import networkx as nx
from networkx.algorithms.approximation import average_clustering

# This approximation has to be be exact in regular graphs
# with no triangles or with all possible triangles.


def test_petersen():
    # Actual coefficient is 0
    G = nx.petersen_graph()
    assert (average_clustering(G, trials=int(len(G) / 2)) ==
                 nx.average_clustering(G))


def test_petersen_seed():
    # Actual coefficient is 0
    G = nx.petersen_graph()
    assert (average_clustering(G, trials=int(len(G) / 2), seed=1) ==
                 nx.average_clustering(G))


def test_tetrahedral():
    # Actual coefficient is 1
    G = nx.tetrahedral_graph()
    assert (average_clustering(G, trials=int(len(G) / 2)) ==
                 nx.average_clustering(G))


def test_dodecahedral():
    # Actual coefficient is 0
    G = nx.dodecahedral_graph()
    assert (average_clustering(G, trials=int(len(G) / 2)) ==
                 nx.average_clustering(G))


def test_empty():
    G = nx.empty_graph(5)
    assert average_clustering(G, trials=int(len(G) / 2)) == 0


def test_complete():
    G = nx.complete_graph(5)
    assert average_clustering(G, trials=int(len(G) / 2)) == 1
    G = nx.complete_graph(7)
    assert average_clustering(G, trials=int(len(G) / 2)) == 1