Change trending statuses to only show one status from each account (#18181)

Calculate trends in temporary sets to avoid having to manage items
that go below the decay threshold while not having any moments
where a half-processed set is accessible to end-users
This commit is contained in:
Eugen Rochko 2022-04-29 22:42:42 +02:00 committed by GitHub
parent 5c691430e5
commit 6476f7e4da
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 82 additions and 83 deletions

View File

@ -64,28 +64,25 @@ class Trends::Base
redis.expire(used_key(at_time), 1.day.seconds) redis.expire(used_key(at_time), 1.day.seconds)
end end
def trim_older_items
redis.zremrangebyscore("#{key_prefix}:all", '-inf', '(0.3')
redis.zremrangebyscore("#{key_prefix}:allowed", '-inf', '(0.3')
end
def score_at_rank(rank) def score_at_rank(rank)
redis.zrevrange("#{key_prefix}:allowed", 0, rank, with_scores: true).last&.last || 0 redis.zrevrange("#{key_prefix}:allowed", 0, rank, with_scores: true).last&.last || 0
end end
# @param [Integer] id def replace_items(suffix, items)
# @param [Float] score tmp_prefix = "#{key_prefix}:tmp:#{SecureRandom.alphanumeric(6)}#{suffix}"
# @param [Hash<String, Boolean>] subsets allowed_items = filter_for_allowed_items(items)
def add_to_and_remove_from_subsets(id, score, subsets = {})
subsets.each_key do |subset|
key = [key_prefix, subset].compact.join(':')
if score.positive? && subsets[subset] redis.pipelined do |pipeline|
redis.zadd(key, score, id) items.each { |item| pipeline.zadd("#{tmp_prefix}:all", item[:score], item[:item].id) }
else allowed_items.each { |item| pipeline.zadd("#{tmp_prefix}:allowed", item[:score], item[:item].id) }
redis.zrem(key, id)
rename_set(pipeline, "#{tmp_prefix}:all", "#{key_prefix}:all#{suffix}", items)
rename_set(pipeline, "#{tmp_prefix}:allowed", "#{key_prefix}:allowed#{suffix}", allowed_items)
end end
end end
def filter_for_allowed_items(items)
raise NotImplementedError
end end
private private
@ -93,4 +90,12 @@ class Trends::Base
def used_key(at_time) def used_key(at_time)
"#{key_prefix}:used:#{at_time.beginning_of_day.to_i}" "#{key_prefix}:used:#{at_time.beginning_of_day.to_i}"
end end
def rename_set(pipeline, from_key, to_key, set_items)
if set_items.empty?
pipeline.del(to_key)
else
pipeline.rename(from_key, to_key)
end
end
end end

View File

@ -8,14 +8,15 @@ class Trends::Links < Trends::Base
review_threshold: 3, review_threshold: 3,
max_score_cooldown: 2.days.freeze, max_score_cooldown: 2.days.freeze,
max_score_halflife: 8.hours.freeze, max_score_halflife: 8.hours.freeze,
decay_threshold: 1,
} }
def register(status, at_time = Time.now.utc) def register(status, at_time = Time.now.utc)
original_status = status.reblog? ? status.reblog : status original_status = status.proper
return unless original_status.public_visibility? && status.public_visibility? && return unless (original_status.public_visibility? && status.public_visibility?) &&
!original_status.account.silenced? && !status.account.silenced? && !(original_status.account.silenced? || status.account.silenced?) &&
!original_status.spoiler_text? !(original_status.spoiler_text? || original_status.sensitive?)
original_status.preview_cards.each do |preview_card| original_status.preview_cards.each do |preview_card|
add(preview_card, status.account_id, at_time) if preview_card.appropriate_for_trends? add(preview_card, status.account_id, at_time) if preview_card.appropriate_for_trends?
@ -61,6 +62,9 @@ class Trends::Links < Trends::Base
private private
def calculate_scores(preview_cards, at_time) def calculate_scores(preview_cards, at_time)
global_items = []
locale_items = Hash.new { |h, key| h[key] = [] }
preview_cards.each do |preview_card| preview_cards.each do |preview_card|
expected = preview_card.history.get(at_time - 1.day).accounts.to_f expected = preview_card.history.get(at_time - 1.day).accounts.to_f
expected = 1.0 if expected.zero? expected = 1.0 if expected.zero?
@ -87,31 +91,21 @@ class Trends::Links < Trends::Base
decaying_score = max_score * (0.5**((at_time.to_f - max_time.to_f) / options[:max_score_halflife].to_f)) decaying_score = max_score * (0.5**((at_time.to_f - max_time.to_f) / options[:max_score_halflife].to_f))
add_to_and_remove_from_subsets(preview_card.id, decaying_score, { next unless decaying_score >= options[:decay_threshold]
all: true,
allowed: preview_card.trendable?,
})
next unless valid_locale?(preview_card.language) global_items << { score: decaying_score, item: preview_card }
locale_items[preview_card.language] << { score: decaying_score, item: preview_card } if valid_locale?(preview_card.language)
add_to_and_remove_from_subsets(preview_card.id, decaying_score, {
"all:#{preview_card.language}" => true,
"allowed:#{preview_card.language}" => preview_card.trendable?,
})
end end
trim_older_items replace_items('', global_items)
# Clean up localized sets by calculating the intersection with the main
# set. We do this instead of just deleting the localized sets to avoid
# having moments where the API returns empty results
redis.pipelined do
Trends.available_locales.each do |locale| Trends.available_locales.each do |locale|
redis.zinterstore("#{key_prefix}:all:#{locale}", ["#{key_prefix}:all:#{locale}", "#{key_prefix}:all"], aggregate: 'max') replace_items(":#{locale}", locale_items[locale])
redis.zinterstore("#{key_prefix}:allowed:#{locale}", ["#{key_prefix}:allowed:#{locale}", "#{key_prefix}:allowed"], aggregate: 'max')
end end
end end
def filter_for_allowed_items(items)
items.select { |item| item[:item].trendable? }
end end
def would_be_trending?(id) def would_be_trending?(id)

View File

@ -7,6 +7,7 @@ class Trends::Statuses < Trends::Base
threshold: 5, threshold: 5,
review_threshold: 3, review_threshold: 3,
score_halflife: 2.hours.freeze, score_halflife: 2.hours.freeze,
decay_threshold: 0.3,
} }
class Query < Trends::Query class Query < Trends::Query
@ -31,7 +32,7 @@ class Trends::Statuses < Trends::Base
end end
def register(status, at_time = Time.now.utc) def register(status, at_time = Time.now.utc)
add(status.proper, status.account_id, at_time) if eligible?(status) add(status.proper, status.account_id, at_time) if eligible?(status.proper)
end end
def add(status, _account_id, at_time = Time.now.utc) def add(status, _account_id, at_time = Time.now.utc)
@ -74,15 +75,13 @@ class Trends::Statuses < Trends::Base
private private
def eligible?(status) def eligible?(status)
original_status = status.proper status.public_visibility? && status.account.discoverable? && !status.account.silenced? && status.spoiler_text.blank? && !status.sensitive? && !status.reply?
original_status.public_visibility? &&
original_status.account.discoverable? && !original_status.account.silenced? &&
original_status.spoiler_text.blank? && !original_status.sensitive? && !original_status.reply?
end end
def calculate_scores(statuses, at_time) def calculate_scores(statuses, at_time)
redis.pipelined do global_items = []
locale_items = Hash.new { |h, key| h[key] = [] }
statuses.each do |status| statuses.each do |status|
expected = 1.0 expected = 1.0
observed = (status.reblogs_count + status.favourites_count).to_f observed = (status.reblogs_count + status.favourites_count).to_f
@ -97,30 +96,24 @@ class Trends::Statuses < Trends::Base
decaying_score = score * (0.5**((at_time.to_f - status.created_at.to_f) / options[:score_halflife].to_f)) decaying_score = score * (0.5**((at_time.to_f - status.created_at.to_f) / options[:score_halflife].to_f))
add_to_and_remove_from_subsets(status.id, decaying_score, { next unless decaying_score >= options[:decay_threshold]
all: true,
allowed: status.trendable? && status.account.discoverable?,
})
next unless valid_locale?(status.language) global_items << { score: decaying_score, item: status }
locale_items[status.language] << { account_id: status.account_id, score: decaying_score, item: status } if valid_locale?(status.language)
add_to_and_remove_from_subsets(status.id, decaying_score, {
"all:#{status.language}" => true,
"allowed:#{status.language}" => status.trendable? && status.account.discoverable?,
})
end end
trim_older_items replace_items('', global_items)
# Clean up localized sets by calculating the intersection with the main
# set. We do this instead of just deleting the localized sets to avoid
# having moments where the API returns empty results
Trends.available_locales.each do |locale| Trends.available_locales.each do |locale|
redis.zinterstore("#{key_prefix}:all:#{locale}", ["#{key_prefix}:all:#{locale}", "#{key_prefix}:all"], aggregate: 'max') replace_items(":#{locale}", locale_items[locale])
redis.zinterstore("#{key_prefix}:allowed:#{locale}", ["#{key_prefix}:allowed:#{locale}", "#{key_prefix}:allowed"], aggregate: 'max')
end end
end end
def filter_for_allowed_items(items)
# Show only one status per account, pick the one with the highest score
# that's also eligible to trend
items.group_by { |item| item[:account_id] }.values.filter_map { |account_items| account_items.select { |item| item[:item].trendable? && item[:item].account.discoverable? }.max_by { |item| item[:score] } }
end end
def would_be_trending?(id) def would_be_trending?(id)

View File

@ -8,6 +8,7 @@ class Trends::Tags < Trends::Base
review_threshold: 3, review_threshold: 3,
max_score_cooldown: 2.days.freeze, max_score_cooldown: 2.days.freeze,
max_score_halflife: 4.hours.freeze, max_score_halflife: 4.hours.freeze,
decay_threshold: 1,
} }
def register(status, at_time = Time.now.utc) def register(status, at_time = Time.now.utc)
@ -26,7 +27,6 @@ class Trends::Tags < Trends::Base
def refresh(at_time = Time.now.utc) def refresh(at_time = Time.now.utc)
tags = Tag.where(id: (recently_used_ids(at_time) + currently_trending_ids(false, -1)).uniq) tags = Tag.where(id: (recently_used_ids(at_time) + currently_trending_ids(false, -1)).uniq)
calculate_scores(tags, at_time) calculate_scores(tags, at_time)
trim_older_items
end end
def request_review def request_review
@ -53,6 +53,8 @@ class Trends::Tags < Trends::Base
private private
def calculate_scores(tags, at_time) def calculate_scores(tags, at_time)
items = []
tags.each do |tag| tags.each do |tag|
expected = tag.history.get(at_time - 1.day).accounts.to_f expected = tag.history.get(at_time - 1.day).accounts.to_f
expected = 1.0 if expected.zero? expected = 1.0 if expected.zero?
@ -79,11 +81,16 @@ class Trends::Tags < Trends::Base
decaying_score = max_score * (0.5**((at_time.to_f - max_time.to_f) / options[:max_score_halflife].to_f)) decaying_score = max_score * (0.5**((at_time.to_f - max_time.to_f) / options[:max_score_halflife].to_f))
add_to_and_remove_from_subsets(tag.id, decaying_score, { next unless decaying_score >= options[:decay_threshold]
all: true,
allowed: tag.trendable?, items << { score: decaying_score, item: tag }
})
end end
replace_items('', items)
end
def filter_for_allowed_items(items)
items.select { |item| item[:item].trendable? }
end end
def would_be_trending?(id) def would_be_trending?(id)