ファイル
btrc-hub/backend/app/services/counts_axes.rb
T
2026-11-08 01:56:45 +09:00

183 行
7.1 KiB
Ruby

class CountsAxes
MAX_AXES = 8
MASKED_TEXT = '********'.freeze
MASKED_VALUE = '__counts_private__'.freeze
PUBLIC_FIELDS = ['resource_id', 'version_no', 'event_type', 'user',
'created_at'].freeze
class TooManyAxes < ArgumentError; end
class PrivateCondition < ArgumentError; end
GRAINS = ['year', 'quarter', 'month', 'week', 'day', 'hour'].freeze
def self.private_field?(field)
!PUBLIC_FIELDS.include?(field) && field != 'user.id'
end
def self.visible_ids_sql(definition)
"SELECT counts_visible.id FROM #{ definition.table } counts_visible WHERE " +
HistoryAccess.sql(definition.model, 'counts_visible')
end
def self.visibility_sql(definition, field)
return 'TRUE' if HistoryAccess.admin?
source = visible_ids_sql(definition)
visible = "v.id IN (#{ source })"
return visible unless field == 'changed_field'
"(#{ visible }) AND (previous.id IS NULL OR previous.id IN (#{ source }))"
end
def self.masked_sql(expression, definition, field)
return expression if HistoryAccess.admin? || !private_field?(field)
visible = visibility_sql(definition, field)
quoted = ApplicationRecord.connection.quote(MASKED_VALUE)
"CASE WHEN #{ visible } THEN #{ expression } ELSE #{ quoted } END"
end
def self.time_expression(expression, grain, timezone)
raise ArgumentError unless GRAINS.include?(grain)
connection = ApplicationRecord.connection
# Use the installed TZInfo database, including DST, rather than MySQL timezone tables.
first = Time.utc(1900)
last = Time.utc(2101)
offset = timezone.period_for_utc(first).utc_total_offset
clauses = timezone.transitions_up_to(last, first).filter_map do |transition|
next if transition.offset.utc_total_offset == offset
previous = offset
offset = transition.offset.utc_total_offset
"WHEN #{ expression } < #{ connection.quote(transition.at.to_time) } THEN #{ previous }"
end
seconds = clauses.empty? ? offset.to_s : "CASE #{ clauses.join(' ') } ELSE #{ offset } END"
local = "DATE_ADD(#{ expression }, INTERVAL (#{ seconds }) SECOND)"
case grain
when 'year' then "DATE_FORMAT(#{ local }, '%Y')"
when 'quarter' then "CONCAT(YEAR(#{ local }), '-Q', QUARTER(#{ local }))"
when 'month' then "DATE_FORMAT(#{ local }, '%Y-%m')"
when 'week' then "DATE_FORMAT(#{ local }, '%x-W%v')"
when 'day' then "DATE_FORMAT(#{ local }, '%Y-%m-%d')"
else "DATE_FORMAT(#{ local }, '%Y-%m-%d %H:00')"
end
end
attr_reader :rows, :columns, :all
def initialize(population)
@population = population
@definition = population.definition
@rows = parse(population.params.fetch('rows', ''))
@columns = parse(population.params.fetch('cols', ''))
@all = rows + columns
raise TooManyAxes if all.length > MAX_AXES
raise ArgumentError if all.map { _1['field'] }.uniq.length != all.length
end
def expressions
@expressions ||= all.map do |axis|
field = axis.fetch('field')
expression = case field
when 'tags'
"CASE WHEN tag_entry.tag_id IS NOT NULL " +
"THEN CONCAT('internal:', tag_entry.tag_id) " +
"ELSE CONCAT('external:', tag_entry.external_tag_id) END"
when 'changed_field' then 'changed.field_key'
else
expression = @definition.expression(field)
@definition.fields[field] == :datetime ? self.class.time_expression(
expression, axis.fetch('grain', 'month'), @population.timezone) : expression
end
if !HistoryAccess.admin? && self.class.private_field?(field)
expression = self.class.masked_sql(
"CAST((#{ expression }) AS CHAR)", @definition, field)
end
expression
end
end
def grouped_sql(scope, indexes = all.each_index.to_a)
keys = indexes.map { "a#{ _1 }" }
selected = indexes.map { "#{ expressions[_1] } AS a#{ _1 }" }
identities = ['v.id']
if rows.any? { _1['field'] == 'tags' }
identities << self.class.masked_sql(
"COALESCE(CONCAT('internal:', tag_entry.tag_id), " +
"CONCAT('external:', tag_entry.external_tag_id))", @definition, 'tags')
if rows.any? { _1['field'] == 'changed_field' }
identities << self.class.masked_sql('changed.field_key', @definition, 'changed_field')
end
end
members = identities.each_index.map { "counts_member_#{ _1 }" }
selected += identities.each_with_index.map { |expression, index|
"#{ expression } AS #{ members[index] }"
}
# Evaluate correlated visibility subqueries before grouping. LIMIT prevents
# derived-table merging, so ONLY_FULL_GROUP_BY sees ordinary projected columns.
projection = scope.except(:select, :order).select(Arel.sql(selected.join(', ')))
.limit(18_446_744_073_709_551_615).to_sql
fields = keys + ["COUNT(DISTINCT #{ members.join(', ') }) AS count"]
grouping = keys.empty? ? '' : " GROUP BY #{ keys.join(', ') }"
"SELECT #{ fields.join(', ') } FROM (#{ projection }) counts_values#{ grouping }"
end
def joins(fields = all.map { _1['field'] })
result = ''
if fields.include?('tags')
document = 'CAST(v.tags_json AS JSON)'
unless HistoryAccess.admin?
visible = self.class.visible_ids_sql(@definition)
result += " LEFT JOIN (#{ visible }) counts_tags_visible " +
'ON counts_tags_visible.id = v.id '
document = "CASE WHEN counts_tags_visible.id IS NOT NULL " +
"THEN #{ document } ELSE JSON_ARRAY(" +
"JSON_OBJECT('tag_id', 0)) END"
end
result += <<~SQL.squish
JOIN JSON_TABLE(#{ document }, '$[*]' COLUMNS (
tag_id BIGINT PATH '$.tag_id' NULL ON EMPTY,
external_tag_id BIGINT PATH '$.external_tag_id' NULL ON EMPTY)) tag_entry
SQL
end
if fields.include?('changed_field')
fields = @definition.changes.map do |field|
"SELECT #{ ApplicationRecord.connection.quote(field) } AS field_key"
end.join(' UNION ALL ')
condition = @definition.changes.map do |field|
"(changed.field_key = '#{ field }' AND #{ @population.changed_sql(field) })"
end.join(' OR ')
unless HistoryAccess.admin?
visible = self.class.visibility_sql(@definition, 'changed_field')
marker = ApplicationRecord.connection.quote(MASKED_VALUE)
fields += " UNION ALL SELECT #{ marker } AS field_key"
condition = "((#{ visible }) AND (#{ condition })) OR " +
"(NOT (#{ visible }) AND changed.field_key = #{ marker })"
end
result += " JOIN (#{ fields }) changed ON (#{ condition })"
end
result
end
private
def parse(value)
axes = QueryState.axes(value)
raise ArgumentError unless axes.is_a?(Array)
axes.each do |axis|
raise ArgumentError unless axis.is_a?(Hash) && (axis.keys - ['field', 'grain']).empty?
raise ArgumentError unless @definition.axes(@population.history).include?(axis['field'])
if @definition.fields[axis['field']] == :datetime
raise ArgumentError unless GRAINS.include?(axis.fetch('grain', 'month'))
axis['grain'] ||= 'month'
elsif axis.key?('grain')
raise ArgumentError
end
end
axes
end
end