checking in work to deepen migration to new mediawikiutils
This commit is contained in:
@@ -16,6 +16,45 @@ from io import StringIO
|
||||
# wikia and wikipedia data DONE
|
||||
# malformed xmls DONE
|
||||
|
||||
# class Test_Persistence_Bug(unittest.TestCase):
|
||||
|
||||
# def setUp(self):
|
||||
# if not os.path.exists("test_output"):
|
||||
# os.mkdir("test_output")
|
||||
|
||||
# self.wiki = 'enwiki-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_segment_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 --collapse-user"
|
||||
# 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)
|
||||
# baseline = pd.read_table(baseline_file)
|
||||
# assert_frame_equal(test,baseline)
|
||||
|
||||
|
||||
class Test_Wikipedia(unittest.TestCase):
|
||||
def setUp(self):
|
||||
if not os.path.exists("test_output"):
|
||||
@@ -47,6 +86,7 @@ class Test_Wikipedia(unittest.TestCase):
|
||||
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)
|
||||
@@ -63,7 +103,7 @@ class Test_Wikipedia(unittest.TestCase):
|
||||
print(call)
|
||||
proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
|
||||
proc.wait()
|
||||
copyfile(self.call_output, test_file)
|
||||
# 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
|
||||
@@ -74,191 +114,192 @@ class Test_Wikipedia(unittest.TestCase):
|
||||
assert_frame_equal(test,baseline)
|
||||
|
||||
|
||||
class Test_Basic(unittest.TestCase):
|
||||
# class Test_Basic(unittest.TestCase):
|
||||
|
||||
def setUp(self):
|
||||
if not os.path.exists("test_output"):
|
||||
os.mkdir("test_output")
|
||||
# def setUp(self):
|
||||
# if not os.path.exists("test_output"):
|
||||
# os.mkdir("test_output")
|
||||
|
||||
self.wiki = 'sailormoon'
|
||||
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.wiki = 'sailormoon'
|
||||
# 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.7z".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"
|
||||
# self.infile = "{0}.xml.7z".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_noargs(self):
|
||||
# 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)
|
||||
# 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()
|
||||
# call = self.base_call.format(self.input_file, self.test_output_dir)
|
||||
# proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
|
||||
# proc.wait()
|
||||
|
||||
copyfile(self.call_output, test_file)
|
||||
# copyfile(self.call_output, test_file)
|
||||
|
||||
baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
|
||||
# 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 = pd.read_table(test_file)
|
||||
# baseline = pd.read_table(baseline_file)
|
||||
# assert_frame_equal(test,baseline)
|
||||
|
||||
|
||||
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)
|
||||
# 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"
|
||||
# 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()
|
||||
# proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
|
||||
# proc.wait()
|
||||
|
||||
copyfile(self.call_output, 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)
|
||||
# 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)
|
||||
# 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()
|
||||
# call = self.base_call.format(self.input_file, self.test_output_dir)
|
||||
# call = call + " --persistence segment"
|
||||
# print(call)
|
||||
# proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
|
||||
# proc.wait()
|
||||
|
||||
|
||||
copyfile(self.call_output, test_file)
|
||||
# copyfile(self.call_output, test_file)
|
||||
|
||||
baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
|
||||
# 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 = pd.read_table(test_file)
|
||||
# print(test)
|
||||
# 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)
|
||||
# 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()
|
||||
# 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)
|
||||
|
||||
copyfile(self.call_output, test_file)
|
||||
# baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
|
||||
|
||||
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 = 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)
|
||||
# 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()
|
||||
# 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()
|
||||
|
||||
|
||||
copyfile(self.call_output, test_file)
|
||||
# copyfile(self.call_output, test_file)
|
||||
|
||||
baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
|
||||
# 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 = pd.read_table(test_file)
|
||||
# baseline = pd.read_table(baseline_file)
|
||||
# assert_frame_equal(test,baseline)
|
||||
|
||||
|
||||
def test_url_encode(self):
|
||||
test_filename = "url-encode_" + self.wikiq_out_name
|
||||
# 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)
|
||||
# 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()
|
||||
# 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()
|
||||
|
||||
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)
|
||||
# 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)
|
||||
|
||||
|
||||
class Test_Malformed(unittest.TestCase):
|
||||
def setUp(self):
|
||||
if not os.path.exists("test_output"):
|
||||
os.mkdir("test_output")
|
||||
# class Test_Malformed(unittest.TestCase):
|
||||
# def setUp(self):
|
||||
# if not os.path.exists("test_output"):
|
||||
# os.mkdir("test_output")
|
||||
|
||||
self.wiki = 'twinpeaks'
|
||||
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.wiki = 'twinpeaks'
|
||||
# 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.7z".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.infile = "{0}.xml.7z".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)
|
||||
|
||||
|
||||
def test_malformed_noargs(self):
|
||||
# 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')
|
||||
# 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(unittest.TestCase):
|
||||
# class Test_Stdout(unittest.TestCase):
|
||||
|
||||
def setUp(self):
|
||||
self.wiki = 'sailormoon'
|
||||
self.wikiq_out_name = self.wiki + ".tsv"
|
||||
# def setUp(self):
|
||||
# self.wiki = 'sailormoon'
|
||||
# self.wikiq_out_name = self.wiki + ".tsv"
|
||||
|
||||
self.infile = "{0}.xml.7z".format(self.wiki)
|
||||
self.base_call = "../wikiq {0} --stdout"
|
||||
self.input_dir = "dumps"
|
||||
self.input_file = os.path.join(".", self.input_dir,self.infile)
|
||||
self.baseline_output_dir = "baseline_output"
|
||||
# self.infile = "{0}.xml.7z".format(self.wiki)
|
||||
# self.base_call = "../wikiq {0} --stdout"
|
||||
# self.input_dir = "dumps"
|
||||
# self.input_file = os.path.join(".", self.input_dir,self.infile)
|
||||
# self.baseline_output_dir = "baseline_output"
|
||||
|
||||
def test_noargs(self):
|
||||
# def test_noargs(self):
|
||||
|
||||
call = self.base_call.format(self.input_file)
|
||||
proc = subprocess.run(call,stdout=subprocess.PIPE,shell=True)
|
||||
outs = proc.stdout.decode("utf8")
|
||||
# call = self.base_call.format(self.input_file)
|
||||
# proc = subprocess.run(call,stdout=subprocess.PIPE,shell=True)
|
||||
# outs = proc.stdout.decode("utf8")
|
||||
|
||||
test_file = "noargs_" + self.wikiq_out_name
|
||||
baseline_file = os.path.join(".", self.baseline_output_dir, test_file)
|
||||
print(baseline_file)
|
||||
test = pd.read_table(StringIO(outs))
|
||||
baseline = pd.read_table(baseline_file)
|
||||
assert_frame_equal(test,baseline)
|
||||
# test_file = "noargs_" + self.wikiq_out_name
|
||||
# baseline_file = os.path.join(".", self.baseline_output_dir, test_file)
|
||||
# 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()
|
||||
|
||||
Reference in New Issue
Block a user