def setUp(self):
MongoDBTestCase.setUp(self)
# Generate some fake workload sessions
for i in xrange(0, NUM_SESSIONS):
sess = self.metadata_db.Session()
sess['session_id'] = i
sess['ip_client'] = "client:%d" % (1234+i)
sess['ip_server'] = "server:5678"
sess['start_time'] = time.time()
sess['end_time'] = time.time() + 5
for j in xrange(0, NUM_OPS_PER_SESSION):
_id = str(random.random())
queryId = long((i<<16) + j)
queryContent = { constants.REPLACE_KEY_DOLLAR_PREFIX + "query":
{ "_id": [ _id ] } }
responseContent = {"_id": _id}
responseId = (queryId<<8)
for f in xrange(0, NUM_FIELDS):
f_name = "field%02d" % f
if f % 2 == 0:
responseContent[f_name] = random.randint(0, 100)
else:
responseContent[f_name] = str(random.randint(1000, 100000))
## FOR
op = workload.Session.operationFactory()
op['collection'] = COLLECTION_NAME
op['type'] = constants.OP_TYPE_QUERY
op['query_id'] = queryId
op['query_content'] = [ queryContent ]
op['resp_content'] = [ responseContent ]
op['resp_id'] = responseId
sess['operations'].append(op)
## FOR (ops)
sess.save()
## FOR (sess)
self.assertEqual(NUM_SESSIONS, self.metadata_db.Session.find().count())
self.reconstructor = Reconstructor(self.metadata_db, self.dataset_db)
def testPostProcess(self):
"""
Check whether we can successfully extract the database schema
into our internal catalog
"""
Reconstructor(self.metadata_db, self.dataset_db).process()
self.converter.postProcess()
col_info = self.metadata_db.Collection.one({"name": COLLECTION_NAME})
# pprint(col_info)
# Workload-derived Attributes
self.assertEqual(NUM_SESSIONS*NUM_OPS_PER_SESSION, col_info['workload_queries'])
self.assertAlmostEqual(1.0, col_info['workload_percent'])
# Fields
# Add one for the '_id' field count
self.assertEqual(NUM_FIELDS + 1, len(col_info['fields']))
for k,field in col_info['fields'].iteritems():
self.assertEqual(NUM_SESSIONS*NUM_OPS_PER_SESSION, field['query_use_count'])
class TestReconstructor(MongoDBTestCase):
def setUp(self):
MongoDBTestCase.setUp(self)
# Generate some fake workload sessions
for i in xrange(0, NUM_SESSIONS):
sess = self.metadata_db.Session()
sess['session_id'] = i
sess['ip_client'] = "client:%d" % (1234+i)
sess['ip_server'] = "server:5678"
sess['start_time'] = time.time()
sess['end_time'] = time.time() + 5
for j in xrange(0, NUM_OPS_PER_SESSION):
_id = str(random.random())
queryId = long((i<<16) + j)
queryContent = { constants.REPLACE_KEY_DOLLAR_PREFIX + "query":
{ "_id": [ _id ] } }
responseContent = {"_id": _id}
responseId = (queryId<<8)
for f in xrange(0, NUM_FIELDS):
f_name = "field%02d" % f
if f % 2 == 0:
responseContent[f_name] = random.randint(0, 100)
else:
responseContent[f_name] = str(random.randint(1000, 100000))
## FOR
op = workload.Session.operationFactory()
op['collection'] = COLLECTION_NAME
op['type'] = constants.OP_TYPE_QUERY
op['query_id'] = queryId
op['query_content'] = [ queryContent ]
op['resp_content'] = [ responseContent ]
op['resp_id'] = responseId
sess['operations'].append(op)
## FOR (ops)
sess.save()
## FOR (sess)
self.assertEqual(NUM_SESSIONS, self.metadata_db.Session.find().count())
self.reconstructor = Reconstructor(self.metadata_db, self.dataset_db)
## DEF
def testProcess(self):
"""
Check whether the reconstructed database includes the fields
that were returned in the query responses
"""
self.reconstructor.process()
fields = [ "field%02d" % f for f in xrange(0, NUM_FIELDS) ]
num_docs = 0
for doc in self.dataset_db[COLLECTION_NAME].find():
# pprint(doc)
for f in fields:
self.assertIn(f, doc.keys())
num_docs += 1
# We should always have one per operation
self.assertEquals(num_docs, NUM_SESSIONS * NUM_OPS_PER_SESSION)
