+class Test_Persistence(unittest.TestCase):
+
+ def setUp(self):
+
+ if not os.path.exists("test_output"):
+ os.mkdir("test_output")
+
+ self.wiki = 'pwr-test'
+ self.wikiq_out_name = self.wiki + ".tsv"
+ self.test_output_dir = os.path.join(".", "test_output")
+ self.call_output = os.path.join(self.test_output_dir, self.wikiq_out_name)
+
+ self.infile = "{0}.xml".format(self.wiki)
+ self.base_call = "../wikiq {0} -o {1}"
+ self.input_dir = "dumps"
+ self.input_file = os.path.join(".", self.input_dir,self.infile)
+ self.baseline_output_dir = "baseline_output"
+
+ def test_sequence_persistence(self):
+ test_filename = "sequence-" + self.wikiq_out_name
+ test_file = os.path.join(self.test_output_dir, test_filename)
+ if os.path.exists(test_file):
+ os.remove(test_file)
+
+ call = self.base_call.format(self.input_file, self.test_output_dir)
+ call = call + " --url-encode --persistence sequence"
+ print(os.path.abspath('.'))
+ print(call)
+ proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
+ proc.wait()
+
+ copyfile(self.call_output, test_file)
+ baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
+
+ # as a test let's make sure that we get equal data frames
+ test = pd.read_table(test_file)
+ self.assertEqual(test['tokens_added'][0],7)
+ self.assertEqual(test['tokens_added'][1],10)
+ self.assertEqual(test['tokens_added'][2],0)
+ self.assertEqual(test['tokens_added'][3],8)
+ self.assertEqual(test['tokens_added'][4],0)
+ self.assertEqual(test['tokens_removed'][0],0)
+ self.assertEqual(test['tokens_removed'][1],0)
+ self.assertEqual(test['tokens_removed'][2],0)
+ self.assertEqual(test['tokens_removed'][3],4)
+ self.assertEqual(test['tokens_removed'][4],0)
+ self.assertEqual(test['token_revs'][0],8*3)
+ self.assertEqual(test['token_revs'][1],0)
+ self.assertEqual(test['token_revs'][2],0)
+ self.assertEqual(test['token_revs'][3],0)
+ self.assertEqual(test['token_revs'][4],0)
+
+ baseline = pd.read_table(baseline_file)
+ assert_frame_equal(test,baseline)
+
+ def test_legacy_persistence(self):
+ test_filename = "legacy-" + self.wikiq_out_name
+ test_file = os.path.join(self.test_output_dir, test_filename)
+ if os.path.exists(test_file):
+ os.remove(test_file)
+
+ call = self.base_call.format(self.input_file, self.test_output_dir)
+ call = call + " --url-encode --persistence legacy"
+ print(os.path.abspath('.'))
+ print(call)
+ proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
+ proc.wait()
+
+ copyfile(self.call_output, test_file)
+ baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
+
+ # as a test let's make sure that we get equal data frames
+ test = pd.read_table(test_file)
+ self.assertEqual(test['tokens_added'][0],7)
+ self.assertEqual(test['tokens_added'][1],10)
+ self.assertEqual(test['tokens_added'][2],0)
+ self.assertEqual(test['tokens_added'][3],11)
+ self.assertEqual(test['tokens_added'][4],0)
+ self.assertEqual(test['tokens_removed'][0],0)
+ self.assertEqual(test['tokens_removed'][1],0)
+ self.assertEqual(test['tokens_removed'][2],0)
+ self.assertEqual(test['tokens_removed'][3],7)
+ self.assertEqual(test['tokens_removed'][4],0)
+ self.assertEqual(test['token_revs'][0],7*3)
+ self.assertEqual(test['token_revs'][1],0)
+ self.assertEqual(test['token_revs'][2],0)
+ self.assertEqual(test['token_revs'][3],0)
+ self.assertEqual(test['token_revs'][4],0)
+
+ baseline = pd.read_table(baseline_file)
+ assert_frame_equal(test,baseline)