Example #1
0
 def test_count_all(self):
     """
     Test simply counting everything.
     """
     query = tree([Count(), [MemScan(_RECORDS)]])
     result = list(execute(query))
     self.assertListEqual(result, [tuple([10])])
Example #2
0
 def test_distinct_unsorted(self):
     """
     Distinct is not expected to handle out-of-order repeated values.
     """
     records = [tuple([n]) for n in (0, 1, 1, 0, 0, 1, 1, 1)]
     query = tree([Distinct(), [MemScan(records)]])
     result = list(execute(query))
     self.assertListEqual(result, [tuple([n]) for n in (0, 1, 0, 1)])
Example #3
0
 def test_distinct_sorted(self):
     """
     Typical case: expect input to be sorted.
     """
     records = [tuple([n]) for n in (0, 0, 1, 2, 2, 2, 3, 3)]
     query = tree([Distinct(), [MemScan(records)]])
     result = list(execute(query))
     self.assertListEqual(result, [tuple([n]) for n in (0, 1, 2, 3)])
Example #4
0
 def test_average_all(self):
     """
     Find the average of all numbers 0..9
     """
     query = tree(
         [Average(), [Projection(lambda r: [r[0]]), [MemScan(_RECORDS)]]])
     result = list(execute(query))
     self.assertListEqual(result, [tuple([4.5])])
Example #5
0
 def test_count_selected(self):
     """
     Test counting the output of a selection.
     """
     for p in _PREDICATES:
         query = tree([Count(), [Selection(p), [MemScan(_RECORDS)]]])
         result = list(execute(query))
         expected = [tuple([len([r for r in _RECORDS if p(r)])])]
         self.assertListEqual(result, expected)
 def test_cartesian_product(self):
     """
     When the theta function simply returns True, we achieve cross product.
     """
     query = tree(
         [NestedLoopsJoin(lambda r, s: True),
             [MemScan(self.R_RECORDS)],
             [MemScan(self.S_RECORDS)]])
     result = list(execute(query))
     expected = [r + s for r in self.R_RECORDS for s in self.S_RECORDS]
     self.assertListEqual(result, expected)
 def test_self_join(self):
     """
     It's fine for a table to be joined to itself.
     """
     query = tree(
         [NestedLoopsJoin(lambda r, s: True),
             [MemScan(self.R_RECORDS)],
             [MemScan(self.R_RECORDS)]])
     result = list(execute(query))
     expected = [r + s for r in self.R_RECORDS for s in self.R_RECORDS]
     self.assertListEqual(result, expected)
Example #8
0
 def test_project_various(self):
     mapping_functions = [
         lambda r: [r[1]],  # project to a single column
         lambda r: [r[1], r[0]],  # switch columns
         lambda r: [r[0] + r[1]],  # sums of fields
         lambda r: [r[1] > 8],  # results of boolean operations
     ]
     records = [(n, n * n) for n in range(10)]
     for f in mapping_functions:
         query = tree([Projection(f), [MemScan(records)]])
         result = list(execute(query))
         expected = [tuple(f(r)) for r in records]
         self.assertListEqual(result, expected)
Example #9
0
 def test_select_by_id(self):
     """
     What is the name of the movie with id 5000?
     """
     query = tree([
         Projection(lambda r: tuple([r.name])),
         [
             Selection(lambda r: r.id == '5000'),
             [FileScan(movies_file, MovieRecord)]
         ]
     ])
     self.assertListEqual(['Medium Cool (1969)'],
                          [r[0] for r in execute(query)])
Example #10
0
 def test_select_various_predicates(self):
     predicates = [
         lambda r: r[0] % 2 == 1,  # odd values
         lambda r: r[1] % 2 == 1,  # odd squares
         lambda r: r[0] + r[1] == 30,  # just (5, 25)
         lambda r: r[1] == 'Reginald',  # no records
     ]
     records = [(n, n * n) for n in range(10)]
     for p in predicates:
         query = tree([Selection(p), [MemScan(records)]])
         result = list(execute(query))
         expected = [r for r in records if p(r)]
         self.assertListEqual(result, expected)
Example #11
0
 def test_average_selected(self):
     """
     Find the average of only filtered numbers in the range 0..9
     """
     query = tree([
         Average(),
         [
             Projection(lambda r: [r[0]]),
             [Selection(lambda r: r[0] % 2 == 1), [MemScan(_RECORDS)]]
         ]
     ])
     result = list(execute(query))
     self.assertListEqual(result, [tuple([5])])
Example #12
0
 def test_equijoin(self):
     """
     The join condition can be that the value in one field equals another.
     """
     query = tree(
         [NestedLoopsJoin(lambda r, s: r[1] == s[0]),
             [MemScan(self.R_RECORDS)],
             [MemScan(self.S_RECORDS)]])
     result = list(execute(query))
     expected = [
         (0, 'a', 'a', 'apple'),
         (1, 'b', 'b', 'banana'),
         (1, 'b', 'b', 'burger')
     ]
     self.assertListEqual(result, expected)
Example #13
0
 def test_equijoin(self):
     """
     The a simple join by field field value.
     """
     query = tree([
         MergeJoin(lambda r: r[1], lambda s: s[0]),
         [MemScan(self.R_RECORDS)], [MemScan(self.S_RECORDS)]
     ])
     result = list(execute(query))
     expected = [
         (0, 'a', 'a', 'apple'),
         (1, 'b', 'b', 'banana'),
         (1, 'b', 'b', 'burger'),
         (3, 'd', 'd', 'domato'),
     ]
     self.assertListEqual(result, expected)
Example #14
0
    def test_average_rating(self):
        """
        What is the average rating for movie with id 5000?

        WARNING: super slow
        """
        query = tree([
            Average(),
            [
                Projection(lambda r: tuple([float(r.rating)])),
                [
                    Selection(lambda r: r.movie_id == '5000'),
                    [FileScan(ratings_file, RatingRecord)]
                ]
            ]
        ])
        self.assertAlmostEqual(3.5, next(execute(query))[0], places=2)
Example #15
0
 def test_inequality(self):
     """
     The join condition can be an inequality.
     """
     query = tree(
         [NestedLoopsJoin(lambda r, s: r[0] > s[0]),
             [MemScan(self.R_RECORDS)],
             [MemScan(self.R_RECORDS)]])
     result = list(execute(query))
     expected = [
         (1, 'b', 0, 'a'),
         (1, 'c', 0, 'a'),
         (2, 'c', 0, 'a'),
         (2, 'c', 1, 'b'),
         (2, 'c', 1, 'c'),
     ]
     self.assertListEqual(result, expected)
Example #16
0
 def test_self_join(self):
     """
     It's fine for a table to be joined to itself.
     """
     query = tree([
         MergeJoin(lambda r: r[0], lambda s: s[0]),
         [MemScan(self.S_RECORDS)], [MemScan(self.S_RECORDS)]
     ])
     result = list(execute(query))
     expected = [
         ('a', 'apple', 'a', 'apple'),
         ('b', 'banana', 'b', 'banana'),
         ('b', 'banana', 'b', 'burger'),
         ('b', 'burger', 'b', 'banana'),
         ('b', 'burger', 'b', 'burger'),
         ('d', 'domato', 'd', 'domato'),
     ]
     self.assertListEqual(result, expected)
Example #17
0
 def test_three_way(self):
     """
     The input to a join can be a join.
     """
     query = tree(
         [NestedLoopsJoin(lambda r, s: r[3] == s[0]),
             [NestedLoopsJoin(lambda r, s: r[0] > s[0]),
                 [MemScan(self.R_RECORDS)],
                 [MemScan(self.R_RECORDS)]],
             [MemScan(self.S_RECORDS)]])
     result = list(execute(query))
     expected = [
         (1, 'b', 0, 'a', 'a', 'apple'),
         (1, 'c', 0, 'a', 'a', 'apple'),
         (2, 'c', 0, 'a', 'a', 'apple'),
         (2, 'c', 1, 'b', 'b', 'banana'),
         (2, 'c', 1, 'b', 'b', 'burger'),
     ]
     self.assertListEqual(result, expected)
Example #18
0
    def test_count_rated_movies(self):
        """
        How many distinct movies have a rating?

        WARNING: super slow
        """
        query = tree([
            Count(),
            [
                Distinct(),
                [
                    Sort(lambda r: r[0]),
                    [
                        Projection(lambda r: tuple([r.movie_id])),
                        [FileScan(ratings_file, RatingRecord)]
                    ]
                ]
            ]
        ])
        # TODO: shouldn't this be 26744?
        self.assertEqual(26745, next(execute(query))[0])
Example #19
0
 def test_three_way(self):
     """
     The input to a join can be a join.
     """
     query = tree([
         MergeJoin(lambda r: r[1], lambda s: s[0]),
         [MemScan(self.R_RECORDS)],
         [
             MergeJoin(lambda r: r[0], lambda s: s[0]),
             [MemScan(self.S_RECORDS)], [MemScan(self.S_RECORDS)]
         ]
     ])
     result = list(execute(query))
     expected = [
         (0, 'a', 'a', 'apple', 'a', 'apple'),
         (1, 'b', 'b', 'banana', 'b', 'banana'),
         (1, 'b', 'b', 'banana', 'b', 'burger'),
         (1, 'b', 'b', 'burger', 'b', 'banana'),
         (1, 'b', 'b', 'burger', 'b', 'burger'),
         (3, 'd', 'd', 'domato', 'd', 'domato'),
     ]
     self.assertListEqual(result, expected)