def test_workflow(self):
# set up directories:
src_path = os.path.join(self.temp_rootdir, "src")
os.mkdir(src_path)
counts_path = os.path.join(self.temp_rootdir, "counts")
os.mkdir(counts_path)
report_path = os.path.join(self.temp_rootdir, "report.csv")
data_filepath = os.path.join(self.temp_rootdir, "geoloc.dat")
with open(data_filepath, 'w') as data_file:
data_file.write("Dummy geolocation data.")
# create input:
log_filepath = os.path.join(src_path, "tracking.log")
with open(log_filepath, 'w') as log_file:
log_file.write(self._create_event_log_line())
log_file.write('\n')
log_file.write(
self._create_event_log_line(username="******",
ip=FakeGeoLocation.ip_address_2))
log_file.write('\n')
end_date = '2014-04-01'
task = UsersPerCountryReportWorkflow(
mapreduce_engine='local',
name='test',
src=[src_path],
end_date=datetime.datetime.strptime(end_date, '%Y-%m-%d').date(),
geolocation_data=data_filepath,
counts=counts_path,
report=report_path,
)
worker = luigi.worker.Worker()
worker.add(task)
with patch(
'edx.analytics.tasks.user_location.pygeoip') as mock_pygeoip:
mock_pygeoip.GeoIP = Mock(return_value=FakeGeoLocation())
worker.run()
worker.stop()
output_lines = []
with open(report_path) as report_file:
output_lines = report_file.readlines()
self.assertEquals(len(output_lines), 3)
self.assertEquals(output_lines[0].strip('\n'),
UsersPerCountryReport.create_header(end_date))
expected = UsersPerCountryReport.create_csv_entry(
0.5, 1, FakeGeoLocation.country_name_1,
FakeGeoLocation.country_code_1)
self.assertEquals(output_lines[1].strip('\n'), expected)
expected = UsersPerCountryReport.create_csv_entry(
0.5, 1, FakeGeoLocation.country_name_2,
FakeGeoLocation.country_code_2)
self.assertEquals(output_lines[2].strip('\n'), expected)
def test_workflow(self):
# set up directories:
src_path = os.path.join(self.temp_rootdir, "src")
os.mkdir(src_path)
counts_path = os.path.join(self.temp_rootdir, "counts")
os.mkdir(counts_path)
report_path = os.path.join(self.temp_rootdir, "report.csv")
data_filepath = os.path.join(self.temp_rootdir, "geoloc.dat")
with open(data_filepath, 'w') as data_file:
data_file.write("Dummy geolocation data.")
# create input:
log_filepath = os.path.join(src_path, "tracking.log")
with open(log_filepath, 'w') as log_file:
log_file.write(self._create_event_log_line())
log_file.write('\n')
log_file.write(self._create_event_log_line(username="******", ip=FakeGeoLocation.ip_address_2))
log_file.write('\n')
end_date = '2014-04-01'
task = UsersPerCountryReportWorkflow(
mapreduce_engine='local',
name='test',
src=[src_path],
end_date=datetime.datetime.strptime(end_date, '%Y-%m-%d').date(),
geolocation_data=data_filepath,
counts=counts_path,
report=report_path,
)
worker = luigi.worker.Worker()
worker.add(task)
with patch('edx.analytics.tasks.user_location.pygeoip') as mock_pygeoip:
mock_pygeoip.GeoIP = Mock(return_value=FakeGeoLocation())
worker.run()
worker.stop()
output_lines = []
with open(report_path) as report_file:
output_lines = report_file.readlines()
self.assertEquals(len(output_lines), 3)
self.assertEquals(output_lines[0].strip('\n'), UsersPerCountryReport.create_header(end_date))
expected = UsersPerCountryReport.create_csv_entry(
0.5, 1, FakeGeoLocation.country_name_1, FakeGeoLocation.country_code_1
)
self.assertEquals(output_lines[1].strip('\n'), expected)
expected = UsersPerCountryReport.create_csv_entry(
0.5, 1, FakeGeoLocation.country_name_2, FakeGeoLocation.country_code_2
)
self.assertEquals(output_lines[2].strip('\n'), expected)
def test_report(self):
date = '2014-04-01'
# Output counts in reverse order, to confirm that sorting works.
counts = """
Country_1 Code_1 34 {date}
Country_2 Code_2 43 {date}
""".format(date=date)
output = self.run_task(counts)
output_lines = output.split('\n')
self.assertEquals(output_lines[0], UsersPerCountryReport.create_header(date))
self.assertEquals(
output_lines[1], UsersPerCountryReport.create_csv_entry(float(43) / 77, 43, "Country_2", "Code_2")
)
self.assertEquals(
output_lines[2], UsersPerCountryReport.create_csv_entry(float(34) / 77, 34, "Country_1", "Code_1")
)
# Also confirm the formatting:
for line in output_lines[1:2]:
self.assertTrue(line.startswith('0.'))
def run_task(self, counts):
"""
Run task with fake targets.
Returns:
the task output as a string.
"""
task = UsersPerCountryReport(counts='fake_counts', report='fake_report')
def reformat(string):
"""Reformat string to make it like a TSV."""
return textwrap.dedent(string).strip().replace(' ', '\t')
task.input = MagicMock(return_value=FakeTarget(reformat(counts)))
output_target = FakeTarget()
task.output = MagicMock(return_value=output_target)
task.run()
return output_target.buffer.read()
def test_report(self):
date = '2014-04-01'
# Output counts in reverse order, to confirm that sorting works.
counts = """
Country_1 Code_1 34 {date}
Country_2 Code_2 43 {date}
""".format(date=date)
output = self.run_task(counts)
output_lines = output.split('\n')
self.assertEquals(output_lines[0],
UsersPerCountryReport.create_header(date))
self.assertEquals(
output_lines[1],
UsersPerCountryReport.create_csv_entry(
float(43) / 77, 43, "Country_2", "Code_2"))
self.assertEquals(
output_lines[2],
UsersPerCountryReport.create_csv_entry(
float(34) / 77, 34, "Country_1", "Code_1"))
# Also confirm the formatting:
for line in output_lines[1:2]:
self.assertTrue(line.startswith('0.'))
def run_task(self, counts):
"""
Run task with fake targets.
Returns:
the task output as a string.
"""
task = UsersPerCountryReport(counts='fake_counts',
report='fake_report')
def reformat(string):
"""Reformat string to make it like a TSV."""
return textwrap.dedent(string).strip().replace(' ', '\t')
task.input = MagicMock(return_value=FakeTarget(value=reformat(counts)))
output_target = FakeTarget()
task.output = MagicMock(return_value=output_target)
task.run()
return output_target.buffer.read()
