def diff(A, B, context=False, mods=False):
'''Given two graphs A and B, where it is generally assumed that B is a "newer" version of A,
returns a new graph which captures information about which nodes and edges of A were
removed, added, and remain the same in B.
Specifically, it returns A ∪ B, such that:
1. Nodes in A - B are given the "diffstatus" attribute "removed"
2. Nodes in B - A are given the "diffstatus" attribute "added"
3. Nodes in A ∩ B are given the "diffstatus" attribute "same"
Notice that the union of 1 - 3 equals A ∪ B.
The optional parameter context, when true, will prune the graph so that nodes/edges which are
the same are only present in the diff graph if:
1. An edge incident on/to/from it has been changed, or
2. it is connected to a changed node.
The optional parameter mods, when true, will check for attribute modifications on nodes and
edges, in addition to new/removed nodes. Any nodes/edges that have had their attributes
changed between A and B are marked with the "diffstatus" attribute as "modified."
WARNING: Currently, this method only works if both A and B were generated with unique IDs in a
deterministic fashion; i.e., two identical nodes are given the same ID at both points in time.
This means that diff() will not work on graphs which were generated with automatic random UUIDs.
'''
# must take their union first, then mark appropriate nodes/edges
AB = sn.union(A, B)
# any edges incident on, to, or from the removed nodes will not be in the removed graph,
# since we cannot have edges incident on, to, or from non-existent nodes
removed = sn.difference(A, B)
_mark_nodes_edges_as(AB, removed, 'removed')
_mark_incident_edges_as(AB, removed, 'removed')
added = sn.difference(B, A)
_mark_nodes_edges_as(AB, added, 'added')
_mark_incident_edges_as(AB, added, 'added')
same = sn.intersection(B, A)
_mark_nodes_edges_as(AB, same, 'same')
_check_changed_edges(A, B, AB, same)
if mods:
_check_mods(A, B, AB, same)
if context:
_clear_clutter(AB)
return AB
def test_difference(populated_digraph):
A = populated_digraph
B = populated_digraph.copy()
# remove node C. Consequently, also removes edges (A, C) and (B, C)
B.remove_node('3cd197c2cf5e42dc9ccd0c2adcaf4bc2')
d = B.add_node({"type": "D"}, 'da30015efe3c44dbb0b3b3862cef704a') # add another node D
B.add_edge(d, '3caaa8c09148493dbdf02c574b95526c') # add an edge from D to A
e = A.add_node({"type": "E"}, 'b1b1c6bbbce74a6fb40ee2486cebef26') # add another node
f = A.add_node({"type": "F"}, '3a668c22b43e4521b3c9f042fb2380c2') # add another node
A.add_edge(e, f, {"type": "irregular"}, 'a216de41cca8412fa4b3f432b5d3b0e4') # add edge between the two new nodes
C = sn.difference(A, B) # A - B
correct_nodes = {
uuid.UUID('3cd197c2cf5e42dc9ccd0c2adcaf4bc2'): {
"type": "C",
"id": uuid.UUID('3cd197c2cf5e42dc9ccd0c2adcaf4bc2')
},
uuid.UUID('b1b1c6bbbce74a6fb40ee2486cebef26'): {
"type": "E",
"id": uuid.UUID('b1b1c6bbbce74a6fb40ee2486cebef26')
},
uuid.UUID('3a668c22b43e4521b3c9f042fb2380c2'): {
"type": "F",
"id": uuid.UUID('3a668c22b43e4521b3c9f042fb2380c2')
}
}
assert C.get_nodes() == correct_nodes
correct_edges = {
# e,f
uuid.UUID('a216de41cca8412fa4b3f432b5d3b0e4'): {
"type": "irregular",
"src": uuid.UUID('b1b1c6bbbce74a6fb40ee2486cebef26'),
"dst": uuid.UUID('3a668c22b43e4521b3c9f042fb2380c2'),
"id": uuid.UUID('a216de41cca8412fa4b3f432b5d3b0e4')
}
}
assert C.get_edges() == correct_edges
def test_difference_custom_lambda(populated_digraph):
# add some attributes
populated_digraph.set_node_attribute('3caaa8c09148493dbdf02c574b95526c', 'depth', 0)
populated_digraph.set_node_attribute('2cdfebf3bf9547f19f0412ccdfbe03b7', 'depth', 0)
populated_digraph.set_node_attribute('3cd197c2cf5e42dc9ccd0c2adcaf4bc2', 'depth', 1)
# add a node of type D
d = populated_digraph.add_node({'type': 'D', 'depth': 1}, '63cf70d2762043c29eb5e3e958383f4a')
populated_digraph.set_edge_attribute('5f5f44ec7c0144e29c5b7d513f92d9ab', 'weight', 1)
populated_digraph.set_edge_attribute('f3674fcc691848ebbd478b1bfb3e84c3', 'weight', 2) # (B, A)
populated_digraph.set_edge_attribute('7eb91be54d3746b89a61a282bcc207bb', 'weight', 3)
populated_digraph.set_edge_attribute('c172a3599b7d4ef3bbb688277276b763', 'weight', 5) # (B, C)
# make a copy to change around
new_populated_digraph = populated_digraph.copy()
# add an edge between the new node d and node B with weight 8
new_populated_digraph.add_edge(d, '2cdfebf3bf9547f19f0412ccdfbe03b7', {'weight': 8},
'8ccd176a48284915828e5ac7e13bc43a')
# remove node C. Will remove the edges (B, C) and (A, C) as well
new_populated_digraph.remove_node('3cd197c2cf5e42dc9ccd0c2adcaf4bc2')
new_populated_digraph.remove_edge('f3674fcc691848ebbd478b1bfb3e84c3') # remove edge (B, A)
# custom lambda that defines membership in the usual way, but only for nodes with
# a depth greater than 0, and edges with a weight greater than 2
node_depth_gt_0 = lambda nid, G: sn.node_in(nid, G) and G.get_node_attribute(nid, 'depth') > 0
edge_weight_gt_2 = lambda eid, G: sn.edge_in(eid, G) and G.get_edge_attribute(eid, 'weight') > 2
dg = sn.difference(populated_digraph, new_populated_digraph, node_depth_gt_0, edge_weight_gt_2)
# the first two nodes are "not in B" as we have defined it because their
# depth is not greater than 0
correct_nodes = {
uuid.UUID('2cdfebf3-bf95-47f1-9f04-12ccdfbe03b7'): {
'depth': 0,
'id': uuid.UUID('2cdfebf3-bf95-47f1-9f04-12ccdfbe03b7'),
'type': 'B'
},
uuid.UUID('3caaa8c0-9148-493d-bdf0-2c574b95526c'): {
'depth': 0,
'id': uuid.UUID('3caaa8c0-9148-493d-bdf0-2c574b95526c'),
'type': 'A'
},
uuid.UUID('3cd197c2cf5e42dc9ccd0c2adcaf4bc2'): {
"id": uuid.UUID('3cd197c2cf5e42dc9ccd0c2adcaf4bc2'),
"type": "C",
"depth": 1
}
}
assert dg.get_nodes() == correct_nodes
correct_edges = {
# the new graph DOES have edge (B, A), but its weight 1, which is not > 2,
# so by our definition, (B, A) is "not in" the new graph
uuid.UUID('5f5f44ec-7c01-44e2-9c5b-7d513f92d9ab'): {
'dst': uuid.UUID('2cdfebf3-bf95-47f1-9f04-12ccdfbe03b7'),
'id': uuid.UUID('5f5f44ec-7c01-44e2-9c5b-7d513f92d9ab'),
'src': uuid.UUID('3caaa8c0-9148-493d-bdf0-2c574b95526c'),
'type': 'normal',
'weight': 1
},
# the new graph removed edge (B, A)
uuid.UUID('f3674fcc691848ebbd478b1bfb3e84c3'): {
'id': uuid.UUID('f3674fcc691848ebbd478b1bfb3e84c3'),
'src': uuid.UUID('2cdfebf3bf9547f19f0412ccdfbe03b7'),
'dst': uuid.UUID('3caaa8c09148493dbdf02c574b95526c'),
'weight': 2,
'type': 'normal'
},
# the new graph removed edge (A, C)
uuid.UUID('7eb91be5-4d37-46b8-9a61-a282bcc207bb'): {
'dst': uuid.UUID('3cd197c2-cf5e-42dc-9ccd-0c2adcaf4bc2'),
'id': uuid.UUID('7eb91be5-4d37-46b8-9a61-a282bcc207bb'),
'src': uuid.UUID('3caaa8c0-9148-493d-bdf0-2c574b95526c'),
'type': 'normal',
'weight': 3
},
# new graph removed edge (B, C)
uuid.UUID('c172a359-9b7d-4ef3-bbb6-88277276b763'): {
'dst': uuid.UUID('3cd197c2-cf5e-42dc-9ccd-0c2adcaf4bc2'),
'id': uuid.UUID('c172a359-9b7d-4ef3-bbb6-88277276b763'),
'src': uuid.UUID('2cdfebf3-bf95-47f1-9f04-12ccdfbe03b7'),
'type': 'irregular',
'weight': 5
}
}
assert dg.get_edges() == correct_edges