def prepare_nidb(nidb):
graph = nidb.raw_graph()
for node in graph:
if graph.node[node].get("graphics"):
graph.node[node]['x'] = graph.node[node]['graphics']['x']
graph.node[node]['y'] = graph.node[node]['graphics']['y']
graph.node[node]['device_type'] = graph.node[node]['graphics'][
'device_type']
graph.node[node]['device_subtype'] = graph.node[node]['graphics'][
'device_subtype']
for interface_index in graph.node[node]['_ports']:
try:
interface_id = graph.node[node]["_ports"][interface_index][
'id']
except KeyError: # interface doesn't exist, eg for a lan segment
interface_id = ""
id_brief = shortened_interface(interface_id)
graph.node[node]["_ports"][interface_index]['id_brief'] = id_brief
x = (graph.node[n]['x'] for n in graph)
y = (graph.node[n]['y'] for n in graph)
x_min = min(x)
y_min = min(y)
for n in graph:
graph.node[n]['x'] += -x_min
graph.node[n]['y'] += -y_min
return graph
def prepare_nidb(nidb):
graph = nidb.raw_graph()
for node in graph:
if graph.node[node].get("graphics"):
graph.node[node]['x'] = graph.node[node]['graphics']['x']
graph.node[node]['y'] = graph.node[node]['graphics']['y']
graph.node[node]['device_type'] = graph.node[node]['graphics']['device_type']
graph.node[node]['device_subtype'] = graph.node[node]['graphics']['device_subtype']
for interface_index in graph.node[node]['_interfaces']:
try:
interface_id = graph.node[node]["_interfaces"][interface_index]['id']
except KeyError: # interface doesn't exist, eg for a lan segment
interface_id = ""
id_brief = shortened_interface(interface_id)
graph.node[node]["_interfaces"][interface_index]['id_brief'] = id_brief
x = (graph.node[n]['x'] for n in graph)
y = (graph.node[n]['y'] for n in graph)
x_min = min(x)
y_min = min(y)
for n in graph:
graph.node[n]['x'] += - x_min
graph.node[n]['y'] += - y_min
return graph
def jsonify_anm_with_graphics(anm, nidb=None):
""" Returns a dictionary of json-ified overlay graphs, with graphics data appended to each overlay"""
from collections import defaultdict
import math
anm_json = {}
test_anm_data = {}
graphics_graph = anm["graphics"]._graph.copy()
phy_graph = anm["phy"]._graph # to access ASNs
"""simple layout of deps - more advanced layout could
export to dot and import to omnigraffle, etc
"""
g_deps = anm['_dependencies']
nm_graph = g_deps._graph
# build tree
layers = defaultdict(list)
if len(nm_graph) > 0:
topo_sort = nx.topological_sort(nm_graph)
# trim out any nodes with no sucessors
tree_root = topo_sort[0]
#Note: topo_sort ensures that once reach node, would have reached its predecessors
# start at first element after root
for node in topo_sort:
preds = nm_graph.predecessors(node)
if len(preds):
pred_level = max(nm_graph.node[p].get('level') for p in preds)
else:
# a root node
pred_level = -1 # this node becomes level 0
level = pred_level + 1
nm_graph.node[node]['level'] = level
layers[level].append(node)
data = nm_graph.node[node]
data['y'] = 100 * data['level']
data['device_type'] = "ank_internal"
MIDPOINT = 50 # assign either side of
for nodes in layers.values():
#TODO: since sort is stable, first sort by parent x (avoids zig-zags)
nodes = sorted(nodes,
reverse=True,
key=lambda x: nm_graph.degree(x))
for index, node in enumerate(nodes):
#TODO: work out why weird offset due to the math.pow *
#node_x = MIDPOINT + 125*index * math.pow(-1, index)
node_x = MIDPOINT + 125 * index
nm_graph.node[node]['x'] = node_x
import random
attribute_cache = defaultdict(dict)
# the attributes to copy
# TODO: check behaviour for None if explicitly set
#TODO: need to check if attribute is set in overlay..... using API
copy_attrs = ["x", "y", "asn", "label", "device_type", "device_subtype"]
for node, in_data in phy_graph.nodes(data=True):
out_data = {
key: in_data.get(key)
for key in copy_attrs if key in in_data
}
attribute_cache[node].update(out_data)
# Update for graphics (over-rides phy)
for node, in_data in graphics_graph.nodes(data=True):
out_data = {
key: in_data.get(key)
for key in copy_attrs if key in in_data
}
attribute_cache[node].update(out_data)
# append label from function
for node in anm['phy']:
attribute_cache[node.id]['label'] = str(node)
for overlay_id in anm.overlays():
nm_graph = anm[overlay_id]._graph.copy()
if overlay_id == "_dependencies":
nm_graph = nx.Graph(
nm_graph) #convert to undirected for visual clarify
for node in nm_graph:
node_data = dict(attribute_cache.get(node, {}))
# update with node data from this overlay
#TODO: check is not None won't clobber specifically set in overlay...
graph_node_data = nm_graph.node[node]
overlay_node_data = {
key: graph_node_data.get(key)
for key in graph_node_data
}
node_data.update(overlay_node_data)
# check for any non-set properties
if node_data.get("x") is None:
new_x = random.randint(0, 800)
node_data['x'] = new_x
# store for other graphs to use
log.debug("Allocated random x %s to node %s in overlay %s" %
(new_x, node, overlay_id))
attribute_cache[node]['x'] = new_x
if node_data.get("y") is None:
new_y = random.randint(0, 800)
node_data['y'] = new_y
# store for other graphs to use
attribute_cache[node]['y'] = new_y
log.debug("Allocated random y %s to node %s in overlay %s" %
(new_y, node, overlay_id))
if node_data.get("label") == node:
# try from cache
node_data['label'] = attribute_cache.get(node, {}).get("label")
if node_data.get("label") is None:
node_data['label'] = str(node) # don't need to cache
# store on graph
nm_graph.node[node] = node_data
if nm_graph.is_multigraph():
for u, v, k in nm_graph.edges(keys=True):
# Store key: nx node_link_data ignores it
#anm_graph[u][v][k]['_key'] = k
pass # is this needed? as key itself holds no value?
try:
del nm_graph.node[node]['id']
except KeyError:
pass
if nidb:
nidb_graph = nidb.raw_graph()
if node in nidb:
DmNode_data = nidb_graph.node[node]
try:
#TODO: check why not all nodes have _ports initialised
overlay_interfaces = nm_graph.node[node]["_ports"]
except KeyError:
continue # skip copying interface data for this node
for interface_id in overlay_interfaces.keys():
#TODO: use raw_interfaces here
try:
nidb_interface_id = DmNode_data['_ports'][
interface_id]['id']
except KeyError:
#TODO: check why arrive here - something not initialised?
continue
nm_graph.node[node]['_ports'][interface_id][
'id'] = nidb_interface_id
id_brief = shortened_interface(nidb_interface_id)
nm_graph.node[node]['_ports'][interface_id][
'id_brief'] = id_brief
anm_json[overlay_id] = ank_json_dumps(nm_graph)
test_anm_data[overlay_id] = nm_graph
if nidb:
test_anm_data['nidb'] = prepare_nidb(nidb)
result = json.dumps(test_anm_data,
cls=AnkEncoder,
indent=4,
sort_keys=True)
return result
def jsonify_anm_with_graphics(anm, nidb = None):
""" Returns a dictionary of json-ified overlay graphs, with graphics data appended to each overlay"""
anm_json = {}
test_anm_data = {}
graphics_graph = anm["graphics"]._graph.copy()
phy_graph = anm["phy"]._graph # to access ASNs
import random
from collections import defaultdict
attribute_cache = defaultdict(dict)
# the attributes to copy
# TODO: check behaviour for None if explicitly set
#TODO: need to check if attribute is set in overlay..... using API
copy_attrs = ["x", "y", "asn", "label", "device_type", "device_subtype"]
for node, in_data in phy_graph.nodes(data=True):
out_data = {key: in_data.get(key) for key in copy_attrs
if key in in_data}
attribute_cache[node].update(out_data)
# Update for graphics (over-rides phy)
for node, in_data in graphics_graph.nodes(data=True):
out_data = {key: in_data.get(key) for key in copy_attrs
if key in in_data}
attribute_cache[node].update(out_data)
for overlay_id in anm.overlays():
NmGraph = anm[overlay_id]._graph.copy()
for node in NmGraph:
node_data = dict(attribute_cache.get(node, {}))
# update with node data from this overlay
#TODO: check is not None won't clobber specifically set in overlay...
graph_node_data = NmGraph.node[node]
overlay_node_data = {key: graph_node_data.get(key)
for key in copy_attrs if key in graph_node_data}
node_data.update(overlay_node_data)
# check for any non-set properties
if node_data.get("x") is None:
new_x = random.randint(0,800)
node_data['x'] = new_x
# store for other graphs to use
log.debug("Allocated random x %s to node %s in overlay %s" %
(new_x, node, overlay_id))
attribute_cache[node]['x'] = new_x
if node_data.get("y") is None:
new_y = random.randint(0,800)
node_data['y'] =new_y
# store for other graphs to use
attribute_cache[node]['y'] =new_y
log.debug("Allocated random y %s to node %s in overlay %s" %
(new_y, node, overlay_id))
if node_data.get("label") is None:
node_data['label'] = str(node) # don't need to cache
# store on graph
NmGraph.node[node] = node_data
try:
del NmGraph.node[node]['id']
except KeyError:
pass
if nidb:
nidb_graph = nidb.raw_graph()
if node in nidb:
DmNode_data = nidb_graph.node[node]
try:
#TODO: check why not all nodes have _interfaces initialised
overlay_interfaces = NmGraph.node[node]["_interfaces"]
except KeyError:
continue # skip copying interface data for this node
for interface_id in overlay_interfaces.keys():
try:
nidb_interface_id = DmNode_data['_interfaces'][interface_id]['id']
except KeyError:
#TODO: check why arrive here - something not initialised?
continue
NmGraph.node[node]['_interfaces'][interface_id]['id'] = nidb_interface_id
id_brief = shortened_interface(nidb_interface_id)
NmGraph.node[node]['_interfaces'][interface_id]['id_brief'] = id_brief
anm_json[overlay_id] = ank_json_dumps(NmGraph)
test_anm_data[overlay_id] = NmGraph
if nidb:
test_anm_data['nidb'] = prepare_nidb(nidb)
result = json.dumps(test_anm_data, cls=AnkEncoder, indent = 4, sort_keys = True)
return result
def jsonify_anm_with_graphics(anm, nidb=None):
""" Returns a dictionary of json-ified overlay graphs, with graphics data appended to each overlay"""
from collections import defaultdict
import math
anm_json = {}
test_anm_data = {}
graphics_graph = anm["graphics"]._graph.copy()
phy_graph = anm["phy"]._graph # to access ASNs
"""simple layout of deps - more advanced layout could
export to dot and import to omnigraffle, etc
"""
g_deps = anm['_dependencies']
nm_graph = g_deps._graph
# build tree
layers = defaultdict(list)
nodes_by_layer = {}
if len(nm_graph) > 0:
topo_sort = nx.topological_sort(nm_graph)
# trim out any nodes with no sucessors
tree_root = topo_sort[0]
# Note: topo_sort ensures that once reach node, would have reached its predecessors
# start at first element after root
for node in topo_sort:
preds = nm_graph.predecessors(node)
if len(preds):
pred_level = max(nm_graph.node[p].get('level') for p in preds)
else:
# a root node
pred_level = -1 # this node becomes level 0
level = pred_level + 1
nm_graph.node[node]['level'] = level
layers[level].append(node)
data = nm_graph.node[node]
data['y'] = 100 * data['level']
data['device_type'] = "ank_internal"
MIDPOINT = 50 # assign either side of
for layer, nodes in layers.items():
# TODO: since sort is stable, first sort by parent x (avoids
# zig-zags)
nodes = sorted(nodes, reverse=True,
key=lambda x: nm_graph.degree(x))
for index, node in enumerate(nodes):
# TODO: work out why weird offset due to the math.pow *
#node_x = MIDPOINT + 125*index * math.pow(-1, index)
node_x = MIDPOINT + 125 * index
nm_graph.node[node]['x'] = node_x
nodes_by_layer[node] = layer
import random
attribute_cache = defaultdict(dict)
# the attributes to copy
# TODO: check behaviour for None if explicitly set
# TODO: need to check if attribute is set in overlay..... using API
copy_attrs = ["x", "y", "asn", "label", "device_type", "device_subtype"]
for node, in_data in phy_graph.nodes(data=True):
out_data = {key: in_data.get(key) for key in copy_attrs
if key in in_data}
attribute_cache[node].update(out_data)
# Update for graphics (over-rides phy)
for node, in_data in graphics_graph.nodes(data=True):
out_data = {key: in_data.get(key) for key in copy_attrs
if key in in_data}
attribute_cache[node].update(out_data)
# append label from function
for node in anm['phy']:
attribute_cache[node.id]['label'] = str(node)
overlay_ids = sorted(anm.overlays(),
key=lambda x: nodes_by_layer.get(x, 0))
for overlay_id in overlay_ids:
nm_graph = anm[overlay_id]._graph.copy()
if overlay_id == "_dependencies":
# convert to undirected for visual clarify
nm_graph = nx.Graph(nm_graph)
for node in nm_graph:
node_data = dict(attribute_cache.get(node, {}))
# update with node data from this overlay
# TODO: check is not None won't clobber specifically set in
# overlay...
graph_node_data = nm_graph.node[node]
overlay_node_data = {key: graph_node_data.get(key)
for key in graph_node_data}
node_data.update(overlay_node_data)
# check for any non-set properties
if node_data.get("x") is None:
new_x = random.randint(0, 800)
node_data['x'] = new_x
# store for other graphs to use
log.debug("Allocated random x %s to node %s in overlay %s" %
(new_x, node, overlay_id))
attribute_cache[node]['x'] = new_x
else:
# cache for next time, such as vswitch in l2 for l2_bc
attribute_cache[node]['x'] = node_data['x']
if node_data.get("y") is None:
new_y = random.randint(0, 800)
node_data['y'] = new_y
# store for other graphs to use
attribute_cache[node]['y'] = new_y
log.debug("Allocated random y %s to node %s in overlay %s" %
(new_y, node, overlay_id))
else:
attribute_cache[node]['y'] = node_data['y']
# TODO: may want to re-introduce graphics to store cross-layer data for virtual nodes
# and cache device type and device subtype
# TODO: catch for each, if node not in cache
try:
attribute_cache[node]['device_type'] = node_data['device_type']
except KeyError:
pass # not set
try:
attribute_cache[node][
'device_subtype'] = node_data['device_subtype']
except KeyError:
pass # not set
if node_data.get("label") == node:
# try from cache
node_data['label'] = attribute_cache.get(node, {}).get("label")
if node_data.get("label") is None:
node_data['label'] = str(node) # don't need to cache
# store on graph
nm_graph.node[node] = node_data
if nm_graph.is_multigraph():
for u, v, k in nm_graph.edges(keys=True):
# Store key: nx node_link_data ignores it
#anm_graph[u][v][k]['_key'] = k
pass # is this needed? as key itself holds no value?
try:
del nm_graph.node[node]['id']
except KeyError:
pass
if nidb:
nidb_graph = nidb.raw_graph()
if node in nidb:
DmNode_data = nidb_graph.node[node]
try:
# TODO: check why not all nodes have _ports initialised
overlay_interfaces = nm_graph.node[node]["_ports"]
except KeyError:
continue # skip copying interface data for this node
for interface_id in overlay_interfaces.keys():
# TODO: use raw_interfaces here
try:
nidb_interface_id = DmNode_data[
'_ports'][interface_id]['id']
except KeyError:
# TODO: check why arrive here - something not
# initialised?
continue
nm_graph.node[node]['_ports'][
interface_id]['id'] = nidb_interface_id
id_brief = shortened_interface(nidb_interface_id)
nm_graph.node[node]['_ports'][
interface_id]['id_brief'] = id_brief
anm_json[overlay_id] = ank_json_dumps(nm_graph)
test_anm_data[overlay_id] = nm_graph
if nidb:
test_anm_data['nidb'] = prepare_nidb(nidb)
result = json.dumps(
test_anm_data, cls=AnkEncoder, indent=4, sort_keys=True)
return result