def sample_remote_indexes():
print()
print("SimIndexCollection with remote backend indexes")
processes = []
for i in range(2):
port = 9000 + i
process = Process(target=sim_server.start_sim_index_server,
kwargs={'port': port, 'logRequests': False})
process.daemon = True
processes.append(process)
for process in processes:
process.start()
print("Waiting for servers to start")
time.sleep(1)
remote_index_coll = SimIndexCollection()
for i in range(2):
port = 9000 + i
remote_index_coll.add_shards(
RemoteSimIndex("http://localhost:{}/RPC2".format(port)))
remote_index_coll.set_query_scorer('tfidf')
remote_index_coll.index_urls('http://www.stanford.edu/',
'http://www.berkeley.edu',
'http://www.ucla.edu',
'http://www.mit.edu')
pprint(remote_index_coll.query('stanford university'))
for process in processes:
process.terminate()
def sample_sim_index_collection():
# SimIndexCollection
print()
print("SimIndexCollection: build a collection, index some urls, and query it")
indexes = (MemorySimIndex(), MemorySimIndex())
index_coll = SimIndexCollection()
index_coll.add_shards(*indexes)
index_coll.set_query_scorer('tfidf')
index_coll.index_urls('http://www.stanford.edu/',
'http://www.berkeley.edu',
'http://www.ucla.edu',
'http://www.mit.edu')
pprint(index_coll.query('stanford university'))
def sample_sim_index_collection():
# SimIndexCollection
print()
print(
"SimIndexCollection: build a collection, index some urls, and query it"
)
indexes = (MemorySimIndex(), MemorySimIndex())
index_coll = SimIndexCollection()
index_coll.add_shards(*indexes)
index_coll.set_query_scorer('tfidf')
index_coll.index_urls('http://www.stanford.edu/',
'http://www.berkeley.edu', 'http://www.ucla.edu',
'http://www.mit.edu')
pprint(index_coll.query('stanford university'))
class SimIndexCollectionTest(SimIndexTest, unittest.TestCase):
'''
All tests hitting the SimIndex interface are in the parent class, SimIndexTest
Tests for api's not in parent class are tested separately here. This is
so we can reuse test code across all implementations of SimIndex.
'''
def setUp(self):
print("SimIndexCollectionTest")
self.sim_index = SimIndexCollection()
for i in range(2):
self.sim_index.add_shards(MemorySimIndex())
super(SimIndexCollectionTest, self).setUp()
def tearDown(self):
pass
def sample_remote_indexes():
print()
print("SimIndexCollection with remote backend indexes")
processes = []
for i in range(2):
port = 9000 + i
process = Process(target=sim_server.start_sim_index_server,
kwargs={
'port': port,
'logRequests': False
})
process.daemon = True
processes.append(process)
for process in processes:
process.start()
print("Waiting for servers to start")
time.sleep(1)
remote_index_coll = SimIndexCollection()
for i in range(2):
port = 9000 + i
remote_index_coll.add_shards(
RemoteSimIndex("http://localhost:{}/RPC2".format(port)))
remote_index_coll.set_query_scorer('tfidf')
remote_index_coll.index_urls('http://www.stanford.edu/',
'http://www.berkeley.edu',
'http://www.ucla.edu', 'http://www.mit.edu')
pprint(remote_index_coll.query('stanford university'))
for process in processes:
process.terminate()
class SimIndexRemoteCollectionTest(SimIndexTest, unittest.TestCase):
'''
All tests hitting the SimIndex interface are in the parent class, SimIndexTest
Tests for api's not in parent class are tested separately here. This is
so we can reuse test code across all implementations of SimIndex.
'''
processes = None
def setUp(self):
# setUpClass() may be more efficient for spinning up the servers,
# but this way is more robust (since we'll start each test from a
# clean slate). Otherwise we'd need clear() functionality added.
print("SimIndexRemoteCollectionTest")
# We will create a collection tree of the form:
#
# Root
# / \
# A B
# /\ /\
# 1 2 3 4
self.processes = []
# start leaves
for i in range(4):
port = 9100 + i
process = Process(target=sim_server.start_sim_index_server,
kwargs={'port': port, 'logRequests': False})
process.daemon = True
process.start()
self.processes.append(process)
print("Waiting for leaf servers to start")
time.sleep(0.1)
leaf_nodes = [[],[]]
for i in range(4):
port = 9100 + i
leaf_nodes[i//2].append(RemoteSimIndex(
"http://localhost:{}/RPC2".format(port)))
# start interior nodes (A, B)
for i in range(2):
port = 9200 + i
process = Process(
target=sim_server.start_sim_index_server,
kwargs={ 'port': port,
'backends': leaf_nodes[i],
'root': False,
'logRequests': False
}
)
process.daemon = True
process.start()
self.processes.append(process)
print("Waiting for intermediate servers to start")
time.sleep(0.1)
interior_nodes = []
for i in range(2):
port = 9200 + i
interior_nodes.append(
RemoteSimIndex("http://localhost:{}/RPC2".format(port)))
# root node
self.sim_index = SimIndexCollection(root=True)
self.sim_index.add_shards(*interior_nodes)
super(SimIndexRemoteCollectionTest, self).setUp()
def tearDown(self):
for process in self.processes:
process.terminate()
time.sleep(0.1)