1 | """Module Graph of kytos/pathfinder Kytos Network Application.""" |
||
2 | |||
3 | 1 | from itertools import combinations |
|
4 | |||
5 | 1 | from kytos.core import log |
|
6 | |||
7 | 1 | try: |
|
8 | 1 | import networkx as nx |
|
9 | 1 | from networkx.exception import NodeNotFound, NetworkXNoPath |
|
10 | except ImportError: |
||
11 | PACKAGE = 'networkx>=2.2' |
||
12 | log.error(f"Package {PACKAGE} not found. Please 'pip install {PACKAGE}'") |
||
13 | |||
14 | |||
15 | 1 | class Filter: |
|
16 | """Class responsible for removing items with disqualifying values.""" |
||
17 | |||
18 | 1 | def __init__(self, filter_type, filter_function): |
|
19 | 1 | self._filter_type = filter_type |
|
20 | 1 | self._filter_function = filter_function |
|
21 | |||
22 | 1 | def run(self, value, items): |
|
23 | """Filter out items. Filter chosen is picked at runtime.""" |
||
24 | 1 | if isinstance(value, self._filter_type): |
|
25 | 1 | return filter(self._filter_function(value), items) |
|
26 | |||
27 | 1 | raise TypeError(f"Expected type: {self._filter_type}") |
|
28 | |||
29 | |||
30 | 1 | class KytosGraph: |
|
31 | """Class responsible for the graph generation.""" |
||
32 | |||
33 | 1 | def __init__(self): |
|
34 | 1 | self.graph = nx.Graph() |
|
35 | 1 | self._filter_functions = {} |
|
36 | |||
37 | 1 | def filter_leq(metric): # Lower values are better |
|
38 | 1 | return lambda x: (lambda y: y[2].get(metric, x) <= x) |
|
0 ignored issues
–
show
Comprehensibility
Best Practice
introduced
by
![]() |
|||
39 | |||
40 | 1 | def filter_geq(metric): # Higher values are better |
|
41 | 1 | return lambda x: (lambda y: y[2].get(metric, x) >= x) |
|
0 ignored issues
–
show
Comprehensibility
Best Practice
introduced
by
|
|||
42 | |||
43 | 1 | def filter_eeq(metric): # Equivalence |
|
44 | 1 | return lambda x: (lambda y: y[2].get(metric, x) == x) |
|
0 ignored issues
–
show
Comprehensibility
Best Practice
introduced
by
|
|||
45 | |||
46 | 1 | self._filter_functions["ownership"] = Filter( |
|
47 | str, filter_eeq("ownership")) |
||
48 | 1 | self._filter_functions["bandwidth"] = Filter( |
|
49 | (int, float), filter_geq("bandwidth")) |
||
50 | 1 | self._filter_functions["priority"] = Filter( |
|
51 | (int, float), filter_geq("priority")) |
||
52 | 1 | self._filter_functions["reliability"] = Filter( |
|
53 | (int, float), filter_geq("reliability")) |
||
54 | 1 | self._filter_functions["utilization"] = Filter( |
|
55 | (int, float), filter_leq("utilization")) |
||
56 | 1 | self._filter_functions["delay"] = Filter( |
|
57 | (int, float), filter_leq("delay")) |
||
58 | 1 | self._path_function = nx.all_shortest_paths |
|
59 | |||
60 | 1 | def clear(self): |
|
61 | """Remove all nodes and links registered.""" |
||
62 | 1 | self.graph.clear() |
|
63 | |||
64 | 1 | def update_topology(self, topology): |
|
65 | """Update all nodes and links inside the graph.""" |
||
66 | 1 | self.graph.clear() |
|
67 | 1 | self.update_nodes(topology.switches) |
|
68 | 1 | self.update_links(topology.links) |
|
69 | |||
70 | 1 | def update_nodes(self, nodes): |
|
71 | """Update all nodes inside the graph.""" |
||
72 | 1 | for node in nodes.values(): |
|
73 | 1 | try: |
|
74 | 1 | self.graph.add_node(node.id) |
|
75 | |||
76 | 1 | for interface in node.interfaces.values(): |
|
77 | 1 | self.graph.add_node(interface.id) |
|
78 | 1 | self.graph.add_edge(node.id, interface.id) |
|
79 | |||
80 | except AttributeError: |
||
81 | pass |
||
82 | |||
83 | 1 | def update_links(self, links): |
|
84 | """Update all links inside the graph.""" |
||
85 | 1 | keys = [] |
|
86 | 1 | for link in links.values(): |
|
87 | 1 | if link.is_active(): |
|
88 | 1 | self.graph.add_edge(link.endpoint_a.id, link.endpoint_b.id) |
|
89 | 1 | for key, value in link.metadata.items(): |
|
90 | 1 | keys.append(key) |
|
91 | 1 | endpoint_a = link.endpoint_a.id |
|
92 | 1 | endpoint_b = link.endpoint_b.id |
|
93 | 1 | self.graph[endpoint_a][endpoint_b][key] = value |
|
94 | |||
95 | 1 | self._set_default_metadata(keys) |
|
96 | |||
97 | 1 | def _set_default_metadata(self, keys): |
|
98 | """Set metadata to all links. |
||
99 | |||
100 | Set the value to zero for inexistent metadata in a link to make those |
||
101 | irrelevant in pathfinding. |
||
102 | """ |
||
103 | 1 | for key in keys: |
|
104 | 1 | for endpoint_a, endpoint_b in self.graph.edges: |
|
105 | 1 | if key not in self.graph[endpoint_a][endpoint_b]: |
|
106 | 1 | self.graph[endpoint_a][endpoint_b][key] = 0 |
|
107 | |||
108 | 1 | def get_metadata_from_link(self, endpoint_a, endpoint_b): |
|
109 | """Return the metadata of a link.""" |
||
110 | 1 | return self.graph.edges[endpoint_a, endpoint_b] |
|
111 | |||
112 | 1 | @staticmethod |
|
113 | def _remove_switch_hops(circuit): |
||
114 | """Remove switch hops from a circuit hops list.""" |
||
115 | 1 | for hop in circuit['hops']: |
|
116 | 1 | if len(hop.split(':')) == 8: |
|
117 | 1 | circuit['hops'].remove(hop) |
|
118 | |||
119 | 1 | def shortest_paths(self, source, destination, parameter=None): |
|
120 | """Calculate the shortest paths and return them.""" |
||
121 | 1 | try: |
|
122 | 1 | paths = list(nx.shortest_simple_paths(self.graph, |
|
123 | source, destination, |
||
124 | parameter)) |
||
125 | except (NodeNotFound, NetworkXNoPath): |
||
126 | return [] |
||
127 | 1 | return paths |
|
128 | |||
129 | 1 | def constrained_flexible_paths(self, source, destination, |
|
130 | maximum_misses=None, **metrics): |
||
131 | """Calculate the constrained shortest paths with flexibility.""" |
||
132 | 1 | base = metrics.get("base", {}) |
|
133 | 1 | flexible = metrics.get("flexible", {}) |
|
134 | 1 | default_edge_list = list(self._filter_edges( |
|
135 | self.graph.edges(data=True), **base)) |
||
136 | 1 | length = len(flexible) |
|
137 | 1 | if maximum_misses is None: |
|
138 | 1 | maximum_misses = length |
|
139 | 1 | maximum_misses = min(length, max(0, maximum_misses)) |
|
140 | 1 | results = [] |
|
141 | 1 | paths = [] |
|
142 | 1 | i = 0 |
|
143 | 1 | while (paths == [] and i in range(0, maximum_misses+1)): |
|
144 | 1 | for combo in combinations(flexible.items(), length-i): |
|
145 | 1 | additional = dict(combo) |
|
146 | 1 | paths = self._constrained_shortest_paths( |
|
147 | source, destination, ((u, v) for u, v, d in |
||
0 ignored issues
–
show
|
|||
148 | self._filter_edges(default_edge_list, |
||
149 | **additional))) |
||
150 | 1 | if paths != []: |
|
151 | 1 | results.append( |
|
152 | {"paths": paths, "metrics": {**base, **additional}}) |
||
153 | 1 | i = i + 1 |
|
154 | 1 | return results |
|
155 | |||
156 | 1 | def _constrained_shortest_paths(self, source, destination, edges): |
|
157 | 1 | paths = [] |
|
158 | 1 | try: |
|
159 | 1 | paths = list(self._path_function(self.graph.edge_subgraph(edges), |
|
160 | source, destination)) |
||
161 | 1 | except NetworkXNoPath: |
|
162 | 1 | pass |
|
163 | 1 | except NodeNotFound: |
|
164 | 1 | if source == destination: |
|
165 | 1 | if source in self.graph.nodes: |
|
166 | 1 | paths = [[source]] |
|
167 | 1 | return paths |
|
168 | |||
169 | 1 | def _filter_edges(self, edges, **metrics): |
|
170 | 1 | for metric, value in metrics.items(): |
|
171 | 1 | filter_ = self._filter_functions.get(metric, None) |
|
172 | 1 | if filter_ is not None: |
|
173 | 1 | try: |
|
174 | 1 | edges = filter_.run(value, edges) |
|
175 | 1 | except TypeError as err: |
|
176 | 1 | raise TypeError(f"Error in {metric} value: {err}") |
|
177 | return edges |
||
178 |