export

MayBMS-tpch/uncertain-tpch/Rakefile

@@ -0,0 +1,59 @@
+
+$files = {
+  cust:     [ :c_acctbal, :c_address, :c_comment, :c_custkey, :c_mktsegment, :c_name, :c_nationkey, :c_phone ],
+  lineitem: [ :l_comment, :l_commitdate, :l_discount, :l_extendedprice, :l_linenumber, :l_linestatus, :l_orderkey, :l_partkey, :l_quantity, :l_receiptdate, :l_returnflag, :l_shipdate, :l_shipinstruct, :l_shipmode, :l_suppkey, :l_tax ],
+  nation:   [ :n_nationkey, :n_name, :n_regionkey, :n_comment ],
+  orders:   [ :o_clerk, :o_comment, :o_custkey, :o_orderdate, :o_orderkey, :o_orderpriority, :o_orderstatus, :o_shippriority, :o_totalprice ],
+  part:     [ :p_brand, :p_comment, :p_container, :p_mfgr, :p_name, :p_partkey, :p_retailprice, :p_size, :p_type ],
+  psupp:    [ :ps_availqty, :ps_comment, :ps_partkey, :ps_suppkey, :ps_supplycost ],
+  region:   [ :r_comment, :r_name, :r_regionkey ],
+  supp:     [ :s_acctbal, :s_address, :s_comment, :s_name, :s_nationkey, :s_phone, :s_suppkey ]
+}
+
+def init_data
+  # TUPLE_ID      ->     CONFLICT_GROUP  ->     WORLD        ->        FIELD    -> VALUE
+  Hash.new {|h,k| h[k] = Hash.new {|h,k| h[k] = Hash.new {|h,k| h[k] = Hash.new } } }
+end
+
+def load(data, table, field)
+  puts "Loading #{table}_#{field}.tbl"
+  File.open("input/#{table}_#{field}.tbl") do |f|
+    f.readlines.each do |l|
+      conflict_group, world, tuple_id, value = l.chomp.split(/\|/)
+      data[tuple_id.to_i][conflict_group.to_i][world.to_i][field] = value
+    end
+  end
+end
+
+def render(data)
+  puts "Rendering data..."
+  data.map do |tuple_id, conflict_groups| 
+    conflict_groups = conflict_groups.values.to_a
+    world_ids = conflict_groups.map { |worlds| worlds.keys }
+
+    # compute the cartesian product of the resulting key sets
+    world_groups = [ [nil] ]
+    world_ids.each { |id_set| world_groups = world_groups.product(id_set) }
+    world_groups = world_groups.map { |x| x.flatten[1..-1] }
+
+    # render the corresponding tuple
+    world_groups.map do |per_group_world_ids| 
+      conflict_groups.zip(per_group_world_ids).map { |worlds, world_id| worlds[world_id].to_a }.flatten(1)
+    end.map { |tuple| [tuple_id, tuple.to_h] }
+  end.flatten(1)
+end
+
+$files.each do |table, fields|
+  file "output/#{table}.tbl" do
+    data = init_data
+    fields.each { |field| load(data, table, field) }
+    File.open("output/#{table}.tbl", "w+") do |f|
+      f.puts "tid|#{fields.join("|")}"
+      render(data).each do |tid, tuple| 
+        f.puts "#{tid}|#{fields.map {|f| tuple[f]}.join("|")}"
+      end
+    end
+  end
+
+  task table => "output/#{table}.tbl"
+end