def test_answer_distribution(self):
self.task.launch([
'AnswerDistributionOneFilePerCourseTask',
'--src', as_list_param(self.test_src),
'--dest', url_path_join(self.test_root, 'dst'),
'--name', 'test',
'--output-root', self.test_out,
'--include', as_list_param('"*"'),
'--manifest', url_path_join(self.test_root, 'manifest.txt'),
'--base-input-format', self.input_format,
'--lib-jar', as_list_param(self.oddjob_jar),
'--n-reduce-tasks', str(self.NUM_REDUCERS),
])
self.validate_output()
def test_answer_distribution_mysql(self):
self.task.launch([
'AnswerDistributionToMySQLTaskWorkflow',
'--src', as_list_param(self.test_src),
'--dest', url_path_join(self.test_root, 'dst'),
'--name', 'test',
'--include', as_list_param('"*"'),
'--manifest', url_path_join(self.test_root, 'manifest.txt'),
'--base-input-format', self.input_format,
'--lib-jar', as_list_param(self.oddjob_jar),
'--n-reduce-tasks', str(self.NUM_REDUCERS),
'--credentials', self.export_db.credentials_file_url,
])
self.validate_output()
def test_event_log_exports_using_manifest(self):
config_override = {
'manifest': {
'threshold': 1
}
}
folders = {
'edx': self.PROD_FOLDER,
'edge': self.EDGE_FOLDER
}
for environment in ['edx', 'edge']:
self.task.launch([
'EventExportTask',
'--source', as_list_param(url_path_join(self.test_src, environment)),
'--output-root', self.test_out,
'--config', self.test_config,
'--environment', environment,
'--interval', '2014-05',
'--gpg-key-dir', self.test_gpg_key_dir,
'--gpg-master-key', '*****@*****.**',
'--required-path-text', folders[environment],
'--n-reduce-tasks', str(self.NUM_REDUCERS),
], config_override)
self.validate_output()
def run_and_check(self, interval_type):
self.task.launch([
'StudentEngagementToMysqlTask',
'--source', as_list_param(self.test_src),
'--credentials', self.export_db.credentials_file_url,
'--n-reduce-tasks', str(self.NUM_REDUCERS),
'--interval', '2015-09-01-2015-09-16',
'--interval-type', interval_type,
])
with self.export_db.cursor() as cursor:
cursor.execute(
'SELECT end_date, course_id, username, '
'forum_posts, forum_responses, forum_comments, '
'forum_upvotes_given, forum_upvotes_received '
'FROM student_engagement_{interval_type} WHERE course_id="{course_id}" '
'ORDER BY end_date, username;'
.format(course_id=self.COURSE_ID, interval_type=interval_type)
)
results = cursor.fetchall()
if interval_type == 'weekly':
end_date_expected = datetime.date(2015, 9, 15)
elif interval_type == 'daily':
end_date_expected = datetime.date(2015, 9, 14)
else:
assert False, "Invalid interval type: {}".format(interval_type)
self.assertItemsEqual(results, [
(end_date_expected, self.COURSE_ID, 'audit', 1, 0, 0, 3, 1),
(end_date_expected, self.COURSE_ID, 'honor', 1, 1, 0, 0, 2),
(end_date_expected, self.COURSE_ID, 'staff', 2, 0, 0, 1, 2),
(end_date_expected, self.COURSE_ID, 'verified', 0, 0, 1, 1, 0),
])
def test_location_by_course(self):
self.upload_tracking_log(self.INPUT_FILE, self.START_DATE)
for fixture_file_name in self.SQL_FIXTURES:
self.execute_sql_fixture_file(fixture_file_name)
self.task.launch([
'InsertToMysqlLastCountryPerCourseTask',
'--source', as_list_param(self.test_src),
'--interval', self.DATE_INTERVAL.to_string(),
'--n-reduce-tasks', str(self.NUM_REDUCERS),
])
self.maxDiff = None
with self.export_db.cursor() as cursor:
cursor.execute('SELECT * FROM course_enrollment_location_current ORDER BY country_code, course_id')
results = cursor.fetchall()
# TODO: what happens if the test starts near the UTC day boundary. The task sees that today is day "X", yet this
# code sees the following day since the day boundary was crossed between then and now.
today = datetime.utcnow().date()
self.assertItemsEqual([
row[1:6] for row in results
], [
(today, self.COURSE_ID, None, 1, 1),
(today, self.COURSE_ID, 'UNKNOWN', 0, 1),
(today, self.COURSE_ID, 'IE', 1, 1),
(today, self.COURSE_ID2, 'TH', 1, 1),
(today, self.COURSE_ID, 'TH', 1, 1),
])
def test_event_log_exports_using_manifest(self):
config_override = {'manifest': {'threshold': 1}}
folders = {'prod': self.PROD_FOLDER, 'edge': self.EDGE_FOLDER}
for environment in ['prod', 'edge']:
self.task.launch([
'EventExportTask',
'--source',
as_list_param(url_path_join(self.test_src, environment)),
'--output-root',
self.test_out,
'--config',
self.test_config,
'--environment',
environment,
'--interval',
'2014-05',
'--gpg-key-dir',
self.test_gpg_key_dir,
'--gpg-master-key',
'*****@*****.**',
'--required-path-text',
folders[environment],
'--n-reduce-tasks',
str(self.NUM_REDUCERS),
], config_override)
self.validate_output()
def launch_task(self, output_root, extra_source=None, run_with_validation_events=True):
"""Run the enrollment validation workflow."""
# Widen the interval to include the latest validation events.
interval = self.WIDER_DATE_INTERVAL if run_with_validation_events else self.DATE_INTERVAL
source_pattern = '[\\".*?.log-.*.gz\\"]'
validation_pattern = '".*?enroll_validated_\d{8}\.log\.gz"'
launch_args = [
'EnrollmentValidationWorkflow',
'--interval', interval,
'--validation-root', self.test_validate,
'--validation-pattern', validation_pattern,
'--credentials', self.import_db.credentials_file_url,
'--n-reduce-tasks', str(self.NUM_REDUCERS),
'--pattern', source_pattern,
'--output-root', output_root,
]
# An extra source means we're using synthetic events, so we
# don't want to generate outside the interval in that case.
if extra_source:
launch_args.extend(['--source', '[\\"{}\\",\\"{}\\"]'.format(self.test_src, extra_source)])
else:
launch_args.extend(['--source', as_list_param(self.test_src)])
launch_args.extend(['--generate-before'])
if run_with_validation_events:
launch_args.extend(['--expected-validation', "{}T00".format(self.END_DATE)])
self.task.launch(launch_args)
def test_user_activity(self):
self.maxDiff = None
self.upload_tracking_log(self.INPUT_FILE, self.END_DATE)
self.task.launch([
'InsertToMysqlCourseActivityTask',
'--source', as_list_param(self.test_src),
'--end-date', self.END_DATE.isoformat(),
'--weeks', str(self.NUM_WEEKS),
'--credentials', self.export_db.credentials_file_url,
'--overwrite-n-days', '43',
'--n-reduce-tasks', str(self.NUM_REDUCERS),
'--overwrite-mysql'
])
with self.export_db.cursor() as cursor:
cursor.execute('SELECT course_id, interval_start, interval_end, label, count FROM course_activity ORDER BY course_id, interval_end, label')
results = cursor.fetchall()
# pylint: disable=line-too-long
self.assertItemsEqual([
row for row in results
], [
(self.COURSE_ID2, datetime.datetime(2014, 5, 19, 0, 0), datetime.datetime(2014, 5, 26, 0, 0), 'ACTIVE', 1),
(self.COURSE_ID2, datetime.datetime(2014, 5, 19, 0, 0), datetime.datetime(2014, 5, 26, 0, 0), 'PLAYED_VIDEO', 1),
(self.COURSE_ID2, datetime.datetime(2014, 6, 16, 0, 0), datetime.datetime(2014, 6, 23, 0, 0), 'ACTIVE', 4),
(self.COURSE_ID2, datetime.datetime(2014, 6, 16, 0, 0), datetime.datetime(2014, 6, 23, 0, 0), 'ATTEMPTED_PROBLEM', 1),
(self.COURSE_ID2, datetime.datetime(2014, 6, 16, 0, 0), datetime.datetime(2014, 6, 23, 0, 0), 'PLAYED_VIDEO', 3),
(self.COURSE_ID, datetime.datetime(2014, 6, 9, 0, 0), datetime.datetime(2014, 6, 16, 0, 0), 'ACTIVE', 1),
(self.COURSE_ID, datetime.datetime(2014, 6, 9, 0, 0), datetime.datetime(2014, 6, 16, 0, 0), 'PLAYED_VIDEO', 1),
(self.COURSE_ID, datetime.datetime(2014, 6, 16, 0, 0), datetime.datetime(2014, 6, 23, 0, 0), 'ACTIVE', 4),
(self.COURSE_ID, datetime.datetime(2014, 6, 16, 0, 0), datetime.datetime(2014, 6, 23, 0, 0), 'ATTEMPTED_PROBLEM', 2),
(self.COURSE_ID, datetime.datetime(2014, 6, 16, 0, 0), datetime.datetime(2014, 6, 23, 0, 0), 'PLAYED_VIDEO', 3),
])
def run_task(self, interval_type):
"""Run the CSV-generating task."""
self.task.launch([
'StudentEngagementCsvFileTask',
'--source', as_list_param(self.test_src),
'--output-root', url_path_join(self.test_out, interval_type),
'--n-reduce-tasks', str(self.NUM_REDUCERS),
'--interval', self.interval,
'--interval-type', interval_type,
])
def test_database_import(self):
self.task.launch([
'ImportMysqlToVerticaTask',
'--date', self.DATE,
'--marker-schema', 'acceptance_marker',
'--exclude-field', as_list_param('.*\\.field_to_exclude$'),
'--overwrite',
])
self.validate_output()
def run_obfuscated_package_task(self):
"""Run ObfuscatedPackageTask."""
self.task.launch([
'ObfuscatedPackageTask', '--course', self.filename_safe_course_id,
'--obfuscated-output-root',
url_path_join(self.test_root,
'obfuscated-output'), '--gpg-key-dir',
self.test_gpg_key_dir, '--gpg-master-key', '*****@*****.**',
'--output-root', self.test_out, '--recipient',
as_list_param('*****@*****.**'), '--format-version',
self.FORMAT_VERSION
])
def run_obfuscated_package_task(self):
"""Run ObfuscatedPackageTask."""
self.task.launch([
'ObfuscatedPackageTask',
'--course', self.filename_safe_course_id,
'--obfuscated-output-root', url_path_join(self.test_root, 'obfuscated-output'),
'--gpg-key-dir', self.test_gpg_key_dir,
'--gpg-master-key', '*****@*****.**',
'--output-root', self.test_out,
'--recipient', as_list_param('*****@*****.**'),
'--format-version', self.FORMAT_VERSION
])
def test_database_import(self):
self.task.launch([
'ImportMysqlToVerticaTask',
'--date',
self.DATE,
'--marker-schema',
'acceptance_marker',
'--exclude-field',
as_list_param('.*\\.field_to_exclude$'),
'--overwrite',
])
self.validate_output()
def test_base(self):
self.upload_tracking_log(self.INPUT_FILE, datetime.date(2015, 8, 1))
self.execute_sql_fixture_file('load_auth_userprofile.sql')
self.task.launch([
'TagsDistributionWorkflow',
'--source', as_list_param(self.test_src),
'--interval', '2010-01-01-2020-01-01',
'--n-reduce-tasks', str(self.NUM_REDUCERS),
'--output-root', url_path_join(self.test_out, 'tags_dist_acceptance', ''),
'--database', self.export_db.database_name
])
self.validate_base()
def test_base(self):
self.upload_tracking_log(self.INPUT_FILE, datetime.date(2015, 8, 1))
self.execute_sql_fixture_file('load_auth_userprofile.sql')
self.task.launch([
'TagsDistributionWorkflow', '--source',
as_list_param(self.test_src), '--interval',
'2010-01-01-2020-01-01', '--n-reduce-tasks',
str(self.NUM_REDUCERS), '--output-root',
url_path_join(self.test_out, 'tags_dist_acceptance',
''), '--database', self.export_db.database_name
])
self.validate_base()
def run_task(self, interval_type):
"""Run the CSV-generating task."""
self.task.launch([
'StudentEngagementCsvFileTask',
'--source',
as_list_param(self.test_src),
'--output-root',
url_path_join(self.test_out, interval_type),
'--n-reduce-tasks',
str(self.NUM_REDUCERS),
'--interval',
self.interval,
'--interval-type',
interval_type,
])
def test_user_activity(self):
self.maxDiff = None
self.upload_tracking_log(self.INPUT_FILE, self.END_DATE)
self.task.launch([
'InsertToMysqlCourseActivityTask', '--source',
as_list_param(self.test_src), '--end-date',
self.END_DATE.isoformat(), '--weeks',
str(self.NUM_WEEKS), '--credentials',
self.export_db.credentials_file_url, '--overwrite-n-days', '43',
'--n-reduce-tasks',
str(self.NUM_REDUCERS), '--overwrite-mysql'
])
with self.export_db.cursor() as cursor:
cursor.execute(
'SELECT course_id, interval_start, interval_end, label, count FROM course_activity ORDER BY course_id, interval_end, label'
)
results = cursor.fetchall()
# pylint: disable=line-too-long
self.assertItemsEqual([row for row in results], [
(self.COURSE_ID2, datetime.datetime(2014, 5, 19, 0, 0),
datetime.datetime(2014, 5, 26, 0, 0), 'ACTIVE', 1),
(self.COURSE_ID2, datetime.datetime(2014, 5, 19, 0, 0),
datetime.datetime(2014, 5, 26, 0, 0), 'PLAYED_VIDEO', 1),
(self.COURSE_ID2, datetime.datetime(2014, 6, 16, 0, 0),
datetime.datetime(2014, 6, 23, 0, 0), 'ACTIVE', 4),
(self.COURSE_ID2, datetime.datetime(2014, 6, 16, 0, 0),
datetime.datetime(2014, 6, 23, 0, 0), 'ATTEMPTED_PROBLEM', 1),
(self.COURSE_ID2, datetime.datetime(2014, 6, 16, 0, 0),
datetime.datetime(2014, 6, 23, 0, 0), 'PLAYED_VIDEO', 3),
(self.COURSE_ID, datetime.datetime(2014, 6, 9, 0, 0),
datetime.datetime(2014, 6, 16, 0, 0), 'ACTIVE', 1),
(self.COURSE_ID, datetime.datetime(2014, 6, 9, 0, 0),
datetime.datetime(2014, 6, 16, 0, 0), 'PLAYED_VIDEO', 1),
(self.COURSE_ID, datetime.datetime(2014, 6, 16, 0, 0),
datetime.datetime(2014, 6, 23, 0, 0), 'ACTIVE', 4),
(self.COURSE_ID, datetime.datetime(2014, 6, 16, 0, 0),
datetime.datetime(2014, 6, 23, 0, 0), 'ATTEMPTED_PROBLEM', 2),
(self.COURSE_ID, datetime.datetime(2014, 6, 16, 0, 0),
datetime.datetime(2014, 6, 23, 0, 0), 'PLAYED_VIDEO', 3),
])
def launch_task(self,
output_root,
extra_source=None,
run_with_validation_events=True):
"""Run the enrollment validation workflow."""
# Widen the interval to include the latest validation events.
interval = self.WIDER_DATE_INTERVAL if run_with_validation_events else self.DATE_INTERVAL
source_pattern = '[\\".*?.log-.*.gz\\"]'
validation_pattern = '".*?enroll_validated_\d{8}\.log\.gz"'
launch_args = [
'EnrollmentValidationWorkflow',
'--interval',
interval,
'--validation-root',
self.test_validate,
'--validation-pattern',
validation_pattern,
'--credentials',
self.import_db.credentials_file_url,
'--n-reduce-tasks',
str(self.NUM_REDUCERS),
'--pattern',
source_pattern,
'--output-root',
output_root,
]
# An extra source means we're using synthetic events, so we
# don't want to generate outside the interval in that case.
if extra_source:
launch_args.extend([
'--source',
'[\\"{}\\",\\"{}\\"]'.format(self.test_src, extra_source)
])
else:
launch_args.extend(['--source', as_list_param(self.test_src)])
launch_args.extend(['--generate-before'])
if run_with_validation_events:
launch_args.extend(
['--expected-validation', "{}T00".format(self.END_DATE)])
self.task.launch(launch_args)
def run_and_check(self, interval_type):
self.task.launch([
'StudentEngagementToMysqlTask',
'--source',
as_list_param(self.test_src),
'--credentials',
self.export_db.credentials_file_url,
'--n-reduce-tasks',
str(self.NUM_REDUCERS),
'--interval',
'2015-09-01-2015-09-16',
'--interval-type',
interval_type,
])
with self.export_db.cursor() as cursor:
cursor.execute(
'SELECT end_date, course_id, username, '
'forum_posts, forum_responses, forum_comments, '
'forum_upvotes_given, forum_upvotes_received '
'FROM student_engagement_{interval_type} WHERE course_id="{course_id}" '
'ORDER BY end_date, username;'.format(
course_id=self.COURSE_ID, interval_type=interval_type))
results = cursor.fetchall()
if interval_type == 'weekly':
end_date_expected = datetime.date(2015, 9, 15)
elif interval_type == 'daily':
end_date_expected = datetime.date(2015, 9, 14)
else:
assert False, "Invalid interval type: {}".format(interval_type)
self.assertItemsEqual(results, [
(end_date_expected, self.COURSE_ID, 'audit', 1, 0, 0, 3, 1),
(end_date_expected, self.COURSE_ID, 'honor', 1, 1, 0, 0, 2),
(end_date_expected, self.COURSE_ID, 'staff', 2, 0, 0, 1, 2),
(end_date_expected, self.COURSE_ID, 'verified', 0, 0, 1, 1, 0),
])