def test_view_graph_bundle_flow_partitions_must_be_equal():
material_partition_mn = Partition.Simple('material', ['m', 'n'])
material_partition_XY = Partition.Simple('material', ['X', 'Y'])
nodes = {
'a': ProcessGroup(selection=['a1']),
'b': ProcessGroup(selection=['b1']),
'c': ProcessGroup(selection=['c1']),
'via': Waypoint(),
}
order = [['a', 'b'], ['via'], ['c']]
bundles = [
Bundle('a',
'c',
waypoints=['via'],
flow_partition=material_partition_mn),
Bundle('b',
'c',
waypoints=['via'],
flow_partition=material_partition_XY),
]
# Do partition based on flows stored in bundles
with pytest.raises(ValueError):
G = view_graph(SankeyDefinition(nodes, bundles, order))
bundles[1] = Bundle('b',
'c',
waypoints=['via'],
flow_partition=material_partition_mn)
assert view_graph(SankeyDefinition(nodes, bundles, order))
def test_sankey_definition_as_script_with_partitions():
nodes = {
'a':
ProcessGroup(selection=['a1', 'a2']),
'b':
ProcessGroup(selection=['b1']),
'c':
ProcessGroup(selection=['c1', 'c2'],
partition=Partition.Simple('process', ['c1', 'c2'])),
'via':
Waypoint(partition=Partition.Simple('material', ['m', 'n'])),
}
bundles = [
Bundle('a', 'c', waypoints=['via']),
Bundle('b', 'c', waypoints=['via']),
]
ordering = [[['a', 'b']], [['via']], [['c']]]
sdd = SankeyDefinition(nodes,
bundles,
ordering,
flow_partition=Partition.Simple(
'material', ['m', 'n']))
code = sdd.to_code()
# Check roundtrip
ctx = {}
exec(code, ctx)
assert ctx["sdd"] == sdd
def test_partition_simple_checks_for_duplicates():
with pytest.raises(ValueError):
Partition.Simple('dim1', ['a', 'a'])
with pytest.raises(ValueError):
Partition.Simple('dim1', [
('label1', ['a', 'b']),
'b'
])
def test_results_graph_material_key():
# Mock flow data
flows = pd.DataFrame.from_records([
('a1', 'c1', 'm', 'long', 3),
('a1', 'c1', 'n', 'long', 1),
],
columns=('source', 'target',
'material_type', 'shape',
'value'))
view_graph = LayeredGraph()
view_graph.add_node('a', node=ProcessGroup())
view_graph.add_node('c', node=ProcessGroup())
view_graph.add_edge('a', 'c', bundles=[0])
view_graph.ordering = Ordering([[['a']], [['c']]])
bundle_flows = {0: flows}
material_partition = Partition.Simple('material_type', ['m', 'n'])
shape_partition = Partition.Simple('shape', ['long', 'thin'])
# Partition based on material_type
view_graph.edges['a', 'c']['flow_partition'] = material_partition
Gr, groups = results_graph(view_graph, bundle_flows)
assert sorted(Gr.edges(keys=True, data=True)) == [
('a^*', 'c^*', ('m', '*'), {
'measures': {
'value': 3
},
'original_flows': [0],
'bundles': [0]
}),
('a^*', 'c^*', ('n', '*'), {
'measures': {
'value': 1
},
'original_flows': [1],
'bundles': [0]
}),
]
# Partition based on shape
view_graph.edges['a', 'c']['flow_partition'] = shape_partition
Gr, groups = results_graph(view_graph, bundle_flows)
assert sorted(Gr.edges(keys=True, data=True)) == [
('a^*', 'c^*', ('long', '*'), {
'measures': {
'value': 4
},
'original_flows': [0, 1],
'bundles': [0]
}),
]
def test_results_graph_with_extra_or_not_enough_groups():
# Mock flow data
bundle_flows = {
0:
pd.DataFrame.from_records([
('a1', 'b1', 'm', 3),
('a2', 'b1', 'm', 1),
],
columns=('source', 'target', 'material',
'value'))
}
# Group 'a3' not used. ProcessGroup 'a2' isn't in any group.
node_a = ProcessGroup(partition=Partition.Simple('process', ['a1', 'a3']))
node_b = ProcessGroup(partition=Partition.Simple('process', ['b1']))
view_graph = LayeredGraph()
view_graph.add_node('a', node=node_a)
view_graph.add_node('b', node=node_b)
view_graph.add_edges_from([
('a', 'b', {
'bundles': [0]
}),
])
view_graph.ordering = Ordering([[['a']], [['b']]])
# Do partition based on flows stored in bundles
Gr, groups = results_graph(view_graph, bundle_flows)
assert set(Gr.nodes()) == {'a^a1', 'a^_', 'b^b1'}
assert sorted(Gr.edges(keys=True, data=True)) == [
('a^_', 'b^b1', ('*', '*'), {
'measures': {
'value': 1
},
'original_flows': [1],
'bundles': [0]
}),
('a^a1', 'b^b1', ('*', '*'), {
'measures': {
'value': 3
},
'original_flows': [0],
'bundles': [0]
}),
]
assert Gr.ordering == Ordering([
[['a^a1', 'a^_']],
[['b^b1']],
])
def test_view_graph_adds_waypoints_partition():
nodes = {
'n1': ProcessGroup(selection=['n1']),
'n2': ProcessGroup(selection=['n2']),
}
g = Partition.Simple('test', ['x'])
bundles = [
Bundle('n1', 'n2', default_partition=g),
]
order0 = [['n1'], [], ['n2']]
G = view_graph(SankeyDefinition(nodes, bundles, order0))
assert sorted(nodes_ignoring_elsewhere(G, data=True)) == [
('__n1_n2_1', {
'node': Waypoint(title='', partition=g)
}),
('n1', {
'node': ProcessGroup(selection=['n1'])
}),
('n2', {
'node': ProcessGroup(selection=['n2'])
}),
]
def test_weave_results():
nodes = {
'a':
ProcessGroup(selection=['a1', 'a2']),
'b':
ProcessGroup(selection=['b1']),
'c':
ProcessGroup(selection=['c1', 'c2'],
partition=Partition.Simple('process', ['c1', 'c2'])),
'via':
Waypoint(partition=Partition.Simple('material', ['m', 'n'])),
}
bundles = [
Bundle('a', 'c', waypoints=['via']),
Bundle('b', 'c', waypoints=['via']),
]
ordering = [[['a', 'b']], [['via']], [['c']]]
sdd = SankeyDefinition(nodes, bundles, ordering)
# Dataset
flows = pd.DataFrame.from_records([
('a1', 'c1', 'm', 3),
('a2', 'c1', 'n', 1),
('b1', 'c1', 'm', 1),
('b1', 'c2', 'm', 2),
('b1', 'c2', 'n', 1),
],
columns=('source', 'target', 'material',
'value'))
dim_process = pd.DataFrame({
'id':
list(flows.source.unique()) + list(flows.target.unique())
}).set_index('id')
dataset = Dataset(flows, dim_process)
result = weave(sdd, dataset)
def link(src, tgt, original_flows, value, link_type='*', color='#FBB4AE'):
return SankeyLink(source=src,
target=tgt,
type=link_type,
time='*',
data={'value': value},
title=link_type,
color=color,
original_flows=original_flows)
assert set(n.id for n in result.nodes) == {
'a^*', 'b^*', 'via^m', 'via^n', 'c^c1', 'c^c2'
}
assert sorted(result.links) == [
link('a^*', 'via^m', [0], 3),
link('a^*', 'via^n', [1], 1),
link('b^*', 'via^m', [2, 3], 3),
link('b^*', 'via^n', [4], 1),
link('via^m', 'c^c1', [0, 2], 4),
link('via^m', 'c^c2', [3], 2),
link('via^n', 'c^c1', [1], 1),
link('via^n', 'c^c2', [4], 1),
]
assert result.ordering == Ordering([
[['a^*', 'b^*']],
[['via^m', 'via^n']],
[['c^c1', 'c^c2']],
])
assert result.groups == [
{
'id': 'via',
'title': '',
'type': 'group',
'nodes': ['via^m', 'via^n']
},
{
'id': 'c',
'title': '',
'type': 'process',
'nodes': ['c^c1', 'c^c2']
},
]
# Can also set flow_partition for all bundles at once
sdd2 = SankeyDefinition(nodes,
bundles,
ordering,
flow_partition=Partition.Simple(
'material', ['m', 'n']))
scale = CategoricalScale('type', palette=['red', 'blue'])
scale.set_domain(['m', 'n'])
result = weave(sdd2, dataset, link_color=scale)
assert sorted(result.links) == [
link('a^*', 'via^m', [0], 3, 'm', 'red'),
link('a^*', 'via^n', [1], 1, 'n', 'blue'),
link('b^*', 'via^m', [2, 3], 3, 'm', 'red'),
link('b^*', 'via^n', [4], 1, 'n', 'blue'),
link('via^m', 'c^c1', [0, 2], 4, 'm', 'red'),
link('via^m', 'c^c2', [3], 2, 'm', 'red'),
link('via^n', 'c^c1', [1], 1, 'n', 'blue'),
link('via^n', 'c^c2', [4], 1, 'n', 'blue'),
]
def test_simple_partition_groups():
G = Partition.Simple('dim1', ['x', ('group', ['y', 'z'])])
assert G.labels == ['x', 'group']
assert G.groups == (Group('x', [('dim1', ('x', ))]),
Group('group', [('dim1', ('y', 'z'))]), )
def test_results_graph_overall():
material_partition = Partition.Simple('material', ['m', 'n'])
c_partition = Partition.Simple('process', ['c1', 'c2'])
view_graph = LayeredGraph()
view_graph.add_node('a', node=ProcessGroup(title='Node a'))
view_graph.add_node('b', node=ProcessGroup())
view_graph.add_node('c', node=ProcessGroup(partition=c_partition))
view_graph.add_node('via', node=Waypoint(partition=material_partition))
view_graph.add_edges_from([
('a', 'via', {
'bundles': [0],
'flow_partition': material_partition
}),
('b', 'via', {
'bundles': [1],
'flow_partition': material_partition
}),
('via', 'c', {
'bundles': [0, 1],
'flow_partition': material_partition
}),
])
view_graph.ordering = Ordering([[['a', 'b']], [['via']], [['c']]])
# Mock flow data
bundle_flows = {
0:
pd.DataFrame.from_records([
('a1', 'c1', 'm', 3),
('a2', 'c1', 'n', 1),
],
index=(0, 1),
columns=('source', 'target', 'material',
'value')),
1:
pd.DataFrame.from_records([
('b1', 'c1', 'm', 1),
('b1', 'c2', 'm', 2),
('b1', 'c2', 'n', 1),
],
index=(2, 3, 4),
columns=('source', 'target', 'material',
'value'))
}
# Do partition based on flows stored in bundles
Gr, groups = results_graph(view_graph, bundle_flows)
assert sorted(Gr.nodes(data=True)) == [
('a^*', {
'direction': 'R',
'type': 'process',
'title': 'Node a'
}),
('b^*', {
'direction': 'R',
'type': 'process',
'title': 'b'
}),
('c^c1', {
'direction': 'R',
'type': 'process',
'title': 'c1'
}),
('c^c2', {
'direction': 'R',
'type': 'process',
'title': 'c2'
}),
('via^m', {
'direction': 'R',
'type': 'group',
'title': 'm'
}),
('via^n', {
'direction': 'R',
'type': 'group',
'title': 'n'
}),
]
assert sorted(Gr.edges(keys=True, data=True)) == [
('a^*', 'via^m', ('m', '*'), {
'measures': {
'value': 3
},
'original_flows': [0],
'bundles': [0]
}),
('a^*', 'via^n', ('n', '*'), {
'measures': {
'value': 1
},
'original_flows': [1],
'bundles': [0]
}),
('b^*', 'via^m', ('m', '*'), {
'measures': {
'value': 3
},
'original_flows': [2, 3],
'bundles': [1]
}),
('b^*', 'via^n', ('n', '*'), {
'measures': {
'value': 1
},
'original_flows': [4],
'bundles': [1]
}),
('via^m', 'c^c1', ('m', '*'), {
'measures': {
'value': 4
},
'original_flows': [0, 2],
'bundles': [0, 1]
}),
('via^m', 'c^c2', ('m', '*'), {
'measures': {
'value': 2
},
'original_flows': [3],
'bundles': [0, 1]
}),
('via^n', 'c^c1', ('n', '*'), {
'measures': {
'value': 1
},
'original_flows': [1],
'bundles': [0, 1]
}),
('via^n', 'c^c2', ('n', '*'), {
'measures': {
'value': 1
},
'original_flows': [4],
'bundles': [0, 1]
}),
]
assert Gr.ordering == Ordering([
[['a^*', 'b^*']],
[['via^m', 'via^n']],
[['c^c1', 'c^c2']],
])
# Only includes groups where the title is not the same as the single node
# title
assert groups == [
{
'id': 'via',
'title': '',
'type': 'group',
'nodes': ['via^m', 'via^n']
},
{
'id': 'c',
'title': '',
'type': 'process',
'nodes': ['c^c1', 'c^c2']
},
]
def test_results_graph_time_partition():
time_partition = Partition.Simple('time', [1, 2])
view_graph = LayeredGraph()
view_graph.add_node('a', node=ProcessGroup())
view_graph.add_node('b', node=ProcessGroup())
view_graph.add_edges_from([
('a', 'b', {
'bundles': [0]
}),
])
view_graph.ordering = Ordering([[['a']], [['b']]])
# Mock flow data
bundle_flows = {
0:
pd.DataFrame.from_records([
('a1', 'b1', 'm', 1, 3),
('a2', 'b1', 'n', 1, 1),
('a2', 'b2', 'n', 1, 2),
('a1', 'b1', 'm', 2, 1),
('a1', 'b1', 'n', 2, 3),
],
columns=('source', 'target', 'material',
'time', 'value')),
}
# Do partition based on flows stored in bundles
Gr, groups = results_graph(view_graph,
bundle_flows,
time_partition=time_partition)
assert sorted(Gr.edges(keys=True, data=True)) == [
('a^*', 'b^*', ('*', '1'), {
'measures': {
'value': 6
},
'original_flows': [0, 1, 2],
'bundles': [0]
}),
('a^*', 'b^*', ('*', '2'), {
'measures': {
'value': 4
},
'original_flows': [3, 4],
'bundles': [0]
}),
]
# Now add a material partition too
material_partition = Partition.Simple('material', ['m', 'n'])
Gr, groups = results_graph(view_graph, bundle_flows, material_partition,
time_partition)
assert sorted(Gr.edges(keys=True, data=True)) == [
('a^*', 'b^*', ('m', '1'), {
'measures': {
'value': 3
},
'original_flows': [0],
'bundles': [0]
}),
('a^*', 'b^*', ('m', '2'), {
'measures': {
'value': 1
},
'original_flows': [3],
'bundles': [0]
}),
('a^*', 'b^*', ('n', '1'), {
'measures': {
'value': 3
},
'original_flows': [1, 2],
'bundles': [0]
}),
('a^*', 'b^*', ('n', '2'), {
'measures': {
'value': 3
},
'original_flows': [4],
'bundles': [0]
}),
]
def test_results_graph_elsewhere_stubs():
b_partition = Partition.Simple('process', ['b1', 'b2'])
view_graph = LayeredGraph()
view_graph.add_node('a', node=ProcessGroup())
view_graph.add_node('b',
node=ProcessGroup(partition=b_partition),
from_elsewhere_bundles=[1])
view_graph.add_edge('a', 'b', bundles=[0])
view_graph.ordering = Ordering([[['a']], [['b']]])
# Mock flow data
bundle_flows = {
0:
pd.DataFrame.from_records([
('a1', 'b1', 'm', 3),
('a2', 'b1', 'n', 1),
],
index=(0, 1),
columns=('source', 'target', 'material',
'value')),
1:
pd.DataFrame.from_records([
('x1', 'b1', 'm', 1),
('x3', 'b2', 'n', 5),
],
index=(2, 3),
columns=('source', 'target', 'material',
'value'))
}
# Do partition based on flows stored in bundles
Gr, groups = results_graph(view_graph, bundle_flows)
assert sorted(Gr.nodes(data=True)) == [
('a^*', {
'direction': 'R',
'type': 'process',
'title': 'a'
}),
('b^b1', {
'direction':
'R',
'type':
'process',
'title':
'b1',
'from_elsewhere_edges': [
(('*', '*'), {
'measures': {
'value': 1
},
'original_flows': [2],
'bundles': [1]
}),
]
}),
('b^b2', {
'direction':
'R',
'type':
'process',
'title':
'b2',
'from_elsewhere_edges': [
(('*', '*'), {
'measures': {
'value': 5
},
'original_flows': [3],
'bundles': [1]
}),
]
}),
]
assert sorted(Gr.edges(keys=True, data=True)) == [
('a^*', 'b^b1', ('*', '*'), {
'measures': {
'value': 4
},
'original_flows': [0, 1],
'bundles': [0]
}),
]
