def test_message_number_bins_csv(self, mock_util):
mock_util.correctLastCharCR.side_effect = util.load_from_disk(
self.test_data_dir + "message_number_bins_csv/correctLastCharCR")
mock_util.correct_last_char_list.side_effect = util.load_from_disk(
self.test_data_dir +
"message_number_bins_csv/correct_last_char_list")
mock_util.rec_list_splice.side_effect = util.load_from_disk(
self.test_data_dir + "message_number_bins_csv/rec_list_splice")
bin_matrix_ = util.load_from_disk(self.test_data_dir +
"message_number_bins_csv/bin_matrix")
tot_msgs_ = util.load_from_disk(self.test_data_dir +
"message_number_bins_csv/tot_msgs")
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
bin_matrix, tot_msgs = network.message_number_bins_csv(
self.log_data, self.nicks, self.nick_same_list)
sys.stdout = sys.__stdout__
capturedOutput.close()
self.assertEqual(bin_matrix, bin_matrix_)
self.assertEqual(tot_msgs, tot_msgs_)
def setUp(self):
self.current_directory = os.path.dirname(os.path.realpath(__file__))
self.test_data_dir = self.current_directory + "/../../../../data/test_slack/test_lib/test_analysis/network_test/"
self.log_data = util.load_from_disk(self.test_data_dir + "../log_data")
self.nicks = util.load_from_disk(self.test_data_dir + "../nicks")
self.nick_same_list = util.load_from_disk(self.test_data_dir +
"../nick_same_list")
def test_degree_node_number_csv(self, mock_msg_graph):
msg_num_graph_day_list = util.load_from_disk(
self.test_data_dir + "degree_node_number_csv/msg_day_list")
out_degree_ = util.load_from_disk(self.test_data_dir +
"degree_node_number_csv/out_degree")
in_degree_ = util.load_from_disk(self.test_data_dir +
"degree_node_number_csv/in_degree")
total_degree_ = util.load_from_disk(
self.test_data_dir + "degree_node_number_csv/total_degree")
mock_msg_graph.return_value = msg_num_graph_day_list
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
out_degree, in_degree, total_degree = network.degree_node_number_csv(
self.log_data, self.nicks, self.nick_same_list)
sys.stdout = sys.__stdout__
capturedOutput.close()
mock_msg_graph.assert_called_once_with(self.log_data, self.nicks,
self.nick_same_list, True)
self.assertEqual(out_degree, out_degree_)
self.assertEqual(in_degree, in_degree_)
self.assertEqual(total_degree, total_degree_)
def test_response_time_high_cutoff_percentile(self):
# this test assumes config.CUTOFF_TIME_STRATEGY = "TWO_SIGMA"
cutoff_percentile = 5.0
expected_resp_time = util.load_from_disk(current_directory +
"/data/truncated_rt_5percent")
expected_cutoff_time = 2
resp_time, cutoff_time = channel.response_time(self.log_data,
self.nicks,
self.nick_same_list,
cutoff_percentile)
assert resp_time == expected_resp_time, \
"Error in computing response time with 5% cutoff percentile."
assert cutoff_time == expected_cutoff_time, \
"Error in computing RT cutoff with 5% cutoff percentile."
cutoff_percentile = 10.0
expected_resp_time = util.load_from_disk(
current_directory + "/data/truncated_rt_10percent")
expected_cutoff_time = 2
resp_time, cutoff_time = channel.response_time(self.log_data,
self.nicks,
self.nick_same_list,
cutoff_percentile)
assert resp_time == expected_resp_time, \
"Error in computing response time with 10% cutoff percentile."
assert cutoff_time == expected_cutoff_time, \
"Error in computing RT cutoff with 10% cutoff percentile."
def setUp(self):
self.current_directory = os.path.dirname(os.path.realpath(__file__))
self.log_data = util.load_from_disk(self.current_directory +
"/data/log_data")
self.nicks1 = util.load_from_disk(self.current_directory +
"/data/nicks1")
self.nick_same_list1 = util.load_from_disk(self.current_directory +
"/data/nick_same_list1")
def test_identify_hubs_and_experts(self, mock_keywords, mock_msg_graph):
top_hub_ = util.load_from_disk(current_directory + "/data/top_hub")
top_keyword_overlap_ = util.load_from_disk(current_directory +
"/data/top_keyword_overlap")
top_auth_ = util.load_from_disk(current_directory + "/data/top_auth")
message_graph = util.load_from_disk(current_directory +
"/data/message_graph")
keyword_dict_list = util.load_from_disk(current_directory +
"/data/keyword_dict_list")
user_keyword_freq_dict = util.load_from_disk(
current_directory + "/data/user_keyword_freq_dict")
user_words_dict_list = util.load_from_disk(
current_directory + "/data/user_words_dict_list")
nicks_for_stop_words = util.load_from_disk(
current_directory + "/data/nicks_for_stop_words")
keywords_for_channels = util.load_from_disk(
current_directory + "/data/keywords_for_channels")
# setup mock
mock_msg_graph.return_value = message_graph
mock_keywords.return_value = keyword_dict_list, user_keyword_freq_dict, user_words_dict_list, nicks_for_stop_words, keywords_for_channels
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
message_num_graph, top_hub, top_keyword_overlap, top_auth = network.identify_hubs_and_experts(
self.log_data, self.nicks, self.nick_same_list)
sys.stdout = sys.__stdout__
capturedOutput.close()
self.assertEqual(top_hub, top_hub_)
self.assertEqual(top_keyword_overlap, top_keyword_overlap_)
self.assertEqual(top_auth, top_auth_)
self.assertTrue(nx.is_isomorphic(message_graph, message_num_graph))
def test_conv_len_conv_refr_time(self):
rt_cutoff_time = 1
cutoff_percentile = 0
expected_conv_len = util.load_from_disk(self.current_directory + "/data/conv_len")
expected_conv_ref_time = util.load_from_disk(self.current_directory + "/data/conv_ref_time")
conv_len, conv_ref_time = \
channel.conv_len_conv_refr_time(self.log_data, self.nicks, self.nick_same_list,
rt_cutoff_time, cutoff_percentile)
self.assertEqual(conv_len, expected_conv_len, \
"Error in computing conversation length correctly.")
self.assertEqual(conv_ref_time, expected_conv_ref_time, \
"Error in computing conversation refresh time correctly.")
def test_degree_analysis_on_graph(self):
directed_graph = util.load_from_disk(self.test_data_dir +
'directed_graph')
directed_deg_analysis_expected = util.load_from_disk(
self.test_data_dir + 'directed_deg_analysis_result')
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
directed_deg_analysis = network.degree_analysis_on_graph(
directed_graph)
sys.stdout = sys.__stdout__
capturedOutput.close()
self.assertEqual(directed_deg_analysis, directed_deg_analysis_expected)
def test_message_number_bins_csv(self):
bin_matrix_ = util.load_from_disk(self.test_data_dir +
"message_number_bins_csv/bin_matrix")
tot_msgs_ = util.load_from_disk(self.test_data_dir +
"message_number_bins_csv/tot_msgs")
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
bin_matrix, tot_msgs = network.message_number_bins_csv(
self.log_data, self.nicks, self.nick_same_list)
sys.stdout = sys.__stdout__
capturedOutput.close()
self.assertEqual(bin_matrix, bin_matrix_)
self.assertEqual(tot_msgs, tot_msgs_)
def test_message_number_graph(self, mock_rec_list_splice, mock_correctLastCharCR, mock_get_nick_sen_rec, mock_get_year_month_day,\
mock_get_nick_representative, mock_check_if_msg_line, mock_create_connected_nick_list, mock_to_graph):
to_graph_ret = util.load_from_disk(
current_directory + "/data/message_number_graph/to_graph")
conn_list = list(connected_components(to_graph_ret))
mock_to_graph.return_value = to_graph_ret
mock_rec_list_splice.side_effect = util.load_from_disk(
current_directory + "/data/message_number_graph/rec_list_splice")
mock_create_connected_nick_list.return_value = util.load_from_disk(
current_directory + "/data/message_number_graph/conn_comp_list")
#mock_correct_last_char_list.side_effect = util.load_from_disk(current_directory + "/data/message_number_graph/correct_last_char_list")
mock_check_if_msg_line.side_effect = util.load_from_disk(
current_directory + "/data/message_number_graph/check_if_msg_line")
mock_correctLastCharCR.side_effect = util.load_from_disk(
current_directory + "/data/message_number_graph/correctLastCharCR")
mock_get_nick_sen_rec.side_effect = util.load_from_disk(
current_directory + "/data/message_number_graph/get_nick_sen_rec")
#mock_extend_conversation_list.side_effect = util.load_from_disk(current_directory + "/data/message_number_graph/extend_conversation_list")
mock_get_nick_representative.side_effect = util.load_from_disk(
current_directory +
"/data/message_number_graph/get_nick_representative")
mock_get_year_month_day.side_effect = util.load_from_disk(
current_directory +
"/data/message_number_graph/get_year_month_day")
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
ret = network.message_number_graph(self.log_data,
self.nicks,
self.nick_same_list,
DAY_BY_DAY_ANALYSIS=False)
sys.stdout = sys.__stdout__
capturedOutput.close()
mock_to_graph.assert_called_once_with(self.nick_same_list)
mock_create_connected_nick_list.assert_called_once_with(conn_list)
self.assertTrue(
nx.is_isomorphic(
ret,
util.load_from_disk(
current_directory +
"/data/message_number_graph/aggregate_message_number_graph"
)))
def test_degree_node_number_csv(self):
out_degree_ = util.load_from_disk(self.test_data_dir +
"degree_node_number_csv/out_degree")
in_degree_ = util.load_from_disk(self.test_data_dir +
"degree_node_number_csv/in_degree")
total_degree_ = util.load_from_disk(
self.test_data_dir + "degree_node_number_csv/total_degree")
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
out_degree, in_degree, total_degree = network.degree_node_number_csv(
self.log_data, self.nicks, self.nick_same_list)
sys.stdout = sys.__stdout__
capturedOutput.close()
self.assertEqual(out_degree, out_degree_)
self.assertEqual(in_degree, in_degree_)
self.assertEqual(total_degree, total_degree_)
def test_linux_input_non_existent_file_slack(self):
expected_captured_output = util.load_from_disk(
self.current_directory +
"/data/stdout_captured_linux_input_slack_non_existent_file")
expected_log_data = util.load_from_disk(
self.current_directory +
"/data/log_data_for_test_linux_input_non_existent_file_slack")
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
log_data = reader.linux_input_slack(
self.current_directory + "/data/slackware_with_missing_files/",
"2013-1-1", "2013-1-6")
output = capturedOutput.getvalue()
capturedOutput.close()
sys.stdout = sys.__stdout__
#See https://docs.python.org/2/library/re.html for more details.
output = re.sub(r'(?P<begin>.+ )/.+/(?P<constant>IRCLogParser/.+\n)',
r'\g<begin>\g<constant>', output)
self.assertEqual(expected_log_data, log_data)
self.assertEqual(expected_captured_output, output)
def test_response_time_low_cutoff_percentile(self):
# this test assumes config.CUTOFF_TIME_STRATEGY = "TWO_SIGMA"
cutoff_percentile = 0.0
expected_resp_time = util.load_from_disk(self.current_directory + "/data/resp_time")
expected_cutoff_time = 2
resp_time, cutoff_time = channel.response_time(self.log_data, self.nicks,
self.nick_same_list, cutoff_percentile)
self.assertEqual(resp_time, expected_resp_time, \
"Error in computing response time with 0% cutoff percentile.")
self.assertEqual(cutoff_time, expected_cutoff_time, \
"Error in computing RT cutoff with 0% cutoff percentile.")
cutoff_percentile = 1.0
expected_resp_time = util.load_from_disk(self.current_directory + "/data/truncated_rt_1percent")
expected_cutoff_time = 2
resp_time, cutoff_time = channel.response_time(self.log_data, self.nicks,
self.nick_same_list, cutoff_percentile)
self.assertEqual(resp_time, expected_resp_time, \
"Error in computing response time with 1% cutoff percentile.")
self.assertEqual(cutoff_time, expected_cutoff_time, \
"Error in computing RT cutoff with 1% cutoff percentile.")
def test_identify_hubs_and_experts(self):
top_hub_ = util.load_from_disk(self.test_data_dir + "hits/top_hub")
top_keyword_overlap_ = util.load_from_disk(self.test_data_dir +
"hits/top_keyword_overlap")
top_auth_ = util.load_from_disk(self.test_data_dir + "hits/top_auth")
message_graph = util.load_from_disk(self.test_data_dir +
"hits/message_graph")
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
message_num_graph, top_hub, top_keyword_overlap, top_auth = network.identify_hubs_and_experts(
self.log_data, self.nicks, self.nick_same_list)
sys.stdout = sys.__stdout__
capturedOutput.close()
self.assertEqual(top_hub, top_hub_)
self.assertEqual(top_keyword_overlap, top_keyword_overlap_)
self.assertEqual(top_auth, top_auth_)
self.assertTrue(nx.is_isomorphic(message_graph, message_num_graph))
def test_truncate_table_with_long_table(self):
resp_time = util.load_from_disk(self.current_directory + "/data/resp_time")
expected_cutoff_time = 1067
cutoff_percentile = 5
truncated_table, cutoff_time = channel.truncate_table(resp_time, cutoff_percentile)
self.assertEqual(cutoff_time, expected_cutoff_time, \
"Error in computing percentile cutoff value")
cutoff_percentile = 1
expected_cutoff_time = 1172
truncated_table, cutoff_time = channel.truncate_table(resp_time, cutoff_percentile)
self.assertEqual(cutoff_time, expected_cutoff_time, \
"Error in computing percentile cutoff value")
def test_message_number_graph(self):
to_graph_ret = util.load_from_disk(self.test_data_dir +
"message_number_graph/to_graph")
conn_list = list(connected_components(to_graph_ret))
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
ret = network.message_number_graph(self.log_data,
self.nicks,
self.nick_same_list,
DAY_BY_DAY_ANALYSIS=False)
sys.stdout = sys.__stdout__
capturedOutput.close()
self.assertTrue(
nx.is_isomorphic(
ret,
util.load_from_disk(
self.test_data_dir +
"message_number_graph/aggregate_message_number_graph")))
def test_message_time_graph_day_analysis(self):
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
ret = network.message_time_graph(self.log_data,
self.nicks,
self.nick_same_list,
DAY_BY_DAY_ANALYSIS=True)
expected_graph_list = util.load_from_disk(
self.test_data_dir + "message_time_graph/msg_time_graph_list")
sys.stdout = sys.__stdout__
capturedOutput.close()
self.assertTrue(nx.is_isomorphic(ret[0], expected_graph_list[0]))
def test_message_time_graph(self):
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
ret = network.message_time_graph(self.log_data,
self.nicks,
self.nick_same_list,
DAY_BY_DAY_ANALYSIS=False)
sys.stdout = sys.__stdout__
capturedOutput.close()
self.assertTrue(
nx.is_isomorphic(
ret,
util.load_from_disk(self.test_data_dir +
"message_time_graph/msg_time_aggr_graph")))
def test_message_time_graph_day_analysis(
self, mock_rec_list_splice, mock_correctLastCharCR,
mock_get_year_month_day, mock_get_nick_sen_rec,
mock_check_if_msg_line, mock_create_connected_nick_list,
mock_to_graph):
to_graph_ret = util.load_from_disk(self.test_data_dir +
"message_time_graph/to_graph")
conn_list = list(connected_components(to_graph_ret))
mock_to_graph.return_value = to_graph_ret
mock_rec_list_splice.side_effect = util.load_from_disk(
self.test_data_dir + "message_time_graph/rec_list_splice")
mock_create_connected_nick_list.return_value = util.load_from_disk(
self.test_data_dir + "message_time_graph/conn_comp_list")
mock_check_if_msg_line.side_effect = util.load_from_disk(
self.test_data_dir + "message_time_graph/check_if_msg_line")
mock_correctLastCharCR.side_effect = util.load_from_disk(
self.test_data_dir + "message_time_graph/correctLastCharCR")
mock_get_nick_sen_rec.side_effect = util.load_from_disk(
self.test_data_dir + "message_time_graph/get_nick_sen_rec")
#mock_correct_last_char_list.side_effect = util.load_from_disk(self.test_data_dir + "message_time_graph/correct_last_char_list")
mock_get_year_month_day.side_effect = util.load_from_disk(
self.test_data_dir + "message_time_graph/get_year_month_day")
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
ret = network.message_time_graph(self.log_data,
self.nicks,
self.nick_same_list,
DAY_BY_DAY_ANALYSIS=True)
expected_graph_list = util.load_from_disk(
self.test_data_dir + "message_time_graph/msg_time_graph_list")
sys.stdout = sys.__stdout__
capturedOutput.close()
mock_to_graph.assert_called_once_with(self.nick_same_list)
mock_create_connected_nick_list.assert_called_once_with(conn_list)
self.assertTrue(nx.is_isomorphic(ret[0], expected_graph_list[0]))
def test_linux_input_slack(self):
expected_captured_output = util.load_from_disk(
self.current_directory + "/data/stdout_captured_linux_input_slack")
capturedOutput = StringIO.StringIO()
sys.stdout = capturedOutput
log_data = reader.linux_input_slack(
self.current_directory + "/data/slackware/", self.starting_date,
self.ending_date)
output = capturedOutput.getvalue()
capturedOutput.close()
sys.stdout = sys.__stdout__
#See https://docs.python.org/2/library/re.html for more details.
# string 'Working on: /any_valid_path/IRCLogParser/test/unit-test/test_lib/test_in_out/data/log/2013/01/04/#kubuntu-devel.txt\n' is replaced by
# 'Working on: IRCLogParser/test/unit-test/test_lib/test_in_out/data/log/2013/01/04/#kubuntu-devel.txt\n'
output = re.sub(r'(?P<begin>.+ )/.+/(?P<constant>IRCLogParser/.+\n)',
r'\g<begin>\g<constant>', output)
self.assertEqual(log_data, self.log_data)
self.assertEqual(expected_captured_output, output)
def setUp(self):
self.log_data = util.load_from_disk(current_directory +
"/data/log_data")
self.nicks = util.load_from_disk(current_directory + "/data/nicks")
self.nick_same_list = util.load_from_disk(current_directory +
"/data/nick_same_list")
def setUp(self):
self.current_directory = os.path.dirname(os.path.realpath(__file__))
self.log_data = util.load_from_disk(self.current_directory +
"/data/log_data")
self.starting_date = "2013-1-1"
self.ending_date = "2013-1-31"
def test_load_from_disk(self):
nicks = util.load_from_disk(current_directory + '/data/nicks')
assert nicks == self.expected_nicks, "Failure to load from disk."
def test_load_from_disk(self):
nicks = util.load_from_disk(self.current_directory + '/data/nicks')
self.assertEqual(nicks, self.expected_nicks,
"Failure to load from disk.")
def setUp(self):
self.log_data = util.load_from_disk(current_directory +
"/data/log_data")
self.starting_date = "2013-1-1"
self.ending_date = "2013-1-31"
