X-Git-Url: https://code.communitydata.science/mediawiki_dump_tools.git/blobdiff_plain/e925ac9da10b20ab4ef949263887a1f9158e805f..776b73519ac8af42b64875a5cc27125ceff7c861:/test/Wikiq_Unit_Test.py diff --git a/test/Wikiq_Unit_Test.py b/test/Wikiq_Unit_Test.py index b32d6a9..0826f1f 100644 --- a/test/Wikiq_Unit_Test.py +++ b/test/Wikiq_Unit_Test.py @@ -1,9 +1,10 @@ import unittest import os -import sys import subprocess from shutil import copyfile -import pdb +import pandas as pd +from pandas.util.testing import assert_frame_equal +from io import StringIO # with / without pwr DONE # with / without url encode DONE @@ -32,21 +33,23 @@ class Test_Wikipedia(unittest.TestCase): self.baseline_output_dir = "baseline_output" def test_WP_url_encode(self): + test_filename = "url-encode_" + 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" proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True) proc.wait() - test_file = "url-encode_" + self.wikiq_out_name - copyfile(self.call_output, os.path.join(self.test_output_dir, test_file)) - baseline_file = os.path.join(".", self.baseline_output_dir, test_file) - test_lines = open(os.path.join(self.test_output_dir,test_file)) - baseline_lines = open(baseline_file) - for test, baseline in zip(test_lines, baseline_lines): - self.assertEqual(test,baseline) + copyfile(self.call_output, test_file) + baseline_file = os.path.join(".", self.baseline_output_dir, test_filename) - test_lines.close() - baseline_lines.close() + # as a test let's make sure that we get equal data frames + test = pd.read_table(test_file) + baseline = pd.read_table(baseline_file) + assert_frame_equal(test,baseline) class Test_Basic(unittest.TestCase): @@ -68,77 +71,122 @@ class Test_Basic(unittest.TestCase): def test_noargs(self): + test_filename = "noargs_" + 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) proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True) proc.wait() - test_file = "noargs_" + self.wikiq_out_name - copyfile(self.call_output, os.path.join(self.test_output_dir, test_file)) - baseline_file = os.path.join(".", self.baseline_output_dir, test_file) + copyfile(self.call_output, test_file) + + baseline_file = os.path.join(".", self.baseline_output_dir, test_filename) - test_lines = open(os.path.join(self.test_output_dir,test_file)) - baseline_lines = open(baseline_file) - for test, baseline in zip(test_lines, baseline_lines): - self.assertEqual(test,baseline) + test = pd.read_table(test_file) + baseline = pd.read_table(baseline_file) + assert_frame_equal(test,baseline) - test_lines.close() - baseline_lines.close() def test_collapse_user(self): + test_filename = "collapse-user_" + 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 + " --collapse-user" proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True) proc.wait() - test_file = "collapse-user_" + self.wikiq_out_name - copyfile(self.call_output, os.path.join(self.test_output_dir, test_file)) + copyfile(self.call_output, test_file) - baseline_file = os.path.join(".", self.baseline_output_dir, test_file) + baseline_file = os.path.join(".", self.baseline_output_dir, test_filename) + test = pd.read_table(test_file) + baseline = pd.read_table(baseline_file) + assert_frame_equal(test,baseline) + + def test_pwr_segment(self): + test_filename = "persistence_segment_" + 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 + " --persistence segment" + 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) + + test = pd.read_table(test_file) + baseline = pd.read_table(baseline_file) + assert_frame_equal(test,baseline) + + def test_pwr_legacy(self): + test_filename = "persistence_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 + " --persistence legacy" + proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True) + proc.wait() + + + copyfile(self.call_output, test_file) - test_lines = open(os.path.join(self.test_output_dir,test_file)) - baseline_lines = open(baseline_file) - for test, baseline in zip(test_lines, baseline_lines): - self.assertEqual(test,baseline) + baseline_file = os.path.join(".", self.baseline_output_dir, test_filename) - test_lines.close() - baseline_lines.close() + test = pd.read_table(test_file) + baseline = pd.read_table(baseline_file) + assert_frame_equal(test,baseline) def test_pwr(self): + test_filename = "persistence_" + 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 + " --persistence" proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True) proc.wait() - test_file = "persistence_" + self.wikiq_out_name - copyfile(self.call_output, os.path.join(self.test_output_dir, test_file)) - baseline_file = os.path.join(".", self.baseline_output_dir, test_file) + copyfile(self.call_output, test_file) - test_lines = open(os.path.join(self.test_output_dir,test_file)) - baseline_lines = open(baseline_file) - for test, baseline in zip(test_lines, baseline_lines): - self.assertEqual(test,baseline) + baseline_file = os.path.join(".", self.baseline_output_dir, test_filename) + + test = pd.read_table(test_file) + baseline = pd.read_table(baseline_file) + assert_frame_equal(test,baseline) - test_lines.close() - baseline_lines.close() def test_url_encode(self): + test_filename = "url-encode_" + 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" proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True) proc.wait() - test_file = "url-encode_" + self.wikiq_out_name - copyfile(self.call_output, os.path.join(self.test_output_dir, test_file)) - baseline_file = os.path.join(".", self.baseline_output_dir, test_file) - test_lines = open(os.path.join(self.test_output_dir,test_file)) - baseline_lines = open(baseline_file) - for test, baseline in zip(test_lines, baseline_lines): - self.assertEqual(test,baseline) + copyfile(self.call_output, test_file) + baseline_file = os.path.join(".", self.baseline_output_dir, test_filename) + test = pd.read_table(test_file) + baseline = pd.read_table(baseline_file) + assert_frame_equal(test,baseline) - test_lines.close() - baseline_lines.close() class Test_Malformed(unittest.TestCase): @@ -182,28 +230,14 @@ class Test_Stdout(unittest.TestCase): call = self.base_call.format(self.input_file) proc = subprocess.run(call,stdout=subprocess.PIPE,shell=True) - outs = proc.stdout.decode('utf-8') - + outs = proc.stdout.decode("utf8") + test_file = "noargs_" + self.wikiq_out_name baseline_file = os.path.join(".", self.baseline_output_dir, test_file) - - test_lines = outs.splitlines(True) - baseline_lines = open(baseline_file) - for test, baseline in zip(test_lines, baseline_lines): - self.assertEqual(test,baseline) - - # test_file = "noargs_" + self.wikiq_out_name - # copyfile(self.call_output, os.path.join(self.test_output_dir, test_file)) - - # baseline_file = os.path.join(".", self.baseline_output_dir, test_file) - - # test_lines = open(os.path.join(self.test_output_dir,test_file)) - # baseline_lines = open(baseline_file) - # for test, baseline in zip(test_lines, baseline_lines): - # self.assertEqual(test,baseline) - # test_lines.close() - # baseline_lines.close() - + print(baseline_file) + test = pd.read_table(StringIO(outs)) + baseline = pd.read_table(baseline_file) + assert_frame_equal(test,baseline) if __name__ == '__main__': unittest.main()