145 lines
5.0 KiB
Python
145 lines
5.0 KiB
Python
import unittest
|
|
import os
|
|
import subprocess
|
|
from shutil import copyfile
|
|
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
|
|
# with / without collapse user DONE
|
|
# with output to sdtout DONE
|
|
# note that the persistence radius is 7 by default
|
|
# reading various file formats including
|
|
# 7z, gz, bz2, xml DONE
|
|
# wikia and wikipedia data DONE
|
|
# malformed xmls DONE
|
|
|
|
class Test_Wikiq(unittest.TestCase):
|
|
|
|
def mkoutputdir(self):
|
|
if not os.path.exists("test_output"):
|
|
os.mkdir("test_output")
|
|
|
|
def setuptoutputfiles(self, suffix="xml.7z"):
|
|
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}.{1}".format(self.wiki,suffix)
|
|
self.input_dir = "dumps"
|
|
self.input_file = os.path.join(".", self.input_dir,self.infile)
|
|
self.baseline_output_dir = "baseline_output"
|
|
|
|
def run_and_check_output(self, call, test_filename):
|
|
test_file = os.path.join(self.test_output_dir, test_filename)
|
|
if os.path.exists(test_file):
|
|
os.remove(test_file)
|
|
|
|
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)
|
|
baseline = pd.read_table(baseline_file)
|
|
assert_frame_equal(test,baseline)
|
|
|
|
class Test_Wikipedia(Test_Wikiq):
|
|
def setUp(self):
|
|
print(os.path.abspath("."))
|
|
self.mkoutputdir()
|
|
self.wiki = 'ikwiki-20180301-pages-meta-history'
|
|
self.setuptoutputfiles(suffix="xml.bz2")
|
|
self.base_call = "../bin/wikiq {0} -o {1}"
|
|
|
|
def test_WP_url_encode(self):
|
|
test_filename = "url-encode_" + self.wikiq_out_name
|
|
call = self.base_call.format(self.input_file, self.test_output_dir)
|
|
call = call + " --url-encode"
|
|
self.run_and_check_output(call, test_filename)
|
|
|
|
|
|
class Test_Basic(Test_Wikiq):
|
|
|
|
def setUp(self):
|
|
self.mkoutputdir()
|
|
self.wiki="sailormoon"
|
|
self.setuptoutputfiles()
|
|
self.base_call = "../bin/wikiq {0} -o {1}"
|
|
|
|
def test_noargs(self):
|
|
test_filename = "noargs_" + self.wikiq_out_name
|
|
|
|
call = self.base_call.format(self.input_file, self.test_output_dir)
|
|
print(call)
|
|
self.run_and_check_output(call, test_filename)
|
|
|
|
def test_collapse_user(self):
|
|
test_filename = "collapse-user_" + self.wikiq_out_name
|
|
|
|
call = self.base_call.format(self.input_file, self.test_output_dir)
|
|
call = call + " --collapse-user"
|
|
|
|
self.run_and_check_output(call, test_filename)
|
|
|
|
def test_pwr_legacy(self):
|
|
test_filename = "persistence_legacy_" + self.wikiq_out_name
|
|
|
|
call = self.base_call.format(self.input_file, self.test_output_dir)
|
|
call = call + " --persistence-legacy"
|
|
self.run_and_check_output(call, test_filename)
|
|
|
|
def test_pwr(self):
|
|
test_filename = "persistence_" + self.wikiq_out_name
|
|
|
|
call = self.base_call.format(self.input_file, self.test_output_dir)
|
|
call = call + " --persistence"
|
|
self.run_and_check_output(call, test_filename)
|
|
|
|
def test_url_encode(self):
|
|
test_filename = "url-encode_" + self.wikiq_out_name
|
|
|
|
call = self.base_call.format(self.input_file, self.test_output_dir)
|
|
call = call + " --url-encode"
|
|
self.run_and_check_output(call, test_filename)
|
|
|
|
class Test_Malformed(Test_Wikiq):
|
|
|
|
def setUp(self):
|
|
self.mkoutputdir()
|
|
self.wiki="twinpeaks"
|
|
self.setuptoutputfiles()
|
|
self.base_call = "../bin/wikiq {0} -o {1}"
|
|
|
|
def test_malformed_noargs(self):
|
|
call = self.base_call.format(self.input_file, self.test_output_dir)
|
|
proc = subprocess.Popen(call,stdout=subprocess.PIPE,stderr=subprocess.PIPE, shell=True)
|
|
proc.wait()
|
|
outs, errs = proc.communicate()
|
|
errlines = str(errs).split("\\n")
|
|
self.assertEqual(errlines[-2],'xml.etree.ElementTree.ParseError: no element found: line 1369, column 0')
|
|
|
|
class Test_Stdout(Test_Wikiq):
|
|
|
|
def setUp(self):
|
|
self.mkoutputdir()
|
|
self.wiki = 'sailormoon'
|
|
self.setuptoutputfiles()
|
|
|
|
|
|
def test_noargs(self):
|
|
self.base_call = ["../bin/wikiq", self.input_file, "--stdout"]
|
|
proc = subprocess.Popen(self.base_call, stdout=subprocess.PIPE, stderr=subprocess.PIPE, encoding='utf-8')
|
|
outs = proc.stdout
|
|
test_file = "noargs_" + self.wikiq_out_name
|
|
baseline_file = os.path.join(".", self.baseline_output_dir, test_file)
|
|
test = pd.read_table(outs)
|
|
baseline = pd.read_table(baseline_file)
|
|
assert_frame_equal(test,baseline)
|
|
|
|
if __name__ == '__main__':
|
|
unittest.main()
|