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
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)
+
+ # 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)
+
+ def test_WP_namespaces(self):
+ print(os.path.abspath('.'))
+ test_filename = "namespaces_" + 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 + " -n 0 -n 1"
+ print(call)
+ proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
+ proc.wait()
+ copyfile(self.call_output, test_file)
+ baseline_file = os.path.join(os.path.abspath("."), 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)
+ num_wrong_ns = sum(~ test.namespace.isin({0,1}))
+ self.assertEqual(num_wrong_ns, 0)
+ baseline = pd.read_table(baseline_file)
+ assert_frame_equal(test,baseline)
+
+ def test_WP_revert_radius(self):
+ print(os.path.abspath('.'))
+ test_filename = "revert_radius_" + 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 + " -n 0 -n 1 -rr 1"
+ print(call)
+ proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
+ proc.wait()
+ copyfile(self.call_output, test_file)
+ baseline_file = os.path.join(os.path.abspath("."), 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)
+ num_wrong_ns = sum(~ test.namespace.isin({0,1}))
+ self.assertEqual(num_wrong_ns, 0)
+ baseline = pd.read_table(baseline_file)
+ assert_frame_equal(test,baseline)
- test_lines.close()
- baseline_lines.close()
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)
- 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_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_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()
- 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)
- test_lines.close()
- baseline_lines.close()
+ 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)
+
+ 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(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):
-
def setUp(self):
if not os.path.exists("test_output"):
os.mkdir("test_output")
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()
projects / mediawiki_dump_tools.git / blobdiff