Mercurial > hg > drupalISMI

comparison importFromOpenMind/importer/compare_models.py @ 48:6625019a0c96

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
old model2neo4j renamed to model2neo4j_restclient. new model2neo4j_client and model2neo4j_import. fixed ismixml2model and compare_models.
author casties
date Tue, 07 Feb 2017 21:06:13 +0100
parents 378dcb66a27f
children
comparison
equal deleted inserted replaced
47:378dcb66a27f 48:6625019a0c96
1 import networkx as nx 1 import networkx as nx
2 import sys 2 import sys
3 import csv 3 import csv
4 from sqlalchemy.sql.expression import false
4 5
5 ## configure behaviour 6 ## configure behaviour
6 7
7 # metworkx graph files 8 # metworkx graph files
8 input1_fn = 'ismi_graph1.gpickle' 9 input1_fn = 'ismi_graph1.gpickle'
10 11
11 # name of type attribute 12 # name of type attribute
12 node_type_attribute = '_type' 13 node_type_attribute = '_type'
13 rel_type_attribute = '_type' 14 rel_type_attribute = '_type'
14 15
16 # also compare attributes
17 check_attributes = True
18 check_attribute_content = False
19
15 # active log levels for logging 20 # active log levels for logging
16 logLevels = {'DEBUG', 'INFO', 'WARNING', 'ERROR', 'SYSMSG'} 21 logLevels = {'DEBUG', 'INFO', 'WARNING', 'ERROR', 'SYSMSG'}
17 #logLevels = {'INFO', 'WARNING', 'ERROR', 'SYSMSG'} 22 #logLevels = {'INFO', 'WARNING', 'ERROR', 'SYSMSG'}
18 #logLevels = {'INFO', 'ERROR', 'SYSMSG'} 23 #logLevels = {'INFO', 'ERROR', 'SYSMSG'}
19 24
20 def log(level, message): 25 def log(level, message):
21 if level in logLevels: 26 if level in logLevels:
22 print("%s: %s"%(level, message)) 27 print("%s: %s"%(level, message))
23 28
24 29
25 def invertRelations(nx_graph): 30 def compare_attributes(attrs1, attrs2):
26 """Add inverse relations to each relation""" 31 """compare two sets of attributes"""
27 32 for a in attrs1.keys():
28 print("Adding inverse relations.") 33 if a.startswith('_n_'):
29 # copy list of edges because we add edges in the loop 34 continue
30 edges = nx.edges(nx_graph)[:] 35
31 # iterate list 36 if a not in attrs2:
32 cnt = 0 37 log('DEBUG', "attribute %s missing in attrs2"%a)
33 for nx_edge in edges: 38 return False
34 (nx_src, nx_tar) = nx_edge 39
35 # get attributes of edge 40 elif check_attribute_content:
36 rel_attrs = nx_graph.edge[nx_src][nx_tar][0][:] 41 val1 = attrs1[a]
37 rel_type = rel_attrs[rel_type_attribute] 42 val2 = attrs2[a]
38 rel_id = rel_attrs['ismi_id'] 43 if isinstance(val1, str):
39 # create new relation 44 val1 = val1.replace('\r', '')
40 rel_attrs[rel_type_attribute] = fixName(rel_type, is_tar_rel=True) 45 val2 = val2.replace('\r', '')
41 rel_attrs['ismi_id': -rel_id] 46
42 nx_graph.add_edge(nx_tar, nx_src, attr_dict=invrel_atts) 47 if val1 != val2:
48 log('DEBUG', "attribute %s different in attrs2: \n%s\n vs \n%s\n"%(a, repr(val1), repr(val2)))
49 return False
50
51 for a in attrs2.keys():
52 if a.startswith('_n_'):
53 continue
54
55 if a not in attrs1:
56 log('DEBUG', "attribute %s missing in attrs1"%a)
57 return False
58
59 return True
43 60
44 cnt += 1
45 if cnt % 100 == 0:
46 print(" %s relations"%cnt)
47
48
49 def compare_nodes(nx_graph1, nx_graph2): 61 def compare_nodes(nx_graph1, nx_graph2):
50 """compare nodes of two graphs""" 62 """compare nodes of two graphs"""
51 63
52 log('INFO', "Compare graph nodes: %s vs %s"%(repr(nx_graph1), repr(nx_graph2))) 64 log('INFO', "Compare graph nodes: %s vs %s"%(repr(nx_graph1), repr(nx_graph2)))
53 cnt = 0 65 cnt = 0
54 missing_nodes1 = [] 66 missing_nodes1 = []
55 missing_nodes2 = [] 67 missing_nodes2 = []
68 attribute_differences = []
56 # iterate all nodes in graph 1 69 # iterate all nodes in graph 1
57 for n in nx.nodes_iter(nx_graph1): 70 for n in nx.nodes_iter(nx_graph1):
58 #attrs = nx_graph.node[n]
59
60 if not nx_graph2.has_node(n): 71 if not nx_graph2.has_node(n):
61 missing_nodes2.append(n) 72 missing_nodes2.append(n)
73
74 else:
75 attrs1 = nx_graph1.node[n]
76 attrs2 = nx_graph2.node[n]
77 if check_attributes and not compare_attributes(attrs1, attrs2):
78 attribute_differences.append(n)
62 79
63 if len(missing_nodes2) > 0: 80 if len(missing_nodes2) > 0:
64 log('WARNING', "%s nodes missing in graph 2"%len(missing_nodes2)) 81 log('WARNING', "%s nodes missing in graph 2"%len(missing_nodes2))
65 log('DEBUG', "nodes: %s"%missing_nodes2) 82 log('DEBUG', "nodes: %s"%missing_nodes2)
66 #log('DEBUG', "nodes: %s"%([nx_graph1.node[n] for n in missing_nodes2])) 83 #log('DEBUG', "nodes: %s"%([nx_graph1.node[n] for n in missing_nodes2]))
73 missing_nodes1.append(n) 90 missing_nodes1.append(n)
74 91
75 if len(missing_nodes1) > 0: 92 if len(missing_nodes1) > 0:
76 log('WARNING', "%s nodes missing in graph 1"%len(missing_nodes1)) 93 log('WARNING', "%s nodes missing in graph 1"%len(missing_nodes1))
77 log('DEBUG', "nodes: %s"%(missing_nodes1)) 94 log('DEBUG', "nodes: %s"%(missing_nodes1))
95
96 if len(attribute_differences) > 0:
97 log('WARNING', "%s nodes with attribute differences"%len(attribute_differences))
98 log('DEBUG', "nodes: %s"%(attribute_differences))
78 99
79 100
80 def compare_relations(nx_graph1, nx_graph2): 101 def compare_relations(nx_graph1, nx_graph2):
81 """compare relations of two graphs""" 102 """compare relations of two graphs"""
82 103