def test_compute_rolling_edge(self):
self.graph.add_edge("spark","apache",Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:01 +0000 2015"))
self.assertEqual(self.graph.compute_rolling_average(),1.0)
epoch = Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:30 +0000 2015")
self.graph.add_edge("apache","hadoop",epoch)
self.graph.add_edge("apache","storm",epoch)
self.graph.add_edge("hadoop","storm",epoch)
self.assertEqual(self.graph.compute_rolling_average(),2.0)
self.graph.add_edge("flink","spark",Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:56 +0000 2015"))
self.assertEqual(self.graph.compute_rolling_average(),2.0)
print "Check if rolling average degree in compute_rolling_edge function is computed correctly [OK]"
def test_extract_tweet_info_success(self):
#Check successful information extraction
line = '{"created_at":"Thu Oct 29 17:51:01 +0000 2015","text":"Spark Summit East this week! #Spark #Apache","entities":{"hashtags":[{"text":"Spark"},{"text":"Apache"}]}}'
hashtags,epoch = self.graph.extract_tweet_info(self.graph.read_json_data(line))
self.assertEquals(hashtags,set([u'apache',u'spark']))
self.assertEquals(epoch,Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:01 +0000 2015"))
print "Check if extract_tweet_info function is executing correctly [OK]"
def test_add_neighbor_execute_correctly(self):
epoch = Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:01 +0000 2015")
self.node3.add_neighbor(self.node4.get_hashtag(),epoch)
#Check if a node 3 is neighbor of node 4
self.assertIn(self.node4.get_hashtag(),self.node3.get_adjacency_structure())
#Check if list adjacent of node 3 has correct timestamp
self.assertIn(epoch,self.node3.get_adjacency_structure()[self.node4.get_hashtag()])
print "Check if add_neighbor function is executing correctly [OK]"
def test_time_window(self):
#Check window when the first new tweet arrives - update_time_window
epoch_first = Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:01 +0000 2015")
self.assertFalse(self.graph.check_time_window(epoch_first))
self.assertEqual(epoch_first,self.graph.first_timestamp_window)
self.assertEqual(epoch_first,self.graph.last_timestamp_window)
self.graph.add_node("a")
#Check window when new tweet does no modify time window
epoch = Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:10 +0000 2015")
self.assertFalse(self.graph.check_time_window(epoch))
self.assertEqual(epoch_first,self.graph.first_timestamp_window)
self.assertEqual(epoch,self.graph.last_timestamp_window)
self.graph.add_node("b")
#Check window when new tweet does no modify time window
epoch = Graph.convert_timestamp_to_epoch("Thu Oct 29 17:53:10 +0000 2015")
self.assertTrue(self.graph.check_time_window(epoch))
self.assertEqual(epoch_first,self.graph.first_timestamp_window)
self.assertEqual(epoch,self.graph.last_timestamp_window)
print "Check if check_time_window function is computed correctly [OK]"
def test_graph_structure_update_window(self):
tweets = ['{"created_at":"Mon Nov 2 00:33:01 +0000 2015","entities":{"hashtags":[{"text":"a"},{"text":"b"}]}}','{"created_at":"Mon Nov 2 00:33:20 +0000 2015","entities":{"hashtags":[{"text":"b"},{"text":"a"}]}}','{"created_at":"Mon Nov 2 00:33:30 +0000 2015","entities":{"hashtags":[{"text":"a"},{"text":"c"}]}}','{"created_at":"Mon Nov 2 00:33:58 +0000 2015","entities":{"hashtags":[{"text":"e"},{"text":"f"}]}}','{"created_at":"Mon Nov 2 00:34:05 +0000 2015","entities":{"hashtags":[{"text":"g"},{"text":"h"}]}}','{"created_at":"Mon Nov 2 00:34:31 +0000 2015","entities":{"hashtags":[{"text":"e"},{"text":"a"}]}}']
for tweet in tweets:
self.graph.process_data(self.graph.read_json_data(tweet))
self.assertEqual(self.graph.first_timestamp_window,Graph.convert_timestamp_to_epoch("Mon Nov 2 00:33:58 +0000 2015"))
self.assertIsNot("b",self.graph.get_nodes())
self.assertIsNot("b",self.graph.get_node("a").get_adjacency_structure())
self.assertIsNot("c",self.graph.get_nodes())
self.assertIsNot("c",self.graph.get_node("a").get_adjacency_structure())
print "Check if a graph was update with tweet window change [OK]"
def test_graph_structure_window(self):
self.graph.process_data(self.graph.read_json_data('{"created_at":"Mon Nov 2 00:33:01 +0000 2015","entities":{"hashtags":[{"text":"a"},{"text":"b"}]}}'))
self.graph.process_data(self.graph.read_json_data('{"created_at":"Mon Nov 2 00:33:20 +0000 2015","entities":{"hashtags":[{"text":"b"},{"text":"a"}]}}'))
self.graph.process_data(self.graph.read_json_data('{"created_at":"Mon Nov 2 00:33:30 +0000 2015","entities":{"hashtags":[{"text":"a"},{"text":"c"}]}}'))
self.graph.process_data(self.graph.read_json_data('{"created_at":"Mon Nov 2 00:33:58 +0000 2015","entities":{"hashtags":[{"text":"e"},{"text":"f"}]}}'))
self.graph.process_data(self.graph.read_json_data('{"created_at":"Mon Nov 2 00:34:05 +0000 2015","entities":{"hashtags":[{"text":"g"},{"text":"h"}]}}'))
epoch = Graph.convert_timestamp_to_epoch("Mon Nov 2 00:33:20 +0000 2015")
self.assertEqual(self.graph.first_timestamp_window,epoch)
self.assertEqual(len(self.graph.get_node("a").get_adjacency_structure()["b"]),1)
self.assertEqual(self.graph.get_node("a").get_adjacency_structure()["b"][0],epoch)
print "Check if update_graph_structure_window function is computed correctly [OK]"
def test_create_graph_by_file(self):
lines ='{"created_at":"Thu Oct 29 17:51:01 +0000 2015","entities":{"hashtags":[{"text":"Spark"},{"text":"Apache"}]}}\n{"created_at":"Thu Oct 29 17:51:30 +0000 2015","entities":{"hashtags":[{"text":"Apache"},{"text":"Hadoop"},{"text":"Storm"}]}}\n{"created_at":"Thu Oct 29 17:51:55 +0000 2015","entities":{"hashtags":[{"text":"Apache"}]}}\n{"created_at":"Thu Oct 29 17:51:56 +0000 2015","entities":{"hashtags":[{"text":"Flink"},{"text":"Spark"}]}}\n{"created_at":"Thu Oct 29 17:51:59 +0000 2015","entities":{"hashtags":[{"text":"HBase"},{"text":"Spark"}]}}\n{"created_at":"Thu Oct 29 17:52:05 +0000 2015","entities":{"hashtags":[{"text":"Hadoop"},{"text":"Apache"}]}}'
current_path = os.getcwd().replace("src","")
#Create input file
with open(current_path + "/tweet_input/test_tweets.txt",'w') as input:
input.write(lines)
# Create graph using the created input file
self.graph.create_graph_by_file(current_path + "/tweet_input/test_tweets.txt",current_path + "/tweet_output/test_tweets_out.txt")
result = self.print_graph()
#Create correct graph
correct_graph = Graph()
epoch = Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:30 +0000 2015")
correct_graph.add_edge("apache","hadoop",epoch)
correct_graph.add_edge("apache","storm",epoch)
correct_graph.add_edge("hadoop","storm",epoch)
correct_graph.add_edge("apache","hadoop",Graph.convert_timestamp_to_epoch("Thu Oct 29 17:52:05 +0000 2015"))
correct_graph.add_edge("hbase","spark",Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:59 +0000 2015"))
correct_graph.add_edge("flink","spark",Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:56 +0000 2015"))
correct_result = self.print_graph()
# Check if the two graphs are equal
self.assertEqual(result,correct_result)
#Check output file is correct
correct_output = ["1.00","2.00","2.00","2.00","2.00","1.67"]
i = 0
with open(current_path + "/tweet_output/test_tweets_out.txt") as output_file:
for line in output_file:
self.assertEqual(line.strip().split()[0],correct_output[i])
i+=1
# Delete test files
os.remove(current_path + "/tweet_input/test_tweets.txt")
os.remove(current_path + "/tweet_output/test_tweets_out.txt")
print "Check the graph and the output file produced by tweets file [OK]"
def test_convert_timestamp_to_epoch(self):
with self.assertRaises(ValueError) as context:
Graph.convert_timestamp_to_epoch("this is a test")
msg = "Timestamp has a incorrect format"
self.assertTrue(msg in context.exception)
print "Check convert_timestamp_to_epoch function with wrong timestamp format [OK]"
def create_edge(self,ht1,ht2,timestamp): epoch = Graph.convert_timestamp_to_epoch(timestamp) self.graph.add_edge(ht1,ht2,epoch) return epoch
def test_extract_tweet_info_success_no_hashtag_field(self):
line = '{"created_at":"Thu Oct 29 17:51:01 +0000 2015","entities":{}}'
hashtags,epoch = self.graph.extract_tweet_info(self.graph.read_json_data(line))
self.assertEquals(hashtags,set())
self.assertEquals(epoch,Graph.convert_timestamp_to_epoch("Thu Oct 29 17:51:01 +0000 2015"))
print "Check if extract_tweet_info function is executing with tweets with no hashtags field [OK]"
