Py2neo Write a report on the contents of a graph database, describe what nodes are in, what sort of properties they (seem to) have, and what edges come in and out of each

database_report.py

import py2neo
import datetime

#where we write it
f_name = 'DBREPORT_%s.txt' % datetime.datetime.today().strftime('%Y-%m-%d')


#overwrite anything previous
with open(f_name,'wb') as f:
	f.write('REPORT COMPILATION STARTED AT %s' %  datetime.datetime.now())



def doline(line):
	'''writes a line, also prints it'''
	l = ''.join(line)
	with open(f_name,'ab+') as f:
		f.write(l)
	print l

def describe_nodeset(labels, graph):
	'''yields lines to describe stuff on a particular set of nodes'''
	labs = ':'.join(labels)
	props = graph.cypher.execute('MATCH (n:%s)  return n limit 1' % labs)[0].n.properties
	yield 'Arbitrary Node has properties:'
	for i,j in props.items():
		#truncate for readiblity
		if len(`j`) > 20:
			yield '\t',`i`,':',`j`[:20]
		else:
			yield '\t',`i`,':',`j`
	edges = graph.cypher.execute('''MATCH (n:%s) 
							OPTIONAL MATCH (n)-[r]->(m)
							RETURN 
								type(r) as type,
								labels(m) as target,
								count(distinct r) as count 
						'''% labs)
	yield 'Nodeset has outbound edges:'
	yield '\t|edgetype|target_labset|count(rels)'
	for i in edges:
		if i.count > 0:
			yield '\t',`i.type`, '|',`':'.join(list(i.target or ['NO LABELS']))`, '|',`i.count`
	edges = graph.cypher.execute('''MATCH (n:%s) 
							OPTIONAL MATCH (n)<-[r]-(m)
							RETURN 
								type(r) as type,
								labels(m) as target,
								count(distinct r) as count 
						'''% labs)
	yield 'Nodeset has inbound edges:'
	yield '\t|edgetype|source_labset|count(rels)'
	for i in edges:
		if i.count > 0:
			yield '\t',`i.type`, '|',`':'.join(list(i.target or ['NO LABELS']))`, '|',`i.count`



def report_nodes(graph):
	'''we identify sets of nodes as having the same SET of labels
		for each such set, we give some basic stats about these things
		within your graph'''
	nodes = graph.cypher.execute('MATCH (n) return labels(n) as labset, count(distinct n) as count')
	for i in nodes:
		yield 'Labels: %s, count %d' % (':'.join(i.labset or ['NO LABELS']), i.count)
		for line in describe_nodeset(i.labset, graph):
			yield line


def write_report(graph):
	doline('WRITING REPORT FOR DATABASE %s' % graph.uri)
	for i in report_nodes(graph):
		doline(i)



if __name__ == '__main__':
	#usage:
	#   python database_report.py local
	#-will produce and print out a full report
	#-on the database specified at graphs['local']
	#
	#or do a report on a different graph by providing the uri to that graph
	#as an argument:
	#   python database_report.py http://neo4j:neo4j@localhost:7474/db/data/
	graphs = {
	    'local':'http://neo4j:neo4j@localhost:7474/db/data/',
	    'live':'https://<USERNAME>:<PASSWORD>@<DB_URL>/db/data',
	}
	#populate graphs with a set of neos you care about
	from sys import argv
	graph = py2neo.Graph(graphs.get(argv[1], argv[1]))
	write_report(graph)