Mercurial > hg > drupalISMI

comparison importFromOpenMind/importer/model2neo4j.py @ 23:45a823b5bf33

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
updated ismi2model importer and model2neo4j exporter.
author casties
date Wed, 23 Sep 2015 19:18:59 +0200
parents
children 5bdcb5805d29
comparison
equal deleted inserted replaced
22:3585573999b6 23:45a823b5bf33
1 import networkx as nx
2 from neo4jrestclient.client import GraphDatabase, Node
3 import sys
4
5 ## configure behaviour
6
7 # metworkx graph file
8 input_fn = 'ismi_graph.gpickle'
9
10 # label added to all nodes
11 project_label = '_ismi4'
12
13 # neo4j base URL
14 neo4jBaseURL = "http://localhost:7474/db/data/"
15
16
17 ## setup
18
19 n4j_nodes = {}
20 n4j_relations = {}
21
22
23 def fixName(name, is_src_rel=False, is_tar_rel=False, att_from_rel=False):
24 # these are too embarrassing...
25 if 'FLORUIT' in name:
26 name = name.replace('FLORUIT', 'FLOURISH')
27
28 elif 'floruit' in name:
29 name = name.replace('floruit', 'flourish')
30
31 if is_src_rel:
32 #name = name + '>'
33 pass
34
35 if is_tar_rel:
36 name = '<' + name
37
38 if att_from_rel:
39 # clean up relations as attribute names
40 name = name.replace('is_', '')
41 name = name.replace('has_', '')
42 name = name.replace('was_', '')
43 name = name.replace('_of', '')
44
45 return name
46
47
48 def copyNodes(nx_graph, n4j_graph):
49 """copy all nodes from nx_graph to n4j_graph"""
50
51 print("Copying nodes to Neo4J")
52 cnt = 0
53 for node_id in nx.nodes_iter(nx_graph):
54 attrs = nx_graph.node[node_id]
55 type = attrs['type']
56 ismi_id = attrs['ismi_id']
57 # create node with attributes
58 n4j_node = n4j_graph.nodes.create(**attrs)
59 # add labels
60 n4j_node.labels.add([project_label, type])
61 # save reference
62 n4j_nodes[ismi_id] = n4j_node
63
64 cnt += 1
65 if cnt % 100 == 0:
66 print(" %s"%cnt)
67
68
69 def copyRelations(nx_graph, n4j_graph):
70 """copy all relations from nx_graph to n4j_graph"""
71
72 print("Copying relations to Neo4J")
73 cnt = 0
74 for nx_edge in nx.edges_iter(nx_graph):
75 (nx_src, nx_tar) = nx_edge
76 # get attributes of edge
77 attrs = nx_graph.edge[nx_src][nx_tar][0]
78 type = attrs['type']
79 # get ismi_id of source and target nodes
80 src_id = nx_graph.node[nx_src]['ismi_id']
81 tar_id = nx_graph.node[nx_tar]['ismi_id']
82 # get Neo4J nodes
83 src = n4j_nodes.get(src_id, None)
84 if src is None:
85 print("ERROR: src node %s missing!"%src_id)
86 break
87
88 tar = n4j_nodes.get(tar_id, None)
89 if tar is None:
90 print("ERROR: tar node %s missing!"%tar_id)
91 break
92
93 # create Neo4J relation
94 n4j_rel = n4j_graph.relationships.create(src, type, tar)
95 # add attributes
96 n4j_rel.properties = attrs
97
98 cnt += 1
99 if cnt % 100 == 0:
100 print(" %s"%cnt)
101
102
103 ## main
104
105 print("Copy graph from networkx to Neo4J")
106
107 # read commandline parameters
108 if len(sys.argv) > 1:
109 input_fn = sys.argv[1]
110
111 # read networkx graph from pickle
112 print("Reading graph from %s"%input_fn)
113 nx_graph = nx.read_gpickle(input_fn)
114 print("Graph info: %s"%nx.info(nx_graph))
115
116 # open neo4j graph db
117 print("Opening Neo4J db at %s"%neo4jBaseURL)
118 n4j_graph = GraphDatabase(neo4jBaseURL, username="neo4j", password="neo5j")
119
120 copyNodes(nx_graph, n4j_graph)
121
122 copyRelations(nx_graph, n4j_graph)
123
124 print("Done.")