Mercurial > hg > drupalISMI

annotate importFromOpenMind/importer/compare_models.py @ 47:378dcb66a27f

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
new compare_models comparing the existence of nodes and relations in two graphs. fixed bugs in ismixml2model.
author casties
date Mon, 06 Feb 2017 18:44:43 +0100
parents
children 6625019a0c96
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
47
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
1 import networkx as nx
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
2 import sys
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
3 import csv
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
4
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
5 ## configure behaviour
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
6
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
7 # metworkx graph files
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
8 input1_fn = 'ismi_graph1.gpickle'
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
9 input2_fn = 'ismi_graph2.gpickle'
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
10
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
11 # name of type attribute
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
12 node_type_attribute = '_type'
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
13 rel_type_attribute = '_type'
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
14
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
15 # active log levels for logging
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
16 logLevels = {'DEBUG', 'INFO', 'WARNING', 'ERROR', 'SYSMSG'}
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
17 #logLevels = {'INFO', 'WARNING', 'ERROR', 'SYSMSG'}
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
18 #logLevels = {'INFO', 'ERROR', 'SYSMSG'}
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
19
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
20 def log(level, message):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
21 if level in logLevels:
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
22 print("%s: %s"%(level, message))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
23
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
24
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
25 def invertRelations(nx_graph):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
26 """Add inverse relations to each relation"""
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
27
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
28 print("Adding inverse relations.")
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
29 # copy list of edges because we add edges in the loop
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
30 edges = nx.edges(nx_graph)[:]
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
31 # iterate list
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
32 cnt = 0
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
33 for nx_edge in edges:
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
34 (nx_src, nx_tar) = nx_edge
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
35 # get attributes of edge
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
36 rel_attrs = nx_graph.edge[nx_src][nx_tar][0][:]
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
37 rel_type = rel_attrs[rel_type_attribute]
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
38 rel_id = rel_attrs['ismi_id']
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
39 # create new relation
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
40 rel_attrs[rel_type_attribute] = fixName(rel_type, is_tar_rel=True)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
41 rel_attrs['ismi_id': -rel_id]
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
42 nx_graph.add_edge(nx_tar, nx_src, attr_dict=invrel_atts)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
43
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
44 cnt += 1
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
45 if cnt % 100 == 0:
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
46 print(" %s relations"%cnt)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
47
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
48
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
49 def compare_nodes(nx_graph1, nx_graph2):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
50 """compare nodes of two graphs"""
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
51
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
52 log('INFO', "Compare graph nodes: %s vs %s"%(repr(nx_graph1), repr(nx_graph2)))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
53 cnt = 0
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
54 missing_nodes1 = []
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
55 missing_nodes2 = []
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
56 # iterate all nodes in graph 1
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
57 for n in nx.nodes_iter(nx_graph1):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
58 #attrs = nx_graph.node[n]
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
59
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
60 if not nx_graph2.has_node(n):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
61 missing_nodes2.append(n)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
62
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
63 if len(missing_nodes2) > 0:
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
64 log('WARNING', "%s nodes missing in graph 2"%len(missing_nodes2))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
65 log('DEBUG', "nodes: %s"%missing_nodes2)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
66 #log('DEBUG', "nodes: %s"%([nx_graph1.node[n] for n in missing_nodes2]))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
67
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
68 # iterate all nodes in graph 2
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
69 for n in nx.nodes_iter(nx_graph2):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
70 #attrs = nx_graph.node[n]
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
71
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
72 if not nx_graph1.has_node(n):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
73 missing_nodes1.append(n)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
74
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
75 if len(missing_nodes1) > 0:
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
76 log('WARNING', "%s nodes missing in graph 1"%len(missing_nodes1))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
77 log('DEBUG', "nodes: %s"%(missing_nodes1))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
78
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
79
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
80 def compare_relations(nx_graph1, nx_graph2):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
81 """compare relations of two graphs"""
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
82
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
83 log('INFO', "Compare graph relations: %s vs %s"%(repr(nx_graph1), repr(nx_graph2)))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
84 cnt = 0
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
85 missing_rels1 = []
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
86 missing_rels2 = []
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
87 # iterate all edges in graph 1
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
88 for (s, t) in nx.edges_iter(nx_graph1):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
89
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
90 if not nx_graph2.has_edge(s, t):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
91 missing_rels2.append((s,t))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
92
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
93 if len(missing_rels2) > 0:
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
94 log('WARNING', "%s relations missing in graph 2"%len(missing_rels2))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
95 log('DEBUG', "relations: %s"%missing_rels2)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
96 #log('DEBUG', "nodes: %s"%([nx_graph1.node[n] for n in missing_nodes2]))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
97
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
98 # iterate all nodes in graph 2
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
99 for (s, t) in nx.edges_iter(nx_graph2):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
100 #attrs = nx_graph.node[n]
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
101
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
102 if not nx_graph1.has_edge(s, t):
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
103 missing_rels1.append((s,t))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
104
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
105 if len(missing_rels1) > 0:
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
106 log('WARNING', "%s relations missing in graph 1"%len(missing_rels1))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
107 log('DEBUG', "relations: %s"%(missing_rels1))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
108
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
109
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
110 ## main
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
111
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
112 print("Modify networkx graph")
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
113
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
114 # read commandline parameters
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
115 if len(sys.argv) > 2:
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
116 input1_fn = sys.argv[1]
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
117 input2_fn = sys.argv[2]
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
118
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
119 # read networkx graph from pickle
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
120 print("Reading graph 1 from %s"%input1_fn)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
121 nx_graph1 = nx.read_gpickle(input1_fn)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
122 print("Graph 1 info: %s"%nx.info(nx_graph1))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
123
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
124 print("Reading graph 2 from %s"%input2_fn)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
125 nx_graph2 = nx.read_gpickle(input2_fn)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
126 print("Graph 2 info: %s"%nx.info(nx_graph2))
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
127
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
128 # operate
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
129 compare_nodes(nx_graph1, nx_graph2)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
130 compare_relations(nx_graph1, nx_graph2)
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
131
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
132
378dcb66a27f new compare_models comparing the existence of nodes and relations in two graphs.
casties
parents:
diff changeset
133 print("Done.")