("topics.category_id IN (?)", category_ids)
else
# try a possible tag match
tag_id = Tag.where_name(category_slug).pick(:id)
if (tag_id)
posts.where(<<~SQL, tag_id)
topics.id IN (
SELECT DISTINCT(tt.topic_id)
FROM topic_tags tt
WHERE tt.tag_id = ?
)
SQL
else
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
# a bit yucky but we got to add the term back in
elsif match.to_s.length >= 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
end
end
end
end
advanced_filter(/\Agroup:(.+)\z/i) do |posts, match|
group_query =
Group
.visible_groups(@guardian.user)
.members_visible_groups(@guardian.user)
.where("groups.name ILIKE ? OR (groups.id = ? AND groups.id > 0)", match, match.to_i)
DiscoursePluginRegistry.search_groups_set_query_callbacks.each do |cb|
group_query = cb.call(group_query, @term, @guardian)
end
group_id = group_query.pick(:id)
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
advanced_filter(/\Agroup_messages:(.+)\z/i) do |posts, match|
group_id =
Group
.visible_groups(@guardian.user)
.members_visible_groups(@guardian.user)
.where(has_messages: true)
.where("name ilike ? OR (id = ? AND id > 0)", match, match.to_i)
.pick(:id)
if group_id
posts.where(
"posts.topic_id IN (SELECT topic_id FROM topic_allowed_groups WHERE group_id = ?)",
group_id,
)
else
posts.where("1 = 0")
end
end
advanced_filter(/\Auser:(.+)\z/i) do |posts, match|
user_id =
User
.where(staged: false)
.where("username_lower = ? OR id = ?", match.downcase, match.to_i)
.pick(:id)
if user_id
posts.where("posts.user_id = ?", user_id)
else
posts.where("1 = 0")
end
end
advanced_filter(/\A\@(\S+)\z/i) do |posts, match|
username = User.normalize_username(match)
user_id = User.not_staged.where(username_lower: username).pick(:id)
user_id = @guardian.user&.id if !user_id && username == "me"
2016-05-12 16:43:44 +08:00
if user_id
posts.where("posts.user_id = ?", user_id)
2016-05-12 16:43:44 +08:00
else
posts.where("1 = 0")
end
end
advanced_filter(/\Abefore:(.*)\z/i) do |posts, match|
if date = Search.word_to_date(match)
posts.where("posts.created_at < ?", date)
else
posts
end
end
advanced_filter(/\Aafter:(.*)\z/i) do |posts, match|
if date = Search.word_to_date(match)
posts.where("posts.created_at > ?", date)
else
posts
end
end
advanced_filter(/\Atags?:([\p{L}\p{M}0-9,\-_+]+)\z/i) do |posts, match|
search_tags(posts, match, positive: true)
end
advanced_filter(/\A\-tags?:([\p{L}\p{M}0-9,\-_+]+)\z/i) do |posts, match|
search_tags(posts, match, positive: false)
end
advanced_filter(/\Afiletypes?:([a-zA-Z0-9,\-_]+)\z/i) do |posts, match|
file_extensions = match.split(",").map(&:downcase)
posts.where(
"posts.id IN (
2018-08-04 07:56:26 +08:00
SELECT post_id
FROM topic_links
WHERE extension IN (:file_extensions)
UNION
2018-08-04 07:56:26 +08:00
SELECT upload_references.target_id
2018-08-04 07:56:26 +08:00
FROM uploads
JOIN upload_references ON upload_references.target_type = 'Post' AND upload_references.upload_id = uploads.id
2018-08-04 07:56:26 +08:00
WHERE lower(uploads.extension) IN (:file_extensions)
)",
file_extensions: file_extensions,
)
end
advanced_filter(/\Amin_views:(\d+)\z/i) do |posts, match|
posts.where("topics.views >= ?", match.to_i)
end
advanced_filter(/\Amax_views:(\d+)\z/i) do |posts, match|
posts.where("topics.views <= ?", match.to_i)
end
def apply_filters(posts)
@filters.each do |block, match|
if block.arity == 1
posts = instance_exec(posts, &block) || posts
else
posts = instance_exec(posts, match, &block) || posts
end
end if @filters
posts
end
def apply_order(
posts,
aggregate_search: false,
allow_relevance_search: true,
type_filter: "all_topics"
)
if @order == :latest
if aggregate_search
posts = posts.order("MAX(posts.created_at) DESC")
else
posts = posts.reorder("posts.created_at DESC")
end
elsif @order == :oldest
if aggregate_search
posts = posts.order("MAX(posts.created_at) ASC")
else
posts = posts.reorder("posts.created_at ASC")
end
elsif @order == :latest_topic
if aggregate_search
posts = posts.order("MAX(topics.created_at) DESC")
else
posts = posts.order("topics.created_at DESC")
end
elsif @order == :oldest_topic
if aggregate_search
posts = posts.order("MAX(topics.created_at) ASC")
else
posts = posts.order("topics.created_at ASC")
end
elsif @order == :views
if aggregate_search
posts = posts.order("MAX(topics.views) DESC")
else
posts = posts.order("topics.views DESC")
end
elsif @order == :likes
if aggregate_search
posts = posts.order("MAX(posts.like_count) DESC")
else
posts = posts.order("posts.like_count DESC")
end
elsif allow_relevance_search
posts = sort_by_relevance(posts, type_filter: type_filter, aggregate_search: aggregate_search)
end
if @order
advanced_order = Search.advanced_orders&.fetch(@order, nil)
posts = advanced_order.call(posts) if advanced_order
end
posts
end
private
2013-02-06 03:16:51 +08:00
def search_tags(posts, match, positive:)
return if match.nil?
match.downcase!
modifier = positive ? "" : "NOT"
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
HAVING to_tsvector(#{default_ts_config}, #{Search.wrap_unaccent("array_to_string(array_agg(lower(tags.name)), ' ')")}) @@ to_tsquery(#{default_ts_config}, ?)
)",
Search.unaccent(tags.join("&")),
)
else
tags = match.split(",")
posts.where(
"topics.id #{modifier} IN (
SELECT DISTINCT(tt.topic_id)
FROM topic_tags tt, tags
WHERE tt.tag_id = tags.id AND lower(tags.name) IN (?)
)",
tags,
)
end
end
def process_advanced_search!(term)
term
.to_s
.scan(/(([^" \t\n\x0B\f\r]+)?(("[^"]+")?))/)
.to_a
.map do |(word, _)|
next if word.blank?
found = false
Search.advanced_filters.each do |matcher, block|
case_insensitive_matcher =
Regexp.new(matcher.source, matcher.options | Regexp::IGNORECASE)
cleaned = word.gsub(/["']/, "")
if cleaned =~ case_insensitive_matcher
(@filters ||= []) << [block, $1]
found = true
end
end
if word == "l"
@order = :latest
nil
elsif word =~ /\Aorder:\w+\z/i
@order = word.downcase.gsub("order:", "").to_sym
nil
elsif word =~ /\Ain:title\z/i || word == "t"
@in_title = true
nil
elsif word =~ /\Atopic:(\d+)\z/i
topic_id = $1.to_i
if topic_id > 1
topic = Topic.find_by(id: topic_id)
@search_context = topic if @guardian.can_see?(topic)
end
2018-06-07 13:28:18 +08:00
nil
elsif word =~ /\Ain:all\z/i
@search_all_topics = true
nil
elsif word =~ /\Ain:personal\z/i
@search_pms = true
nil
elsif word =~ /\Ain:messages\z/i
@search_pms = true
nil
elsif word =~ /\Ain:personal-direct\z/i
@search_pms = true
nil
elsif word =~ /\Ain:all-pms\z/i
@search_all_pms = true
nil
elsif word =~ /\Agroup_messages:(.+)\z/i
@search_pms = true
nil
elsif word =~ /\Apersonal_messages:(.+)\z/i
if user = User.find_by_username($1)
@search_pms = true
@search_context = user
end
nil
elsif word =~ /\Ainclude:(invisible|unlisted)\z/i
@include_invisible = true if @guardian.can_see_unlisted_topics?
2018-06-07 13:28:18 +08:00
nil
else
found ? nil : word
end
2018-06-07 13:28:18 +08:00
end
.compact
.join(" ")
end
def find_grouped_results
if @results.type_filter.present?
if Search.facets.exclude?(@results.type_filter)
raise Discourse::InvalidAccess.new("invalid type filter")
end
# calling protected methods
send("#{@results.type_filter}_search")
else
if @term.present? && !@search_context
user_search
category_search
tags_search
groups_search
end
topic_search
2018-06-07 13:28:18 +08:00
end
2013-02-06 03:16:51 +08:00
@results
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
end
2018-06-07 13:28:18 +08:00
# If we're searching for a single topic
def single_topic(id)
if @opts[:restrict_to_archetype].present?
archetype =
(
if @opts[:restrict_to_archetype] == Archetype.default
Archetype.default
else
Archetype.private_message
end
)
post =
posts_scope.joins(:topic).find_by(
"topics.id = :id AND topics.archetype = :archetype AND posts.post_number = 1",
id: id,
archetype: archetype,
)
else
post = posts_scope.find_by(topic_id: id, post_number: 1)
end
return nil unless @guardian.can_see?(post)
2013-05-14 05:04:41 +08:00
@results.add(post)
@results
end
def secure_category_ids
return @secure_category_ids unless @secure_category_ids.nil?
@secure_category_ids = @guardian.secure_category_ids
end
2013-02-06 03:16:51 +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
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)
.order("topics_month DESC")
.secured(@guardian)
2017-08-01 08:28:16 +08:00
.limit(limit)
2013-05-14 05:04:41 +08:00
categories.each { |category| @results.add(category) }
2018-06-07 13:28:18 +08:00
end
def user_search
return if SiteSetting.hide_user_profiles_from_public && !@guardian.user
users =
User
.includes(:user_search_data)
.references(:user_search_data)
2015-11-19 04:06:59 +08:00
.where(active: true)
.where(staged: false)
.where("user_search_data.search_data @@ #{ts_query("simple")}")
.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)
if !SiteSetting.enable_listing_suspended_users_on_search && !@guardian.user&.admin
users = users.where(suspended_at: nil)
end
users = DiscoursePluginRegistry.apply_modifier(:search_user_search, users)
users_custom_data_query =
DB.query(<<~SQL, user_ids: users.pluck(:id), term: "%#{@original_term.downcase}%")
SELECT user_custom_fields.user_id, user_fields.name, user_custom_fields.value FROM user_custom_fields
INNER JOIN user_fields ON user_fields.id = REPLACE(user_custom_fields.name, 'user_field_', '')::INTEGER AND user_fields.searchable IS TRUE
WHERE user_id IN (:user_ids)
AND user_custom_fields.name LIKE 'user_field_%'
AND user_custom_fields.value ILIKE :term
SQL
users_custom_data =
users_custom_data_query.reduce({}) do |acc, row|
acc[row.user_id] = Array.wrap(acc[row.user_id]) << { name: row.name, value: row.value }
acc
end
users.each do |user|
user.custom_data = users_custom_data[user.id] || []
@results.add(user)
end
2018-06-07 13:28:18 +08:00
end
def groups_search
group_query =
Group.visible_groups(@guardian.user, "groups.name ASC", include_everyone: false).where(
"groups.name ILIKE :term OR groups.full_name ILIKE :term",
term: "%#{@term}%",
)
DiscoursePluginRegistry.search_groups_set_query_callbacks.each do |cb|
group_query = cb.call(group_query, @term, @guardian)
end
groups = group_query.limit(limit)
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)
hidden_tag_names = DiscourseTagging.hidden_tag_names(@guardian)
tags.each { |tag| @results.add(tag) if !hidden_tag_names.include?(tag.name) }
2018-06-07 13:28:18 +08:00
end
2017-08-25 23:52:18 +08:00
def exclude_topics_search
if @term.present?
user_search
category_search
tags_search
groups_search
end
end
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)
posts =
Post.where(post_type: Topic.visible_post_types(@guardian.user), hidden: false).joins(
:post_search_data,
:topic,
)
if type_filter != "private_messages"
posts = posts.joins("LEFT JOIN categories ON categories.id = topics.category_id")
end
is_topic_search = @search_context.present? && @search_context.is_a?(Topic)
posts = posts.where("topics.visible") unless is_topic_search || @include_invisible
if type_filter == "private_messages" || (is_topic_search && @search_context.private_message?)
posts =
posts.where(
"topics.archetype = ? AND post_search_data.private_message",
Archetype.private_message,
)
posts = posts.private_posts_for_user(@guardian.user) unless @guardian.is_admin?
elsif type_filter == "all_topics"
private_posts =
posts.where(
"topics.archetype = ? AND post_search_data.private_message",
Archetype.private_message,
).private_posts_for_user(@guardian.user)
posts =
posts.where(
"topics.archetype <> ? AND NOT post_search_data.private_message",
Archetype.private_message,
).or(private_posts)
else
posts =
posts.where(
"topics.archetype <> ? AND NOT post_search_data.private_message",
Archetype.private_message,
)
2018-06-07 13:28:18 +08:00
end
if @term.present?
if is_topic_search
term_without_quote = @term
term_without_quote = $1 if @term =~ /"(.+)"/
term_without_quote = $1 if @term =~ /'(.+)'/
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
posts = posts.where(post_number: 1) if @in_title
posts = posts.where("post_search_data.search_data @@ #{ts_query(weight_filter: weights)}")
exact_terms = @term.scan(Regexp.new(PHRASE_MATCH_REGEXP_PATTERN)).flatten
exact_terms.each do |exact|
posts =
posts.where("posts.raw ilike :exact OR topics.title ilike :exact", exact: "%#{exact}%")
end
end
2018-06-07 13:28:18 +08:00
end
posts = apply_filters(posts)
# If we have a search context, prioritize those posts first
posts =
if @search_context.present?
if @search_context.is_a?(User)
if type_filter == "private_messages"
if @guardian.is_admin? && !@search_all_pms
posts.private_posts_for_user(@search_context)
else
posts
end
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 is_topic_search
posts = posts.where("topics.id = ?", @search_context.id)
posts = posts.order("posts.post_number ASC") unless @order
posts
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)
end
else
posts = categories_ignored(posts) unless @category_filter_matched
posts
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 type_filter != "private_messages"
posts =
if secure_category_ids.present?
posts.where(
"(categories.id IS NULL) OR (NOT categories.read_restricted) OR (categories.id IN (?))",
secure_category_ids,
).references(:categories)
else
posts.where("(categories.id IS NULL) OR (NOT categories.read_restricted)").references(
:categories,
)
end
end
posts =
apply_order(
posts,
aggregate_search: aggregate_search,
allow_relevance_search: !is_topic_search,
type_filter: type_filter,
)
posts = posts.offset(offset)
posts.limit(limit)
end
def weights
# A is for title
# B is for category
# C is for tags
# D is for cooked
@in_title ? "A" : (SiteSetting.tagging_enabled ? "ABCD" : "ABD")
end
def sort_by_relevance(posts, type_filter:, aggregate_search:)
exact_rank = nil
if SiteSetting.prioritize_exact_search_title_match
exact_rank = ts_rank_cd(weight_filter: "A", prefix_match: false)
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
rank = ts_rank_cd(weight_filter: weights)
if type_filter != "private_messages"
category_search_priority = <<~SQL
(
CASE categories.search_priority
WHEN #{Searchable::PRIORITIES[:very_high]}
THEN 3
WHEN #{Searchable::PRIORITIES[:very_low]}
THEN 1
ELSE 2
END
)
SQL
rank_sort_priorities = [["topics.archived", 0.85], ["topics.closed", 0.9]]
rank_sort_priorities =
DiscoursePluginRegistry.apply_modifier(
:search_rank_sort_priorities,
rank_sort_priorities,
self,
)
category_priority_weights = <<~SQL
(
CASE categories.search_priority
WHEN #{Searchable::PRIORITIES[:low]}
THEN #{SiteSetting.category_search_priority_low_weight.to_f}
WHEN #{Searchable::PRIORITIES[:high]}
THEN #{SiteSetting.category_search_priority_high_weight.to_f}
ELSE 1.0
END
*
CASE
#{rank_sort_priorities.sort_by { |_, pri| -pri }.map { |k, v| "WHEN #{k} THEN #{v}" }.join("\n")}
ELSE 1.0
END
)
SQL
posts =
if aggregate_search
posts.order("MAX(#{category_search_priority}) DESC")
else
posts.order("#{category_search_priority} DESC")
end
if @term.present? && exact_rank
posts =
if aggregate_search
posts.order("MAX(#{exact_rank} * #{category_priority_weights}) DESC")
else
posts.order("#{exact_rank} * #{category_priority_weights} DESC")
end
end
data_ranking =
if @term.blank?
"(#{category_priority_weights})"
else
"(#{rank} * #{category_priority_weights})"
end
posts =
if aggregate_search
posts.order("MAX(#{data_ranking}) DESC")
else
posts.order("#{data_ranking} DESC")
end
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
posts.order("topics.bumped_at DESC")
2014-05-15 22:31:45 +08:00
end
def ts_rank_cd(weight_filter:, prefix_match: true)
<<~SQL
TS_RANK_CD(
#{SiteSetting.search_ranking_weights.present? ? "'#{SiteSetting.search_ranking_weights}'," : ""}
post_search_data.search_data,
#{@term.blank? ? "" : ts_query(weight_filter: weight_filter, prefix_match: prefix_match)},
#{SiteSetting.search_ranking_normalization}|32
)
SQL
end
def categories_ignored(posts)
posts.where(<<~SQL, Searchable::PRIORITIES[:ignore])
(categories.search_priority IS NULL OR categories.search_priority IS NOT NULL AND categories.search_priority <> ?)
SQL
end
def self.default_ts_config
"'#{Search.ts_config}'"
2018-06-07 13:28:18 +08:00
end
def default_ts_config
self.class.default_ts_config
2018-06-07 13:28:18 +08:00
end
def self.ts_query(term:, ts_config: nil, joiner: nil, weight_filter: nil, prefix_match: true)
to_tsquery(
ts_config: ts_config,
term: set_tsquery_weight_filter(term, weight_filter, prefix_match: prefix_match),
)
end
def self.to_tsquery(ts_config: nil, term:, joiner: nil)
ts_config = ActiveRecord::Base.connection.quote(ts_config) if ts_config
escaped_term = "'#{escape_string(unaccent(term))}'"
tsquery = "TO_TSQUERY(#{ts_config || default_ts_config}, #{escaped_term})"
# PG 14 and up default to using the followed by operator
# this restores the old behavior
tsquery = "REGEXP_REPLACE(#{tsquery}::text, '<->|<\\d+>', '&', 'g')::tsquery"
tsquery = "REPLACE(#{tsquery}::text, '&', '#{escape_string(joiner)}')::tsquery" if joiner
tsquery
end
def self.set_tsquery_weight_filter(term, weight_filter, prefix_match: true)
"'#{self.escape_string(term)}':#{prefix_match ? "*" : ""}#{weight_filter}"
end
def self.escape_string(term)
PG::Connection.escape_string(term).gsub('\\', '\\\\\\')
end
def ts_query(ts_config = nil, weight_filter: nil, prefix_match: true)
2015-09-18 15:16:37 +08:00
@ts_query_cache ||= {}
@ts_query_cache[
"#{ts_config || default_ts_config} #{@term} #{weight_filter} #{prefix_match}"
] ||= Search.ts_query(
term: @term,
ts_config: ts_config,
weight_filter: weight_filter,
prefix_match: prefix_match,
)
2018-06-07 13:28:18 +08:00
end
2015-09-18 15:16:37 +08:00
def wrap_rows(query)
"SELECT *, row_number() over() row_number FROM (#{query.to_sql}) xxx"
2015-09-18 15:49:57 +08:00
end
def aggregate_post_sql(opts)
min_id =
if SiteSetting.search_recent_regular_posts_offset_post_id > 0
if %w[all_topics private_message].include?(opts[:type_filter])
0
else
SiteSetting.search_recent_regular_posts_offset_post_id
end
else
# This is kept around for backwards compatibility.
# TODO: Drop this code path after Discourse 2.7 has been released.
Search.min_post_id
end
min_or_max = @order == :latest ? "max" : "min"
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
query =
if @order == :likes
# likes are a pain to aggregate so skip
posts_query(limit, type_filter: opts[:type_filter]).select("topics.id", "posts.post_number")
2018-06-07 13:28:18 +08:00
else
posts_query(limit, aggregate_search: true, type_filter: opts[:type_filter]).select(
"topics.id",
"#{min_or_max}(posts.post_number) post_number",
).group("topics.id")
end
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)
2018-06-07 13:28:18 +08:00
return { default: wrap_rows(high_set), remaining: wrap_rows(low_set) }
end
# double wrapping so we get correct row numbers
{ default: wrap_rows(query) }
2018-06-07 13:28:18 +08:00
end
def aggregate_posts(post_sql)
return [] unless post_sql
posts_scope(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
def aggregate_search(opts = {})
post_sql = aggregate_post_sql(opts)
added = 0
aggregate_posts(post_sql[:default]).each do |p|
@results.add(p)
added += 1
end
aggregate_posts(post_sql[:remaining]).each { |p| @results.add(p) } if added < limit
2018-06-07 13:28:18 +08:00
end
def private_messages_search
raise Discourse::InvalidAccess.new("anonymous can not search PMs") unless @guardian.user
aggregate_search(type_filter: "private_messages")
end
def all_topics_search
aggregate_search(type_filter: "all_topics")
end
2013-02-06 03:16:51 +08:00
def topic_search
if @search_context.is_a?(Topic)
posts =
posts_scope(posts_eager_loads(posts_query(limit))).where(
"posts.topic_id = ?",
@search_context.id,
)
posts.each { |post| @results.add(post) }
2018-06-07 13:28:18 +08:00
else
aggregate_search
2018-06-07 13:28:18 +08:00
end
end
def posts_eager_loads(query)
query = query.includes(:user, :post_search_data)
topic_eager_loads = [{ category: :parent_category }]
2018-06-07 13:28:18 +08:00
topic_eager_loads << :tags if SiteSetting.tagging_enabled
Search.custom_topic_eager_loads.each do |custom_loads|
topic_eager_loads.concat(
custom_loads.is_a?(Array) ? custom_loads : custom_loads.call(search_pms: @search_pms).to_a,
)
end
query.includes(topic: topic_eager_loads)
2018-06-07 13:28:18 +08:00
end
PERF: Limit characters used to generate headline for search blurb. We determined using the following benchmark script that limiting to 2500 chars would mean a maximum of 25ms spent generating headlines. ``` require 'benchmark/ips' string = <<~STRING Far far away, behind the word mountains... STRING def sql_excerpt(string, l = 1000000) DB.query_single(<<~SQL) SELECT TS_HEADLINE('english', left('#{string}', #{l}), PLAINTO_TSQUERY('mountains')) SQL end def ruby_excerpt(string) output = DB.query_single("SELECT '#{string}'")[0] Search::GroupedSearchResults::TextHelper.excerpt(output, 'mountains', radius: 100) end puts "Ruby Excerpt: #{ruby_excerpt(string)}" puts "SQL Excerpt: #{sql_excerpt(string)}" puts Benchmark.ips do |x| x.time = 10 [1000, 2500, 5000, 10000, 20000, 50000].each do |l| short_string = string[0..l] x.report("ts_headline excerpt #{l}") do sql_excerpt(short_string, l) end x.report("actionview excerpt #{l}") do ruby_excerpt(short_string) end end x.compare! end ``` ``` actionview excerpt 1000: 20570.7 i/s actionview excerpt 2500: 17863.1 i/s - 1.15x (± 0.00) slower actionview excerpt 5000: 14228.9 i/s - 1.45x (± 0.00) slower actionview excerpt 10000: 10906.2 i/s - 1.89x (± 0.00) slower actionview excerpt 20000: 6255.0 i/s - 3.29x (± 0.00) slower ts_headline excerpt 1000: 4337.5 i/s - 4.74x (± 0.00) slower actionview excerpt 50000: 3222.7 i/s - 6.38x (± 0.00) slower ts_headline excerpt 2500: 2240.4 i/s - 9.18x (± 0.00) slower ts_headline excerpt 5000: 1258.7 i/s - 16.34x (± 0.00) slower ts_headline excerpt 10000: 667.2 i/s - 30.83x (± 0.00) slower ts_headline excerpt 20000: 348.7 i/s - 58.98x (± 0.00) slower ts_headline excerpt 50000: 131.9 i/s - 155.91x (± 0.00) slower ```
2020-08-07 14:36:12 +08:00
# Limited for performance reasons since `TS_HEADLINE` is slow when the text
# document is too long.
MAX_LENGTH_FOR_HEADLINE = 2500
def posts_scope(default_scope = Post.all)
if SiteSetting.use_pg_headlines_for_excerpt
search_term = @term.present? ? Search.escape_string(@term) : nil
ts_config = default_ts_config
default_scope
.joins("INNER JOIN post_search_data pd ON pd.post_id = posts.id")
.joins("INNER JOIN topics t1 ON t1.id = posts.topic_id")
.select(
"TS_HEADLINE(
#{ts_config},
t1.fancy_title,
PLAINTO_TSQUERY(#{ts_config}, '#{search_term}'),
'StartSel=''<span class=\"#{HIGHLIGHT_CSS_CLASS}\">'', StopSel=''</span>'', HighlightAll=true'
) AS topic_title_headline",
"TS_HEADLINE(
#{ts_config},
LEFT(
TS_HEADLINE(
#{ts_config},
LEFT(pd.raw_data, #{MAX_LENGTH_FOR_HEADLINE}),
PLAINTO_TSQUERY(#{ts_config}, '#{search_term}'),
'ShortWord=0, MaxFragments=1, MinWords=50, MaxWords=51, StartSel='''', StopSel='''''
),
#{Search::GroupedSearchResults::BLURB_LENGTH}
),
PLAINTO_TSQUERY(#{ts_config}, '#{search_term}'),
'HighlightAll=true, StartSel=''<span class=\"#{HIGHLIGHT_CSS_CLASS}\">'', StopSel=''</span>'''
) AS headline",
"LEFT(pd.raw_data, 50) AS leading_raw_data",
"RIGHT(pd.raw_data, 50) AS trailing_raw_data",
default_scope.arel.projections,
)
else
default_scope
end
end
def log_query?(readonly_mode)
SiteSetting.log_search_queries? && @opts[:search_type].present? && !readonly_mode &&
@opts[:type_filter] != "exclude_topics"
end
def min_search_term_length
return @opts[:min_search_term_length] if @opts[:min_search_term_length]
if SiteSetting.search_tokenize_chinese
return SiteSetting.defaults.get("min_search_term_length", "zh_CN")
end
if SiteSetting.search_tokenize_japanese
return SiteSetting.defaults.get("min_search_term_length", "ja")
end
SiteSetting.min_search_term_length
end
2013-02-06 03:16:51 +08:00
end