def test_invalidate_cache_midchain(flowmachine_connect):
"""
Test that invalidating a query in the middle of a chain drops the
top of the chain and this link, but not the bottom.
"""
dl1 = daily_location("2016-01-01")
dl1.store().result()
hl1 = HomeLocation(daily_location("2016-01-01"),
daily_location("2016-01-02"))
hl1.store().result()
hl2 = HomeLocation(daily_location("2016-01-03"),
daily_location("2016-01-04"))
flow = Flows(hl1, hl2)
flow.store().result()
hl1.invalidate_db_cache()
assert dl1.is_stored
assert not hl1.is_stored
assert not flow.is_stored
in_cache = bool(
flowmachine_connect.fetch(
f"SELECT * FROM cache.cached WHERE query_id='{dl1.md5}'"))
assert in_cache
in_cache = bool(
flowmachine_connect.fetch(
f"SELECT * FROM cache.cached WHERE query_id='{hl1.md5}'"))
assert not in_cache
in_cache = bool(
flowmachine_connect.fetch(
f"SELECT * FROM cache.cached WHERE query_id='{flow.md5}'"))
assert not in_cache
has_deps = bool(
flowmachine_connect.fetch("SELECT * FROM cache.dependencies"))
assert has_deps # Daily location deps should remain
def test_invalidate_cascade(flowmachine_connect):
"""
Test that invalidation does not cascade if cascade=False.
"""
dl1 = daily_location("2016-01-01")
dl1.store().result()
hl1 = HomeLocation(daily_location("2016-01-01"),
daily_location("2016-01-02"))
hl1.store().result()
hl2 = HomeLocation(daily_location("2016-01-03"),
daily_location("2016-01-04"))
flow = Flows(hl1, hl2)
flow.store().result()
dl1.invalidate_db_cache(cascade=False)
assert not dl1.is_stored
assert hl1.is_stored
assert flow.is_stored
in_cache = bool(
flowmachine_connect.fetch(
f"SELECT * FROM cache.cached WHERE query_id='{dl1.md5}'"))
assert not in_cache
in_cache = bool(
flowmachine_connect.fetch(
f"SELECT * FROM cache.cached WHERE query_id='{hl1.md5}'"))
assert in_cache
has_deps = bool(
flowmachine_connect.fetch("SELECT * FROM cache.dependencies"))
assert has_deps
def test_invalidate_cascade(flowmachine_connect):
"""
Test that invalidation does not cascade if cascade=False.
"""
dl1 = daily_location("2016-01-01")
dl1.store().result()
hl1 = ModalLocation(daily_location("2016-01-01"),
daily_location("2016-01-02"))
hl1.store().result()
hl2 = ModalLocation(daily_location("2016-01-03"),
daily_location("2016-01-04"))
flow = Flows(hl1, hl2)
flow.store().result()
assert dl1.is_stored
assert hl1.is_stored
assert flow.is_stored
dl1.invalidate_db_cache(cascade=False)
assert not dl1.is_stored
assert hl1.is_stored
assert flow.is_stored
assert not cache_table_exists(get_db(), dl1.query_id)
assert cache_table_exists(get_db(), hl1.query_id)
has_deps = bool(get_db().fetch("SELECT * FROM cache.dependencies"))
assert has_deps
def test_invalidate_cache_midchain(flowmachine_connect):
"""
Test that invalidating a query in the middle of a chain drops the
top of the chain and this link, but not the bottom.
"""
dl1 = daily_location("2016-01-01")
dl1.store().result()
hl1 = ModalLocation(daily_location("2016-01-01"),
daily_location("2016-01-02"))
hl1.store().result()
hl2 = ModalLocation(daily_location("2016-01-03"),
daily_location("2016-01-04"))
flow = Flows(hl1, hl2)
flow.store().result()
assert dl1.is_stored
assert hl1.is_stored
assert flow.is_stored
hl1.invalidate_db_cache()
assert dl1.is_stored
assert not hl1.is_stored
assert not flow.is_stored
assert cache_table_exists(get_db(), dl1.query_id)
assert not cache_table_exists(get_db(), hl1.query_id)
assert not cache_table_exists(get_db(), flow.query_id)
has_deps = bool(get_db().fetch("SELECT * FROM cache.dependencies"))
assert has_deps # Daily location deps should remain
def setUp(self):
self.dl1 = daily_location("2016-01-01")
self.dl2 = daily_location("2016-01-02")
self.dl3 = daily_location("2016-01-07")
self.flowA = Flows(self.dl1, self.dl2)
self.flowB = Flows(self.dl1, self.dl3)
self.relfl = self.flowA - self.flowB
self.df_rel = self.relfl.get_dataframe()
def test_gets_mixins():
"""
Test that a random sample gets applicable mixins.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
flow = Flows(dl1, dl2)
assert isinstance(flow.random_sample(size=10), GraphMixin)
def flows(flowmachine_connect):
"""Fixture providing two flows."""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
dl3 = daily_location("2016-01-07")
flow_a = Flows(dl1, dl2)
flow_b = Flows(dl1, dl3)
yield flow_a, flow_b
def test_flow_math_store(op, exemplar_level_param, flowmachine_connect):
"""
Storing works for flows added together at all levels
"""
dl1 = daily_location("2016-01-01", **exemplar_level_param)
dl2 = daily_location("2016-01-02", **exemplar_level_param)
fl = op(Flows(dl1, dl2), Flows(dl1, dl2))
fl.store().result()
assert flowmachine_connect.has_table(*fl.table_name.split(".")[::-1])
def test_outflow_value(get_dataframe):
"""
One of the values for the outflows.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
flow = Flows(dl1, dl2)
outflow = flow.outflow()
df = get_dataframe(outflow)
assert df.set_index("name_from").ix["Sankhuwasabha"][0] == 24
def test_undirected():
"""
to_networkx() raises a warning if using duplicate edges are detected.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
flow = Flows(dl1, dl2)
with pytest.warns(UserWarning):
graph = flow.to_networkx(directed_graph=False)
assert "Sankhuwasabha" in graph.neighbors("Arghakhanchi")
def test_inflow_value(get_dataframe):
"""
One of the values for the outflows.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
flow = Flows(dl1, dl2)
inflow = flow.inflow()
df = get_dataframe(inflow)
assert df.set_index("name_to").ix["Okhaldhunga"][0] == 20
def test_inflow_value(get_dataframe):
"""
One of the values for the outflows.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
flow = Flows(dl1, dl2)
inflow = flow.inflow()
df = get_dataframe(inflow)
assert df.set_index("pcod_to").loc["524 1 03 13"][0] == 20
def test_nx_object():
"""
to_networkx() creates a networkx.Graph() object.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
flow = Flows(dl1, dl2)
graph = flow.to_networkx()
assert graph.has_node("524 3 09 50")
assert "524 5 14 73" in graph.neighbors("524 3 09 50")
assert "524 1 02 09" not in graph.neighbors("524 3 09 50")
def test_nx_object():
"""
to_networkx() creates a networkx.Graph() object.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
flow = Flows(dl1, dl2)
graph = flow.to_networkx()
assert graph.has_node("Arghakhanchi")
assert "Dadeldhura" in graph.neighbors("Arghakhanchi")
assert "Sankhuwasabha" not in graph.neighbors("Arghakhanchi")
def test_do_cache_nested(flowmachine_connect):
"""
Test that a query containing nested subqueries can be cached.
"""
hl1 = ModalLocation(daily_location("2016-01-01"),
daily_location("2016-01-02"))
hl2 = ModalLocation(daily_location("2016-01-03"),
daily_location("2016-01-04"))
flow = Flows(hl1, hl2)
flow._db_store_cache_metadata()
assert cache_table_exists(flowmachine_connect, flow.md5)
def test_a_mul_b_no_b_flow(get_dataframe):
"""
No row is returned if the Flows() is not in both A and B.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
dl3 = daily_location("2016-01-07")
flowA = Flows(dl1, dl2)
flowB = Flows(dl1, dl3)
relfl = flowA * flowB
df_rel = get_dataframe(relfl)
with pytest.raises(IndexError):
diff = df_rel[(df_rel.name_from == "Humla")
& (df_rel.name_to == "Kapilbastu")]["count"].values[0]
def test_a_mul_b(get_dataframe):
"""
Multiplying one Flows() by another gives an expected value.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
dl3 = daily_location("2016-01-07")
flowA = Flows(dl1, dl2)
flowB = Flows(dl1, dl3)
relfl = flowA * flowB
df_rel = get_dataframe(relfl)
diff = df_rel[(df_rel.name_from == "Bajhang")
& (df_rel.name_to == "Myagdi")]["count"].values[0]
assert 12 == diff
def test_a_plus_b_no_b_flow(get_dataframe):
"""
Adding a Flows() where it does not exist on the other side returns the original Flows() count.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
dl3 = daily_location("2016-01-07")
flowA = Flows(dl1, dl2)
flowB = Flows(dl1, dl3)
df_rel = get_dataframe(flowA + flowB)
diff = df_rel[(df_rel.name_from == "Humla") & (df_rel.name_to == "Kapilbastu")][
"count"
].values[0]
assert 2 == diff
def test_a_plus_b(get_dataframe):
"""
Adding two Flows() together produces a known value.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
dl3 = daily_location("2016-01-07")
flowA = Flows(dl1, dl2)
flowB = Flows(dl1, dl3)
df_rel = get_dataframe(flowA + flowB)
diff = df_rel[(df_rel.name_from == "Bajhang") & (df_rel.name_to == "Myagdi")][
"count"
].values[0]
assert 7 == diff
def test_store_cache_nested(flowmachine_connect):
"""
Test that storing a query with nested subqueries also caches it.
"""
hl1 = ModalLocation(daily_location("2016-01-01"),
daily_location("2016-01-02"))
hl2 = ModalLocation(daily_location("2016-01-03"),
daily_location("2016-01-04"))
flow = Flows(hl1, hl2)
flow.store().result()
# Should be stored
assert flow.is_stored
assert cache_table_exists(get_db(), flow.query_id)
def test_average_self(get_dataframe):
"""
Adding a Flows() to itself and dividing is the same as the original Flows().
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
dl3 = daily_location("2016-01-07")
flowA = Flows(dl1, dl2)
flowB = Flows(dl1, dl3)
avged = get_dataframe((flowA + flowA) / 2)
orig = get_dataframe(flowA)
compare = (avged.set_index(["name_from",
"name_to"]).sort_index() == orig.set_index(
["name_from", "name_to"]).sort_index())
assert compare.all().values[0]
def test_inoutflow_with_double_column_location():
"""
Test that flowmachine.Inflow can handle a location with
more than one column.
"""
dl1 = daily_location("2016-01-01", spatial_unit=make_spatial_unit("versioned-site"))
dl2 = daily_location("2016-01-02", spatial_unit=make_spatial_unit("versioned-site"))
flow = Flows(dl1, dl2)
expected_columns = ["site_id_to", "version_to", "lon_to", "lat_to", "total"]
assert flow.inflow().column_names == expected_columns
expected_columns = ["site_id_from", "version_from", "lon_from", "lat_from", "total"]
assert flow.outflow().column_names == expected_columns
def test_a_div_b_no_b_flow(get_dataframe):
"""
Rows where there is not an exact match are not returned.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
dl3 = daily_location("2016-01-07")
flowA = Flows(dl1, dl2)
flowB = Flows(dl1, dl3)
relfl = flowA / flowB
df_rel = get_dataframe(relfl)
with pytest.raises(IndexError):
diff = df_rel[(df_rel.name_from == "Humla") & (df_rel.name_to == "Kapilbastu")][
"count"
].values[0]
def test_do_cache_nested(flowmachine_connect):
"""
Test that a query containing nested subqueries can be cached.
"""
hl1 = HomeLocation(daily_location("2016-01-01"),
daily_location("2016-01-02"))
hl2 = HomeLocation(daily_location("2016-01-03"),
daily_location("2016-01-04"))
flow = Flows(hl1, hl2)
flow._db_store_cache_metadata()
in_cache = bool(
flowmachine_connect.fetch(
f"SELECT * FROM cache.cached WHERE query_id='{flow.md5}'"))
assert in_cache
def test_deps_cache_broken_chain():
"""
Test that a Query -> not_cached -> cached chain will
return a dependency on cached.
"""
dl1 = daily_location("2016-01-01")
dl1.store().result()
hl1 = ModalLocation(daily_location("2016-01-01"),
daily_location("2016-01-02"))
hl2 = ModalLocation(daily_location("2016-01-03"),
daily_location("2016-01-04"))
flow = Flows(hl1, hl2)
dep = dl1.query_id
assert 8 == len(flow._get_stored_dependencies())
assert dep in [x.query_id for x in flow._get_stored_dependencies()]
def test_summation(get_dataframe):
"""
Summing a list of Flows() yields a known value.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
dl3 = daily_location("2016-01-07")
flowA = Flows(dl1, dl2)
flowB = Flows(dl1, dl3)
flows = [flowA, flowB, flowA]
summed = sum(flows)
res = get_dataframe(summed)
result = res[(res.name_from == "Bajhang") & (res.name_to == "Myagdi")][
"count"
].values[0]
assert 10 == result
def test_flows_raise_error(self):
"""
Flows() raises error if location levels are different.
"""
dl1 = daily_location("2016-01-01", level="admin3")
dl2 = daily_location("2016-01-01", level="admin2")
with self.assertRaises(ValueError):
Flows(dl1, dl2)
def test_store_cache_nested(flowmachine_connect):
"""
Test that storing a query with nested subqueries also caches it.
"""
hl1 = HomeLocation(daily_location("2016-01-01"),
daily_location("2016-01-02"))
hl2 = HomeLocation(daily_location("2016-01-03"),
daily_location("2016-01-04"))
flow = Flows(hl1, hl2)
flow.store().result()
# Should be stored
assert flow.is_stored
in_cache = bool(
flowmachine_connect.fetch(
f"SELECT * FROM cache.cached WHERE query_id='{flow.md5}'"))
assert in_cache
def test_deps_cache_chain():
"""
Test that a Query -> cached1 -> cached2 chain will
return only a dependency on cached1.
"""
dl1 = daily_location("2016-01-01")
hl1 = ModalLocation(daily_location("2016-01-01"),
daily_location("2016-01-02"))
hl1.store().result()
hl2 = ModalLocation(daily_location("2016-01-03"),
daily_location("2016-01-04"))
flow = Flows(hl1, hl2)
bad_dep = dl1.query_id
good_dep = hl1.query_id
assert 6 == len(flow._get_stored_dependencies())
assert good_dep in [x.query_id for x in flow._get_stored_dependencies()]
assert bad_dep not in [x.query_id for x in flow._get_stored_dependencies()]
def test_bad_divisor():
"""
Dividing by something which isn't a Flows() or scalar raises a ValueError.
"""
dl1 = daily_location("2016-01-01")
dl2 = daily_location("2016-01-02")
flowA = Flows(dl1, dl2)
with pytest.raises(TypeError):
relfl = flowA / "A"
