Merge pull request #328 from DevelopingCoder/set_group_cnt

Added lset group_count operator

lib/bud/lattice-lib.rb

@@ -248,6 +248,26 @@ def inspect
     Bud::BoolLattice.new(@v.member? i)
   end
 
+  monotone :group_count do |key_cols|
+    # Assume key_cols for now gives indices
+    rv = Hash.new(Bud::MaxLattice.new(0))
+    @v.each do |t|
+      unless t.class == Array
+        raise Bud::TypeError, "group_count only works if lset elements are type Array"
+      end
+
+      key = []
+      key_cols.each do |ind|
+        if ind >= t.length
+          raise Bud::Error, "lset element in group_count does not have column index #{ind}"
+        end
+        key << t[ind]
+      end
+      rv[key] += 1
+    end
+    Bud::MapLattice.new(rv)
+  end
+
   morph :pro do |&blk|
     # We don't use Set#map, since it returns an Array (ugh).
     rv = Set.new

test/tc_lattice.rb

@@ -942,6 +942,34 @@ class SetProduct
   end
 end
 
+class SetSimpleGroupCnt
+  # Groups by first column
+  include Bud
+
+  state do
+    lset :s1
+    lmap :res
+  end
+
+  bloom do
+    res <= s1.group_count([0])
+  end
+end
+
+class SetMultipleGroupCnt
+  # Groups by first column and third column
+  include Bud
+
+  state do
+    lset :s1
+    lmap :res
+  end
+
+  bloom do
+    res <= s1.group_count([0, 2])
+  end
+end
+
 class SetEqjoin
   include Bud
 
@@ -1141,6 +1169,67 @@ def test_set_method_compose
     assert_equal(true, i.done.current_value.reveal)
   end
 
+  def test_set_simple_groupcnt
+    i = SetSimpleGroupCnt.new
+    i.tick
+    expected = Hash.new(Bud::MaxLattice.new(0))
+    assert_equal(expected, i.res.current_value.reveal)
+
+    i.s1 <+ [['a1', 1]]
+    expected[['a1']] = Bud::MaxLattice.new(1)
+    i.tick
+    assert_equal(expected, i.res.current_value.reveal)
+
+    i.s1 <+ [['a2', 2]]
+    expected[['a2']] = Bud::MaxLattice.new(1)
+    i.tick
+    assert_equal(expected, i.res.current_value.reveal)
+
+    i.s1 <+ [['a1', 3], ['a1', 4]]
+    expected[['a1']] = Bud::MaxLattice.new(3)
+    i.tick
+    assert_equal(expected, i.res.current_value.reveal)
+  end
+
+  def test_set_multiple_groupcnt
+    i = SetMultipleGroupCnt.new
+    i.tick
+    expected = Hash.new(Bud::MaxLattice.new(0))
+    assert_equal(expected, i.res.current_value.reveal)
+
+    i.s1 <+ [['a1', 'b1', 'c1', 1]]
+    expected[['a1', 'c1']] = Bud::MaxLattice.new(1)
+    i.tick
+    assert_equal(expected, i.res.current_value.reveal)
+
+    i.s1 <+ [['a2', 'b2', 'c2', 2]]
+    expected[['a2', 'c2']] = Bud::MaxLattice.new(1)
+    i.tick
+    assert_equal(expected, i.res.current_value.reveal)
+
+    i.s1 <+ [['a1', 'b3', 'c1', 3], ['a1', 'b4', 'c1', 4]]
+    expected[['a1', 'c1']] = Bud::MaxLattice.new(3)
+    i.tick
+    assert_equal(expected, i.res.current_value.reveal)
+  end
+
+  def test_set_error_groupcnt
+    i = SetSimpleGroupCnt.new
+    i.tick
+
+    i.s1 <+ ['a1']
+    assert_raises(Bud::TypeError) do
+      i.tick
+    end
+
+    i = SetSimpleGroupCnt.new
+    i.tick
+    i.s1 <+ [['a1', 1], []]
+    assert_raises(Bud::Error) do
+      i.tick
+    end
+  end
+
   def test_set_product
     i = SetProduct.new
     i.tick