test_recommend.py

import recommend

import networkx

import random
import unittest

"""
Some non-extensive tests for recommend.py
"""

def fresh_random(mth):
    def decorated_mth(*args, **kwds):
        random.seed("random seed")
        return mth(*args, **kwds)
    return decorated_mth

class TestRecommend(unittest.TestCase):
    def setUp(self):
        default_graph_edges = [
            (1, 2, {"weight": 2.0}),
            (2, 1, {"weight": 1.0}),
            (1, 3, {"weight": 3.0}),
            (3, 4, {"weight": 1.0}),
            (3, 5, {"weight": 1.0}),
            (3, 6, {"weight": 2.0}),
            (6, 2, {"weight": 1.0}),
            (2, 7, {"weight": 1.0}),
            (7, 8, {"weight": 1.0}),
            (7, 9, {"weight": 1.0}),
            (7, 10, {"weight": 1.0}),
            (10, 11, {"weight": 1.0})
        ]

        recommend._graph = networkx.DiGraph()
        recommend._graph.add_edges_from(default_graph_edges)

    def test_get_min_weight_neighbor(self):
        self.assertEquals(recommend._get_min_weight_neighbor(2, "in"), 6)
        self.assertEquals(recommend._get_min_weight_neighbor(5, "in"), 3)
        self.assertEquals(recommend._get_min_weight_neighbor(6, "out"), 2)
        self.assertEquals(recommend._get_min_weight_neighbor(3, "out"), 4)
        self.assertRaises(ValueError, recommend._get_min_weight_neighbor, 9, "out")

    def test_get_max_weight_neighbor(self):
        self.assertEquals(recommend._get_max_weight_neighbor(1, "in"), 2)
        self.assertEquals(recommend._get_max_weight_neighbor(3, "in"), 1)
        self.assertEquals(recommend._get_max_weight_neighbor(6, "out"), 2)
        self.assertEquals(recommend._get_max_weight_neighbor(1, "out"), 3)
        self.assertRaises(ValueError, recommend._get_max_weight_neighbor, 9, "out")

    @fresh_random
    def test_compute_candidates(self):
        # no climbing, last item is the random choice
        recommend.MIN_CANDIDATES = 0
        self.assertEquals(recommend._compute_candidates(1, 1), {2: 2.0, 3: 3.0, 4: 2.0})
        self.assertEquals(recommend._compute_candidates(1, 2), {2: 2.0, 3: 3.0, 4: 0.5, 5: 0.5, 6: 1.0, 7: 0.5, 8: 0.5})
        self.assertEquals(recommend._compute_candidates(1, 3), {2: 2.0, 3: 3.0, 4: 0.5, 5: 0.5, 6: 1.0, 7: 0.5, 8: 1.0 / 3, 9: 1.0 / 3, 10: 1.0 / 3, 11: 1.0 / 3})
        self.assertEquals(recommend._compute_candidates(6, 1), {2: 1.0, 8: 1.0})
        self.assertEquals(recommend._compute_candidates(6, 2), {1: 0.5, 2: 1.0, 7: 0.5, 11: 0.5})

        # exercise climbing, last two items are the random choices for the
        # parent and the node
        recommend.MIN_CANDIDATES = 3
        self.assertEquals(recommend._compute_candidates(5, 1), {4: 1.0, 6: 2.0, 7: 1.0, 11: 1.0})
        self.assertEquals(recommend._compute_candidates(11, 2), {3: 1.0, 7: 1.0})

    @fresh_random
    def test_next(self):
        recommend.MIN_CANDIDATES = 0
        self.assertEquals(recommend.next(1, 2), 3)

        # add node to the graph, return random node
        self.assertEquals(recommend.next(20), 11)
        self.assertTrue(recommend._graph.has_node(20))

    @fresh_random
    def test_min_candidates(self):
        recommend.MIN_CANDIDATES = float("inf")
        self.assertEquals(recommend.next(10, 3), None)

        recommend.MIN_CANDIDATES = 0
        self.assertEquals(recommend.next(10, 3), 11)

    def test_return_default(self):
        recommend.MIN_CANDIDATES = float("inf")

        default = object()
        self.assertIs(recommend.next(10, 3, default), default)

    def test_max_out_degree(self):
        recommend.MAX_OUT_DEGREE = 1

        recommend.positive(1, 4)
        self.assertTrue(recommend._graph.has_edge(1, 4))
        self.assertFalse(recommend._graph.has_edge(1, 3))
        self.assertFalse(recommend._graph.has_edge(1, 2))

    def test_max_in_degree(self):
        recommend.MAX_IN_DEGREE = 1

        recommend.positive(5, 2)
        self.assertTrue(recommend._graph.has_edge(5, 2))
        self.assertFalse(recommend._graph.has_edge(1, 2))
        self.assertFalse(recommend._graph.has_edge(6, 2))

if __name__ == "__main__":
    recommend.GRAPH_DOT_FILE = None
    recommend.GRAPH_PERSISTENCE_FILE = None
    recommend.MIN_GRAPH_SIZE = 0

    unittest.main()