def create_graph(GRAPHDB):
# Do we have a node that has a 'name' property, which is set to the value 'Neo'?
# We've probably been here before.
data, metadata = cypher.execute(GRAPHDB, "START n=node(*) where n.name='Neo' return n")
if not data:
# Create two nodes, one for us and one for you.
# Make sure they both have 'name' properties with values.
from_node, to_node = GRAPHDB.create({"name": "Neo"}, {"name": "you"})
# create a 'loves' relationship from the 'from' node to the 'to' node
GRAPHDB.create((from_node, "loves", to_node),)
def saveToNeo4j(found_pages, found_links):
print "++: saving to neo4j"
title_index = GRAPHDB.get_or_create_index(neo4j.Node, "TitleIndex")
i=0
for link in found_links:
pageA, pageB = link
nodeA = title_index.get_or_create("title", pageA, {"title": pageA})
nodeB = title_index.get_or_create("title", pageB, {"title": pageB})
GRAPHDB.create((nodeA, "links_to", nodeB))
if not i % 100:
print i
i += 1
def saveToNeo4jBatch(found_pages, found_links):
url_index = GRAPHDB.get_or_create_index(neo4j.Node, "UrlIndex")
pageToNode = {}
for page in found_pages:
name = getNameFromLink(page)
node = GRAPHDB.create({"name":name, "url":page})[0]
# TODO: add labels based on infobox
pageToNode[page] = node
# save links
i = 0
batch = neo4j.WriteBatch(GRAPHDB)
for link in found_links:
pageA, pageB = link
nodeA = pageToNode.get(pageA)
nodeB = pageToNode.get(pageB)
batch.get_or_create_path(nodeA, "links_to", nodeB)
if not i % 100:
print "i: " + str(i)
batch.run()
batch = neo4j.WriteBatch(GRAPHDB)
i += 1
batch.run()
print "total num links created: " + str(i)
def handle_row_inner(inner_row):
nodeB = inner_row[0]
weight = random.random()
print "w: " + str(weight)
if nodeA != nodeB:
GRAPHDB.create((nodeA, "RELEVANCY", nodeB, {"weight": weight}),)