module Sequel module Plugins # The many_through_many plugin allow you to create an association to multiple objects using multiple join tables. # For example, assume the following associations: # # Artist.many_to_many :albums # Album.many_to_many :tags # # The many_through_many plugin would allow this: # # Artist.plugin :many_through_many # Artist.many_through_many :tags, [[:albums_artists, :artist_id, :album_id], [:albums, :id, :id], [:albums_tags, :album_id, :tag_id]] # # Which will give you the tags for all of the artist's albums. # # Let's break down the 2nd argument of the many_through_many call: # # [[:albums_artists, :artist_id, :album_id], # [:albums, :id, :id], # [:albums_tags, :album_id, :tag_id]] # # This argument is an array of arrays with three elements. Each entry in the main array represents a JOIN in SQL: # # * The first element in each array represents the name of the table to join. # * The second element in each array represents the column used to join to the previous table. # * The third element in each array represents the column used to join to the next table. # # So the "Artist.many_through_many :tags" is translated into something similar to: # # FROM artists # JOIN albums_artists ON (artists.id = albums_artists.artist_id) # JOIN albums ON (albums_artists.album_id = albums.id) # JOIN albums_tags ON (albums.id = albums_tag.album_id) # JOIN tags ON (albums_tags.tag_id = tags.id) # # The "artists.id" and "tags.id" criteria come from other association options (defaulting to the primary keys of the current and # associated tables), but hopefully you can see how each argument in the array is used in the JOIN clauses. # # Here are some more examples: # # # Same as Artist.many_to_many :albums # Artist.many_through_many :albums, [[:albums_artists, :artist_id, :album_id]] # # # All artists that are associated to any album that this artist is associated to # Artist.many_through_many :artists, [[:albums_artists, :artist_id, :album_id], [:albums, :id, :id], [:albums_artists, :album_id, :artist_id]] # # # All albums by artists that are associated to any album that this artist is associated to # Artist.many_through_many :artist_albums, [[:albums_artists, :artist_id, :album_id], [:albums, :id, :id], \ # [:albums_artists, :album_id, :artist_id], [:artists, :id, :id], [:albums_artists, :artist_id, :album_id]], \ # :class=>:Album # # # All tracks on albums by this artist # Artist.many_through_many :tracks, [[:albums_artists, :artist_id, :album_id], [:albums, :id, :id]], \ # :right_primary_key=>:album_id # # Often you don't want the current object to appear in the array of associated objects. This is easiest to handle via an :after_load hook: # # Artist.many_through_many :artists, [[:albums_artists, :artist_id, :album_id], [:albums, :id, :id], [:albums_artists, :album_id, :artist_id]], # :after_load=>proc{|artist, associated_artists| associated_artists.delete(artist)} # # You can also handle it by adding a dataset block that excludes the current record (so it won't be retrieved at all), but # that won't work when eagerly loading, which is why the :after_load proc is recommended instead. # # It's also common to not want duplicate records, in which case the :distinct option can be used: # # Artist.many_through_many :artists, [[:albums_artists, :artist_id, :album_id], [:albums, :id, :id], [:albums_artists, :album_id, :artist_id]], # :distinct=>true module ManyThroughMany # The AssociationReflection subclass for many_through_many associations. class ManyThroughManyAssociationReflection < Sequel::Model::Associations::ManyToManyAssociationReflection Sequel::Model::Associations::ASSOCIATION_TYPES[:many_through_many] = self # The default associated key alias(es) to use when eager loading # associations via eager. def default_associated_key_alias self[:uses_left_composite_keys] ? (0...self[:through].first[:left].length).map{|i| :"x_foreign_key_#{i}_x"} : :x_foreign_key_x end %w'associated_key_table predicate_key edges final_edge final_reverse_edge reverse_edges'.each do |meth| class_eval(<<-END, __FILE__, __LINE__+1) def #{meth} cached_fetch(:#{meth}){calculate_edges[:#{meth}]} end END end # Many through many associations don't have a reciprocal def reciprocal nil end private # Make sure to use unique table aliases when lazy loading or eager loading def calculate_reverse_edge_aliases(reverse_edges) aliases = [associated_class.table_name] reverse_edges.each do |e| table_alias = e[:table] if aliases.include?(table_alias) i = 0 table_alias = loop do ta = :"#{table_alias}_#{i}" break ta unless aliases.include?(ta) i += 1 end end aliases.push(e[:alias] = table_alias) end end # Transform the :through option into a list of edges and reverse edges to use to join tables when loading the association. def calculate_edges es = [{:left_table=>self[:model].table_name, :left_key=>self[:left_primary_key_column]}] self[:through].each do |t| es.last.merge!(:right_key=>t[:left], :right_table=>t[:table], :join_type=>t[:join_type]||self[:graph_join_type], :conditions=>(t[:conditions]||[]).to_a, :block=>t[:block]) es.last[:only_conditions] = t[:only_conditions] if t.include?(:only_conditions) es << {:left_table=>t[:table], :left_key=>t[:right]} end es.last.merge!(:right_key=>right_primary_key, :right_table=>associated_class.table_name) edges = es.map do |e| h = {:table=>e[:right_table], :left=>e[:left_key], :right=>e[:right_key], :conditions=>e[:conditions], :join_type=>e[:join_type], :block=>e[:block]} h[:only_conditions] = e[:only_conditions] if e.include?(:only_conditions) h end reverse_edges = es.reverse.map{|e| {:table=>e[:left_table], :left=>e[:left_key], :right=>e[:right_key]}} reverse_edges.pop calculate_reverse_edge_aliases(reverse_edges) final_reverse_edge = reverse_edges.pop final_reverse_alias = final_reverse_edge[:alias] h = {:final_edge=>edges.pop, :final_reverse_edge=>final_reverse_edge, :edges=>edges, :reverse_edges=>reverse_edges, :predicate_key=>qualify(final_reverse_alias, edges.first[:right]), :associated_key_table=>final_reverse_edge[:alias], } h.each{|k, v| cached_set(k, v)} h end end module ClassMethods # Create a many_through_many association. Arguments: # * name - Same as associate, the name of the association. # * through - The tables and keys to join between the current table and the associated table. # Must be an array, with elements that are either 3 element arrays, or hashes with keys :table, :left, and :right. # The required entries in the array/hash are: # :table (first array element) :: The name of the table to join. # :left (middle array element) :: The key joining the table to the previous table. Can use an # array of symbols for a composite key association. # :right (last array element) :: The key joining the table to the next table. Can use an # array of symbols for a composite key association. # If a hash is provided, the following keys are respected when using eager_graph: # :block :: A proc to use as the block argument to join. # :conditions :: Extra conditions to add to the JOIN ON clause. Must be a hash or array of two pairs. # :join_type :: The join type to use for the join, defaults to :left_outer. # :only_conditions :: Conditions to use for the join instead of the ones specified by the keys. # * opts - The options for the associaion. Takes the same options as many_to_many. def many_through_many(name, through, opts={}, &block) associate(:many_through_many, name, opts.merge(through.is_a?(Hash) ? through : {:through=>through}), &block) end private # Create the association methods and :eager_loader and :eager_grapher procs. def def_many_through_many(opts) name = opts[:name] model = self opts[:read_only] = true opts[:after_load].unshift(:array_uniq!) if opts[:uniq] opts[:cartesian_product_number] ||= 2 opts[:through] = opts[:through].map do |e| case e when Array raise(Error, "array elements of the through option/argument for many_through_many associations must have at least three elements") unless e.length == 3 {:table=>e[0], :left=>e[1], :right=>e[2]} when Hash raise(Error, "hash elements of the through option/argument for many_through_many associations must contain :table, :left, and :right keys") unless e[:table] && e[:left] && e[:right] e else raise(Error, "the through option/argument for many_through_many associations must be an enumerable of arrays or hashes") end end left_key = opts[:left_key] = opts[:through].first[:left] uses_lcks = opts[:uses_left_composite_keys] = left_key.is_a?(Array) left_keys = Array(left_key) left_pk = (opts[:left_primary_key] ||= self.primary_key) opts[:eager_loader_key] = left_pk unless opts.has_key?(:eager_loader_key) left_pks = opts[:left_primary_keys] = Array(left_pk) lpkc = opts[:left_primary_key_column] ||= left_pk opts[:left_primary_key_columns] ||= Array(lpkc) opts[:dataset] ||= lambda do ds = opts.associated_dataset opts.reverse_edges.each{|t| ds = ds.join(t[:table], Array(t[:left]).zip(Array(t[:right])), :table_alias=>t[:alias], :qualify=>:deep)} ft = opts.final_reverse_edge ds.join(ft[:table], Array(ft[:left]).zip(Array(ft[:right])) + opts.predicate_keys.zip(left_pks.map{|k| send(k)}), :table_alias=>ft[:alias], :qualify=>:deep) end left_key_alias = opts[:left_key_alias] ||= opts.default_associated_key_alias opts[:eager_loader] ||= lambda do |eo| h = eo[:id_map] rows = eo[:rows] rows.each{|object| object.associations[name] = []} ds = opts.associated_class opts.reverse_edges.each{|t| ds = ds.join(t[:table], Array(t[:left]).zip(Array(t[:right])), :table_alias=>t[:alias], :qualify=>:deep)} ft = opts.final_reverse_edge ds = ds.join(ft[:table], Array(ft[:left]).zip(Array(ft[:right])) + [[opts.predicate_key, h.keys]], :table_alias=>ft[:alias], :qualify=>:deep) ds = model.eager_loading_dataset(opts, ds, nil, eo[:associations], eo) case opts.eager_limit_strategy when :window_function delete_rn = true rn = ds.row_number_column ds = apply_window_function_eager_limit_strategy(ds, opts) when :correlated_subquery ds = apply_correlated_subquery_eager_limit_strategy(ds, opts) do |xds| dsa = ds.send(:dataset_alias, 2) opts.reverse_edges.each{|t| xds = xds.join(t[:table], Array(t[:left]).zip(Array(t[:right])), :table_alias=>t[:alias])} xds.join(ft[:table], Array(ft[:left]).zip(Array(ft[:right])) + left_keys.map{|k| [k, SQL::QualifiedIdentifier.new(ft[:table], k)]}, :table_alias=>dsa, :qualify=>:deep) end end ds.all do |assoc_record| assoc_record.values.delete(rn) if delete_rn hash_key = if uses_lcks left_key_alias.map{|k| assoc_record.values.delete(k)} else assoc_record.values.delete(left_key_alias) end next unless objects = h[hash_key] objects.each{|object| object.associations[name].push(assoc_record)} end if opts.eager_limit_strategy == :ruby limit, offset = opts.limit_and_offset rows.each{|o| o.associations[name] = o.associations[name].slice(offset||0, limit) || []} end end join_type = opts[:graph_join_type] select = opts[:graph_select] graph_block = opts[:graph_block] only_conditions = opts[:graph_only_conditions] use_only_conditions = opts.include?(:graph_only_conditions) conditions = opts[:graph_conditions] opts[:eager_grapher] ||= proc do |eo| ds = eo[:self] iq = eo[:implicit_qualifier] opts.edges.each do |t| ds = ds.graph(t[:table], t.fetch(:only_conditions, (Array(t[:right]).zip(Array(t[:left])) + t[:conditions])), :select=>false, :table_alias=>ds.unused_table_alias(t[:table]), :join_type=>t[:join_type], :qualify=>:deep, :implicit_qualifier=>iq, &t[:block]) iq = nil end fe = opts.final_edge ds.graph(opts.associated_class, use_only_conditions ? only_conditions : (Array(opts.right_primary_key).zip(Array(fe[:left])) + conditions), :select=>select, :table_alias=>eo[:table_alias], :qualify=>:deep, :join_type=>join_type, &graph_block) end def_association_dataset_methods(opts) end end module DatasetMethods private # Use a subquery to filter rows to those related to the given associated object def many_through_many_association_filter_expression(op, ref, obj) lpks = ref[:left_primary_key_columns] lpks = lpks.first if lpks.length == 1 lpks = ref.qualify(model.table_name, lpks) edges = ref.edges first, rest = edges.first, edges[1..-1] ds = model.db[first[:table]].select(*Array(ref.qualify(first[:table], first[:right]))) rest.each{|e| ds = ds.join(e[:table], e.fetch(:only_conditions, (Array(e[:right]).zip(Array(e[:left])) + e[:conditions])), :table_alias=>ds.unused_table_alias(e[:table]), :qualify=>:deep, &e[:block])} last_alias = if rest.empty? first[:table] else last_join = ds.opts[:join].last last_join.table_alias || last_join.table end meths = if obj.is_a?(Sequel::Dataset) ref.qualify(obj.model.table_name, ref.right_primary_keys) else ref.right_primary_key_methods end exp = association_filter_key_expression(ref.qualify(last_alias, Array(ref.final_edge[:left])), meths, obj) if exp == SQL::Constants::FALSE association_filter_handle_inversion(op, exp, Array(lpks)) else ds = ds.where(exp).exclude(SQL::BooleanExpression.from_value_pairs(ds.opts[:select].zip([]), :OR)) association_filter_handle_inversion(op, SQL::BooleanExpression.from_value_pairs(lpks=>ds), Array(lpks)) end end end end end end