class TestMimirDatasetAnnotations(unittest.TestCase):
def setUp(self):
"""Create empty server directory."""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.mkdir(SERVER_DIR)
self.fileserver = FileSystemFilestore(FILESERVER_DIR)
self.db = MimirDatastore(DATASTORE_DIRECTORY)
def tearDown(self):
"""Delete server directory."""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
def test_dataset_annotations(self):
"""Run test for Mimir datastore."""
dh = self.db.load_dataset(
f_handle=self.fileserver.upload_file(DATA_FILE))
ds = self.db.get_dataset(dh.identifier)
rows = ds.fetch_rows()
print(ds.row_ids)
for row in rows:
print(str(row.identifier) + '\t' + str(row.values))
for row_id in ds.row_ids:
for anno in ds.get_annotations(column_id=1, row_id=row_id):
print(str(row_id) + '\t' + anno.key + '=' + str(anno.value))
def setUp(self):
"""Create empty server directory."""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.mkdir(SERVER_DIR)
self.fileserver = FileSystemFilestore(FILESERVER_DIR)
self.db = MimirDatastore(DATASTORE_DIRECTORY)
def setUp(self):
"""Create an instance of the Mimir processor for an empty server
directory.
"""
# Drop directory if it exists
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.makedirs(SERVER_DIR)
self.datastore = MimirDatastore(DATASTORE_DIR)
self.filestore = FileSystemFilestore(FILESTORE_DIR)
def setUp(self):
"""Create an instance of the Mimir processor for an empty server
directory.
"""
# Drop directory if it exists
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.makedirs(SERVER_DIR)
self.processor = MimirProcessor()
self.datastore = MimirDatastore(DATASTORE_DIR)
self.filestore = FileSystemFilestore(FILESTORE_DIR)
self.available_lenses = set(mimir.getAvailableLensTypes())
def set_up(self, engine):
"""Create an empty file server repository."""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.mkdir(SERVER_DIR)
# Setup file server
self.fs = FileSystemFilestore(FILESERVER_DIR)
# Setup the respective datastore and Vizual engine
if engine == ENGINEENV_DEFAULT:
self.datastore = FileSystemDatastore(DATASTORE_DIR)
elif engine == ENGINEENV_MIMIR:
self.datastore = MimirDatastore(DATASTORE_DIR)
class TestSQLProcessor(unittest.TestCase):
def setUp(self):
"""Create an instance of the Mimir processor for an empty server
directory.
"""
# Drop directory if it exists
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.makedirs(SERVER_DIR)
self.datastore = MimirDatastore(DATASTORE_DIR)
self.filestore = FileSystemFilestore(FILESTORE_DIR)
def tearDown(self):
"""Clean-up by dropping the server directory.
"""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
def test_run_sql_query(self):
"""Test running a SQL query without materializing the result."""
f_handle = self.filestore.upload_file(CSV_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
cmd = sql_cell(source='SELECT grade_or_service_category FROM ' +
DATASET_NAME + ' WHERE program = \'GENERAL EDUCATION\'',
validate=True)
result = SQLTaskProcessor().compute(
command_id=cmd.command_id,
arguments=cmd.arguments,
context=TaskContext(datasets={DATASET_NAME: ds.identifier},
datastore=self.datastore,
filestore=self.filestore))
self.assertTrue(result.is_success)
self.assertIsNone(result.provenance.read)
self.assertIsNone(result.provenance.write)
self.assertTrue(len(result.outputs.stdout) > 0)
self.assertEqual(len(result.outputs.stderr), 0)
# Materialize result
cmd = sql_cell(source='SELECT grade_or_service_category FROM ' +
DATASET_NAME + ' WHERE program = \'GENERAL EDUCATION\'',
output_dataset='ge',
validate=True)
result = SQLTaskProcessor().compute(
command_id=cmd.command_id,
arguments=cmd.arguments,
context=TaskContext(datasets={DATASET_NAME: ds.identifier},
datastore=self.datastore,
filestore=self.filestore))
self.assertTrue(result.is_success)
self.assertIsNone(result.provenance.read)
self.assertIsNotNone(result.provenance.write)
self.assertTrue('ge' in result.provenance.write)
self.assertTrue(len(result.outputs.stdout) > 0)
self.assertEqual(len(result.outputs.stderr), 0)
def get_datastore(self, identifier):
"""Get the datastore instance for the project with the given identifier.
Paramaters
----------
identifier: string
Unique identifier for datastore
Returns
-------
vizier.datastore.base.Datastore
"""
datastore_dir = os.path.join(self.base_path, identifier)
return MimirDatastore(datastore_dir)
def test_mimir_config(self):
"""Run workflows for Mimir configurations."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
import vizier.mimir as mimir
self.run_workflow(MimirDatastore(DATASTORE_DIR))
def datastore_init(self):
"""Test initalizing a datastore with existing datasets."""
self.setup_fileserver()
ds = self.db.load_dataset(self.fileserver.upload_file(CSV_FILE))
self.db = MimirDatastore(DATASTORE_DIR)
def set_up(self):
"""Create empty data store directory."""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.mkdir(SERVER_DIR)
self.db = MimirDatastore(DATASTORE_DIR)
class TestMimirDatastore(unittest.TestCase):
def setup_fileserver(self):
"""Create a fresh file server."""
if os.path.isdir(FILESERVER_DIR):
shutil.rmtree(FILESERVER_DIR)
os.mkdir(FILESERVER_DIR)
self.fileserver = FileSystemFilestore(FILESERVER_DIR)
def set_up(self):
"""Create empty data store directory."""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.mkdir(SERVER_DIR)
self.db = MimirDatastore(DATASTORE_DIR)
def tear_down(self):
"""Delete data store directory.
"""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
def test_mimir_datastore(self):
"""Run test for Mimir datastore."""
self.set_up()
self.dataset_load()
self.tear_down()
self.set_up()
self.datastore_init()
self.tear_down()
self.set_up()
self.dataset_read()
self.tear_down()
self.set_up()
self.dataset_column_index()
self.tear_down()
self.tear_down()
def datastore_init(self):
"""Test initalizing a datastore with existing datasets."""
self.setup_fileserver()
ds = self.db.load_dataset(self.fileserver.upload_file(CSV_FILE))
self.db = MimirDatastore(DATASTORE_DIR)
def dataset_column_index(self):
"""Test the column by id index of the dataset handle."""
self.setup_fileserver()
ds = self.db.load_dataset(self.fileserver.upload_file(CSV_FILE))
# Ensure that the project data has three columns and two rows
self.assertEqual(ds.column_by_id(0).name.upper(), 'NAME')
self.assertEqual(ds.column_by_id(1).name.upper(), 'AGE')
self.assertEqual(ds.column_by_id(2).name.upper(), 'SALARY')
with self.assertRaises(ValueError):
ds.column_by_id(5)
ds.columns.append(DatasetColumn(identifier=5, name='NEWNAME'))
self.assertEqual(ds.column_by_id(5).name.upper(), 'NEWNAME')
with self.assertRaises(ValueError):
ds.column_by_id(4)
def dataset_load(self):
"""Test create and delete dataset."""
self.setup_fileserver()
ds = self.db.load_dataset(self.fileserver.upload_file(CSV_FILE))
# Ensure that the project data has three columns and two rows
self.assertEqual(len(ds.columns), 3)
self.assertEqual(len(ds.fetch_rows()), 2)
self.assertEqual(ds.row_count, 2)
def dataset_read(self):
"""Test reading a dataset."""
self.setup_fileserver()
dh = self.db.load_dataset(self.fileserver.upload_file(CSV_FILE))
ds = self.db.get_dataset(dh.identifier)
ds_rows = ds.fetch_rows()
self.assertEqual(dh.identifier, ds.identifier)
self.assertEqual(len(dh.columns), len(ds.columns))
self.assertEqual(len(dh.fetch_rows()), len(ds_rows))
self.assertEqual(len(dh.fetch_rows()), len(ds_rows))
self.assertEqual(dh.row_count, len(ds_rows))
# Name,Age,Salary
# Alice,23,35K
# Bob,32,30K
self.assertEqual(ds.column_index('Name'), 0)
self.assertEqual(ds.column_index('Age'), 1)
self.assertEqual(ds.column_index('Salary'), 2)
row = ds_rows[0]
self.assertEqual(row.values[0], 'Alice')
self.assertEqual(int(row.values[1]), 23)
self.assertEqual(row.values[2], '35K')
row = ds_rows[1]
self.assertEqual(row.values[0], 'Bob')
self.assertEqual(int(row.values[1]), 32)
self.assertEqual(row.values[2], '30K')
class TestDatasetPaginationReader(unittest.TestCase):
def set_up(self, engine):
"""Create an empty file server repository."""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.mkdir(SERVER_DIR)
# Setup file server
self.fs = FileSystemFilestore(FILESERVER_DIR)
# Setup the respective datastore and Vizual engine
if engine == ENGINEENV_DEFAULT:
self.datastore = FileSystemDatastore(DATASTORE_DIR)
elif engine == ENGINEENV_MIMIR:
self.datastore = MimirDatastore(DATASTORE_DIR)
def tear_down(self, engine):
"""Clean-up by dropping file server directory.
"""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
def test_default_engine(self):
"""Test functionality for the default setup."""
self.run_tests(ENGINEENV_DEFAULT)
def test_mimir_engine(self):
"""Test functionality for the Mimir setup."""
import vizier.mimir as mimir # noqa: F401
self.run_tests(ENGINEENV_MIMIR)
def run_tests(self, engine):
"""Run sequence of tests for given configuration."""
self.set_up(engine)
ds = self.datastore.load_dataset(self.fs.upload_file(CSV_FILE_1))
rows = ds.fetch_rows()
self.assertEqual(len(rows), 7)
rows = ds.fetch_rows(offset=1)
self.assertEqual(len(rows), 6)
self.assertEqual(rows[0].values[0], 'Bob')
self.assertEqual(rows[5].values[0], 'Gertrud')
rows = ds.fetch_rows(limit=2)
self.assertEqual(len(rows), 2)
self.assertEqual(rows[0].values[0], 'Alice')
self.assertEqual(rows[1].values[0], 'Bob')
rows = ds.fetch_rows(offset=4, limit=3)
self.assertEqual(len(rows), 3)
self.assertEqual(rows[0].values[0], 'Eileen')
self.assertEqual(rows[2].values[0], 'Gertrud')
rows = ds.fetch_rows(offset=5, limit=3)
self.assertEqual(len(rows), 2)
self.assertEqual(rows[0].values[0], 'Frank')
self.assertEqual(rows[1].values[0], 'Gertrud')
rows = ds.fetch_rows(offset=6, limit=3)
self.assertEqual(len(rows), 1)
self.assertEqual(rows[0].values[0], 'Gertrud')
# Test larger dataset with deletes
ds = self.datastore.load_dataset(self.fs.upload_file(CSV_FILE_2))
rows = ds.fetch_rows(offset=0, limit=10)
self.assertEqual(len(rows), 10)
rows = ds.fetch_rows(offset=10, limit=20)
self.assertEqual(len(rows), 20)
rows = ds.fetch_rows(offset=60, limit=10)
self.assertEqual(len(rows), 3)
self.tear_down(engine)
class TestMimirProcessor(unittest.TestCase):
"""Individual test for Mimir lenses. Run separately since each test has to
initialize and shout down the Mimir gateway.
"""
def setUp(self):
"""Create an instance of the Mimir processor for an empty server
directory.
"""
# Drop directory if it exists
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.makedirs(SERVER_DIR)
self.processor = MimirProcessor()
self.datastore = MimirDatastore(DATASTORE_DIR)
self.filestore = FileSystemFilestore(FILESTORE_DIR)
def tearDown(self):
"""Clean-up by dropping the server directory.
"""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
def test_domain_lens(self):
"""Test DOMAIN lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(INCOMPLETE_CSV_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
col_age = ds.column_by_name('Age')
command = cmd.mimir_domain(DATASET_NAME, col_age.identifier)
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
rows = ds.fetch_rows()
self.assertNotEqual(rows[2].values[ds.column_index('Age')], '')
# Introduce an error. Make sure command formating is correct
command = cmd.mimir_domain('MY DS', 'MY COL')
with self.assertRaises(ValueError):
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
def test_geocode_lens(self):
"""Test GEOCODE lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(GEO_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
# Geocode Lens
command = cmd.mimir_geocode(
DATASET_NAME,
'GOOGLE',
house_nr=ds.column_by_name('STRNUMBER').identifier,
street=ds.column_by_name('STRNAME').identifier,
city=ds.column_by_name('CITY').identifier,
state=ds.column_by_name('STATE').identifier)
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
columns = [c.name for c in ds.columns]
self.assertEqual(len(columns), 6)
self.assertTrue('LATITUDE' in columns)
self.assertTrue('LONGITUDE' in columns)
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
columns = [c.name for c in ds.columns]
self.assertEqual(len(columns), 8)
self.assertTrue('LATITUDE_1' in columns)
self.assertTrue('LONGITUDE_1' in columns)
self.assertEqual(len(ds.columns), 8)
def test_key_repair_lens(self):
"""Test KEY REPAIR lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(KEY_REPAIR_FILE)
ds1 = self.datastore.load_dataset(f_handle=f_handle)
# Missing Value Lens
command = cmd.mimir_key_repair(DATASET_NAME,
ds1.column_by_name('Empid').identifier)
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds1.identifier}))
self.assertTrue(result.is_success)
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
self.assertEqual(len(ds.columns), 4)
self.assertEqual(ds.row_count, 3)
names = set()
empids = set()
rowids = set()
for row in ds.fetch_rows():
rowids.add(row.identifier)
empids.add(int(row.get_value('empid')))
names.add(row.get_value('name'))
self.assertTrue(1 in empids)
self.assertTrue(2 in rowids)
self.assertTrue('Alice' in names)
self.assertTrue('Carla' in names)
# Test error case and command text
command = cmd.mimir_key_repair('MY DS', 'MY COL')
with self.assertRaises(ValueError):
self.processor.compute(command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(
datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
def test_missing_value_lens(self):
"""Test MISSING_VALUE lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(INCOMPLETE_CSV_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
# Missing Value Lens
command = cmd.mimir_missing_value(
DATASET_NAME,
columns=[{
'column': ds.column_by_name('AGE').identifier
}])
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
rows = ds.fetch_rows()
for row in rows:
self.assertIsNotNone(row.values[1])
self.assertNotEqual(rows[2].values[ds.column_index('Age')], '')
# MISSING VALUE Lens with value constraint
command = cmd.mimir_missing_value(
DATASET_NAME,
columns=[{
'column': ds.column_by_name('AGE').identifier,
'constraint': '> 30'
}],
)
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
rows = ds.fetch_rows()
for row in rows:
self.assertIsNotNone(row.values[1])
self.assertTrue(rows[2].values[ds.column_index('Age')] > 30)
def test_missing_key_lens(self):
"""Test MISSING_KEY lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(INCOMPLETE_CSV_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
# Missing Value Lens
age_col = ds.column_by_name('Age').identifier
command = cmd.mimir_missing_key(DATASET_NAME, age_col)
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
self.assertEqual(len(ds.columns), 3)
rows = ds.fetch_rows()
self.assertEqual(len(rows), 24)
command = cmd.mimir_missing_key(DATASET_NAME,
ds.column_by_name('Salary').identifier)
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
self.assertEqual(len(ds.columns), 3)
rows = ds.fetch_rows()
self.assertEqual(len(rows), 55)
def test_picker_lens(self):
"""Test PICKER lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(PICKER_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
command = cmd.mimir_picker(
DATASET_NAME, [{
'pickFrom': ds.column_by_name('Age').identifier
}, {
'pickFrom': ds.column_by_name('Salary').identifier
}])
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
columns = [c.name for c in ds.columns]
print(columns)
self.assertEqual(len(ds.columns), 5)
self.assertTrue('PICK_ONE_AGE_SALARY' in columns)
# Pick another column, this time with custom name
command = cmd.mimir_picker(
DATASET_NAME, [{
'pickFrom': ds.column_by_name('Age').identifier
}, {
'pickFrom': ds.column_by_name('Salary').identifier
}],
pick_as='My_Column')
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
columns = [c.name for c in ds.columns]
self.assertEqual(len(ds.columns), 6)
self.assertTrue('PICK_ONE_AGE_SALARY' in columns)
self.assertTrue('MY_COLUMN' in columns)
# Pick from a picked column
command = cmd.mimir_picker(
DATASET_NAME,
[{
'pickFrom': ds.column_by_name('Age').identifier
}, {
'pickFrom': ds.column_by_name('PICK_ONE_AGE_SALARY').identifier
}],
pick_as='My_Next_Column')
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
columns = [c.name for c in ds.columns]
self.assertTrue('MY_NEXT_COLUMN' in columns)
def test_schema_matching_lens(self):
"""Test SCHEMA_MATCHING lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(CSV_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
# Missing Value Lens
command = cmd.mimir_schema_matching(DATASET_NAME, [{
'column': 'BDate',
'type': 'int'
}, {
'column': 'PName',
'type': 'varchar'
}], 'new_' + DATASET_NAME)
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write['new_' + DATASET_NAME].identifier)
self.assertEqual(len(ds.columns), 2)
self.assertEqual(ds.row_count, 2)
def test_type_inference_lens(self):
"""Test TYPE INFERENCE lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(INCOMPLETE_CSV_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
# Infer type
command = cmd.mimir_type_inference(DATASET_NAME, 0.6)
result = self.processor.compute(
command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(datastore=self.datastore,
filestore=self.filestore,
datasets={DATASET_NAME: ds.identifier}))
self.assertTrue(result.is_success)
# Get dataset
ds2 = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
self.assertEqual(len(ds2.columns), 3)
self.assertEqual(ds2.row_count, 7)
ds1_rows = ds.fetch_rows()
ds2_rows = ds2.fetch_rows()
for i in range(ds2.row_count):
self.assertEqual(ds1_rows[i].values, ds2_rows[i].values)
class TestDefaultVizualApi(unittest.TestCase):
api: MimirVizualApi
def setUp(self):
"""Create an instance of the default vizier API for an empty server
directory.
"""
# Drop directory if it exists
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.makedirs(SERVER_DIR)
self.api = MimirVizualApi()
self.datastore = MimirDatastore(DATASTORE_DIR)
self.filestore = FileSystemFilestore(FILESTORE_DIR)
def tearDown(self):
"""Clean-up by dropping the server directory.
"""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
def test_api(self):
"""Run all tests after we initialize mimir. Make sure to create a
fresh environment after each test.
"""
self.delete_column()
self.setUp()
self.delete_row()
self.setUp()
self.filter_columns()
self.setUp()
self.insert_column()
self.setUp()
self.insert_row()
self.setUp()
self.load_dataset()
self.setUp()
self.move_column()
self.setUp()
self.move_row()
self.setUp()
self.rename_column()
self.setUp()
self.sequence_of_steps()
self.setUp()
self.sort_dataset()
self.setUp()
self.update_cell()
def delete_column(self):
"""Test functionality to delete a column."""
# Create a new dataset
fh = self.filestore.upload_file(CSV_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
ds_rows = ds.fetch_rows()
# Keep track of column and row identifier
row_ids = [row.identifier for row in ds_rows]
# Delete Age column
col_id = ds.column_by_name('AGE').identifier
result = self.api.delete_column(ds.identifier, col_id, self.datastore)
# Resulting dataset should differ from previous one
self.assertNotEqual(result.dataset.identifier, ds.identifier)
# Retrieve modified dataset and ensure that it cobtains the following
#
# Name, Salary
# ------------
# Alice, 35K
# Bob, 30K
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
# Schema is Name, Salary
self.assertEqual(len(ds.columns), 2)
self.assertEqual(ds.columns[0].name.upper(), 'NAME')
self.assertEqual(ds.columns[1].name.upper(), 'SALARY')
# Make sure that all rows only have two columns
row = ds_rows[0]
self.assertEqual(len(row.values), 2)
self.assertEqual(len(row.values), 2)
self.assertEqual(row.values[0], 'Alice')
self.assertEqual(row.values[1], '35K')
row = ds_rows[1]
self.assertEqual(len(row.values), 2)
self.assertEqual(len(row.values), 2)
self.assertEqual(row.values[0], 'Bob')
self.assertEqual(row.values[1], '30K')
# Ensure that row identifier haven't changed
for i in range(len(ds_rows)):
self.assertEqual(ds_rows[i].identifier, row_ids[i])
# Ensure exception is thrown if dataset identifier is unknown
with self.assertRaises(MimirError):
self.api.delete_column('unknown:uri', 0, self.datastore)
# Ensure exception is thrown if column identifier is unknown
with self.assertRaises(ValueError):
self.api.delete_column(ds.identifier, 100, self.datastore)
def delete_row(self):
"""Test functionality to delete a row."""
# Create a new dataset
fh = self.filestore.upload_file(CSV_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
ds_rows = ds.fetch_rows()
# Keep track of column and row identifier
col_ids = [col.identifier for col in ds.columns]
row_ids = [row.identifier for row in ds_rows]
# Delete second row
result = self.api.delete_row(ds.identifier, row_ids[1], self.datastore)
del row_ids[1]
# Resulting dataset should differ from previous one
self.assertNotEqual(result.dataset.identifier, ds.identifier)
# Retrieve modified dataset and ensure that it contains the following
# data:
#
# Name, Age, Salary
# ------------
# Alice, 23, 35K
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
# Schema is Name, Salary
col_names = ['Name', 'Age', 'Salary']
self.assertEqual(len(ds.columns), len(col_names))
for i in range(len(ds.columns)):
self.assertEqual(ds.columns[i].name.upper(), col_names[i].upper())
# Make sure column identifier haven't changed
for i in range(len(ds.columns)):
self.assertEqual(ds.columns[i].identifier, col_ids[i])
# There should only be one row
self.assertEqual(len(ds_rows), 1)
# Ensure that row identifier haven't changed
for i in range(len(ds_rows)):
self.assertEqual(ds_rows[i].identifier, row_ids[i])
# Ensure exception is thrown if dataset is unknown
with self.assertRaises(MimirError):
self.api.delete_row('unknown:uri', 0, self.datastore)
def filter_columns(self):
"""Test projection of a dataset."""
# Create a new dataset
fh = self.filestore.upload_file(CSV_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
result = self.api.filter_columns(ds.identifier, [2, 0], ['BD', None],
self.datastore)
ds = self.datastore.get_dataset(result.dataset.identifier)
self.assertEqual(len(ds.columns), 2)
self.assertEqual(ds.columns[0].name.upper(), 'BD')
self.assertEqual(ds.columns[1].name.upper(), 'NAME')
rows = ds.fetch_rows()
self.assertEqual(rows[0].values, ['35K', 'Alice'])
self.assertEqual(rows[1].values, ['30K', 'Bob'])
def insert_column(self):
"""Test functionality to insert a columns."""
# Create a new dataset
fh = self.filestore.upload_file(CSV_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
ds_rows = ds.fetch_rows()
# Keep track of column and row identifier
col_ids = [col.identifier for col in ds.columns]
row_ids = [row.identifier for row in ds_rows]
# Insert columns at position 1
col_ids.insert(1, ds.max_column_id() + 1)
result = self.api.insert_column(ds.identifier, 1, 'Height',
self.datastore)
# Resulting dataset should differ from previous one
self.assertNotEqual(result.dataset.identifier, ds.identifier)
# Retrieve dataset and ensure that it has the following schema:
# Name, Height, Age, Salary
ds = self.datastore.get_dataset(result.dataset.identifier)
col_names = ['Name', 'Height', 'Age', 'Salary']
# Ensure that there are four rows
self.assertEqual(len(ds.columns), len(col_names))
print(ds.columns)
for i in range(len(col_names)):
col = ds.columns[i]
self.assertEqual(col.name.upper(), col_names[i].upper())
# Insert columns at last position
col_ids.append(ds.max_column_id() + 1)
col_names.append('Weight')
result = self.api.insert_column(ds.identifier, 4, 'Weight',
self.datastore)
# Resulting dataset should differ from previous one
self.assertNotEqual(result.dataset.identifier, ds.identifier)
# Retrieve dataset and ensure that it has the following schema:
# Name, Height, Age, Salary, Weight
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
# Ensure that there are five rows
self.assertEqual(len(ds.columns), len(col_names))
for i in range(len(col_names)):
col = ds.columns[i]
self.assertEqual(col.name.upper(), col_names[i].upper())
# The cell values for new columns are None all other values are not None
for row in ds_rows:
for i in range(len(ds.columns)):
if i == 1 or i == 4:
self.assertIsNone(row.values[i])
else:
self.assertTrue(row.values[i])
# Ensure that row identifier haven't changed
for i in range(len(ds_rows)):
self.assertEqual(ds_rows[i].identifier, row_ids[i])
# Ensure exception is thrown if dataset identifier is unknown
with self.assertRaises(MimirError):
self.api.insert_column('unknown:uri', 1, 'Height', self.datastore)
# Ensure exception is thrown if column name is invalid
self.api.insert_column(ds.identifier, 1, 'Height_from_ground',
self.datastore)
with self.assertRaises(ValueError):
self.api.insert_column(ds.identifier, 1, 'Height from ground!@#',
self.datastore)
# Ensure exception is thrown if column position is out of bounds
with self.assertRaises(ValueError):
self.api.insert_column(ds.identifier, 100, 'Height',
self.datastore)
def insert_row(self):
"""Test functionality to insert a row."""
fh = self.filestore.upload_file(CSV_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
# Keep track of column and row identifier
ds_rows = ds.fetch_rows()
row_ids = [row.identifier for row in ds_rows]
# Insert row at index position 1
row_ids.insert(1, None)
# Result should indicate that one row was inserted. The identifier of
# the resulting dataset should differ from the identifier of the
# original dataset
result = self.api.insert_row(ds.identifier, 1, self.datastore)
# Resulting dataset should differ from previous one
self.assertNotEqual(result.dataset.identifier, ds.identifier)
# Retrieve modified dataset
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
# Ensure that there are three rows
self.assertEqual(len(ds_rows), 3)
# The second row has empty values for each column
row = ds_rows[1]
self.assertEqual(len(row.values), len(ds.columns))
for i in range(len(ds.columns)):
self.assertIsNone(row.values[i])
# Append row at end current dataset
row_ids.append(None)
result = self.api.insert_row(ds.identifier, 3, self.datastore)
# Resulting dataset should differ from previous one
self.assertNotEqual(result.dataset.identifier, ds.identifier)
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
# Ensure that there are three rows
self.assertEqual(len(ds_rows), 4)
# The next to last row has non-empty values for each column
row = ds_rows[2]
self.assertEqual(len(row.values), len(ds.columns))
for i in range(len(ds.columns)):
self.assertIsNotNone(row.values[i])
# The last row has empty values for each column
row = ds_rows[3]
self.assertEqual(len(row.values), len(ds.columns))
for i in range(len(ds.columns)):
self.assertIsNone(row.values[i])
# Ensure that row ids haven't changed
# ## July 16, 2020 by OK: Bug in mimir that is going to take a bunch of
# ## heavy lifting to fix: https://github.com/UBOdin/mimir-api/issues/11
# for i in range(len(ds_rows)):
# if row_ids[i] is not None:
# self.assertEqual(int(ds_rows[i].identifier), int(row_ids[i]))
# Ensure exception is thrown if dataset identifier is unknown
with self.assertRaises(MimirError):
self.api.insert_row('unknown:uri', 1, self.datastore)
# Ensure no exception is raised
self.api.insert_row(ds.identifier, 4, self.datastore)
def load_dataset(self) -> None:
"""Test functionality to load a dataset."""
# Create a new dataset
fh = self.filestore.upload_file(CSV_FILE)
result = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier)
ds = result.dataset
resources = result.resources
assert (isinstance(ds, DatasetHandle))
ds_rows = ds.fetch_rows()
self.assertEqual(len(ds.columns), 3)
self.assertEqual(len(ds_rows), 2)
for row in ds_rows:
self.assertTrue(isinstance(row.values[1], int))
self.assertIsNotNone(resources)
self.assertEqual(resources[RESOURCE_FILEID], fh.identifier)
self.assertEqual(resources[RESOURCE_DATASET], ds.identifier)
# Delete file handle and oing the same should raise an exception
self.filestore.delete_file(fh.identifier)
with self.assertRaises(ValueError):
self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier)
# Ensure exception is thrown if dataset identifier is unknown
with self.assertRaises(ValueError):
self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id='unknown:uri')
# Test loading file from external resource. Skip if DOWNLOAD_URL is None
if DOWNLOAD_URL is None:
print('Skipping download test')
return
result = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
url=DOWNLOAD_URL,
options=[{
'delimiter': '\t'
}])
ds = result.dataset
resources = result.resources
ds_rows = ds.fetch_rows()
self.assertEqual(len(ds.columns), 4)
self.assertEqual(len(ds_rows), 54)
self.assertIsNotNone(resources)
self.assertEqual(resources[RESOURCE_URL], DOWNLOAD_URL)
self.assertEqual(resources[RESOURCE_DATASET], ds.identifier)
# Attempt to simulate re-running without downloading again. Set the
# Uri to some fake Uri that would raise an exception if an attempt was
# made to download
url = 'some fake uri'
resources[RESOURCE_URL] = url
result = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
url=url,
resources=resources)
prev_id = result.dataset.identifier
self.assertEqual(result.dataset.identifier, prev_id)
# If we re-run with reload flag true a new dataset should be returned
resources[RESOURCE_URL] = DOWNLOAD_URL
result = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
url=DOWNLOAD_URL,
resources=resources,
reload=True,
options=[{
'delimiter': '\t'
}])
self.assertNotEqual(result.dataset.identifier, prev_id)
def move_column(self):
"""Test functionality to move a column."""
# Create a new dataset
fh = self.filestore.upload_file(CSV_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
ds_rows = ds.fetch_rows()
# Keep track of column and row identifier
col_ids = [col.identifier for col in ds.columns]
row_ids = [row.identifier for row in ds_rows]
# Swap first two columns
c = col_ids[0]
del col_ids[0]
col_ids.insert(1, c)
result = self.api.move_column(ds.identifier,
ds.column_by_name('Name').identifier, 1,
self.datastore)
self.assertNotEqual(result.dataset.identifier, ds.identifier)
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
self.assertEqual(ds.columns[0].name.upper(), 'Age'.upper())
self.assertEqual(ds.columns[1].name.upper(), 'Name'.upper())
self.assertEqual(ds.columns[2].name.upper(), 'Salary'.upper())
row = ds_rows[0]
self.assertEqual(row.values[0], 23)
self.assertEqual(row.values[1], 'Alice')
self.assertEqual(row.values[2], '35K')
row = ds_rows[1]
self.assertEqual(row.values[0], 32)
self.assertEqual(row.values[1], 'Bob')
self.assertEqual(row.values[2], '30K')
# Ensure that row ids haven't changed
for i in range(len(ds_rows)):
self.assertEqual(int(ds_rows[i].identifier), int(row_ids[i]))
# Swap last two columns
c = col_ids[1]
del col_ids[1]
col_ids.append(c)
result = self.api.move_column(ds.identifier,
ds.column_by_name('Salary').identifier,
1, self.datastore)
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
self.assertEqual(ds.columns[0].name.upper(), 'Age'.upper())
self.assertEqual(ds.columns[1].name.upper(), 'Salary'.upper())
self.assertEqual(ds.columns[2].name.upper(), 'Name'.upper())
row = ds_rows[0]
self.assertEqual(row.values[0], 23)
self.assertEqual(row.values[1], '35K')
self.assertEqual(row.values[2], 'Alice')
row = ds_rows[1]
self.assertEqual(row.values[0], 32)
self.assertEqual(row.values[1], '30K')
self.assertEqual(row.values[2], 'Bob')
# Ensure that row ids haven't changed
for i in range(len(ds_rows)):
self.assertEqual(int(ds_rows[i].identifier), int(row_ids[i]))
# No changes if source and target position are the same
result = self.api.move_column(ds.identifier, ds.columns[1].identifier,
1, self.datastore)
self.assertEqual(ds.identifier, result.dataset.identifier)
# Ensure exception is thrown if dataset identifier is unknown
with self.assertRaises(MimirError):
self.api.move_column('unknown:uri', 0, 1, self.datastore)
# Raise error if source column is out of bounds
with self.assertRaises(ValueError):
self.api.move_column(ds.identifier, 40, 1, self.datastore)
# Raise error if target position is out of bounds
with self.assertRaises(ValueError):
self.api.move_column(ds.identifier,
ds.column_by_name('Name').identifier, -1,
self.datastore)
with self.assertRaises(ValueError):
self.api.move_column(ds.identifier,
ds.column_by_name('Name').identifier, 4,
self.datastore)
def move_row(self):
"""Test functionality to move a row."""
# Create a new dataset
fh = self.filestore.upload_file(CSV_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
ds_rows = ds.fetch_rows()
# Keep track of column and row identifier
row_ids = [row.identifier for row in ds_rows]
# Swap first two rows
result = self.api.move_row(ds.identifier, row_ids[0], 1,
self.datastore)
row_ids = [row for row in reversed(row_ids)]
self.assertNotEqual(result.dataset.identifier, ds.identifier)
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
self.assertEqual(ds.columns[0].name.upper(), 'Name'.upper())
self.assertEqual(ds.columns[1].name.upper(), 'Age'.upper())
self.assertEqual(ds.columns[2].name.upper(), 'Salary'.upper())
row = ds_rows[0]
self.assertEqual(row.values[0], 'Bob')
self.assertEqual(row.values[1], 32)
self.assertEqual(row.values[2], '30K')
row = ds_rows[1]
self.assertEqual(row.values[0], 'Alice')
self.assertEqual(row.values[1], 23)
self.assertEqual(row.values[2], '35K')
# Ensure that row ids haven't changed
for i in range(len(ds_rows)):
self.assertEqual(ds_rows[i].identifier, row_ids[i])
# Swap last two rows
result = self.api.move_row(ds.identifier, row_ids[1], 0,
self.datastore)
row_ids = [row for row in reversed(row_ids)]
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
self.assertEqual(ds.columns[0].name.upper(), 'Name'.upper())
self.assertEqual(ds.columns[1].name.upper(), 'Age'.upper())
self.assertEqual(ds.columns[2].name.upper(), 'Salary'.upper())
row = ds_rows[0]
self.assertEqual(row.values[0], 'Alice')
self.assertEqual(row.values[1], 23)
self.assertEqual(row.values[2], '35K')
row = ds_rows[1]
self.assertEqual(row.values[0], 'Bob')
self.assertEqual(row.values[1], 32)
self.assertEqual(row.values[2], '30K')
# Ensure that row ids haven't changed
for i in range(len(ds_rows)):
self.assertEqual(int(ds_rows[i].identifier), int(row_ids[i]))
# Move first row to the end
result = self.api.move_row(ds.identifier, row_ids[0], 2,
self.datastore)
row_ids = [row for row in reversed(row_ids)]
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
row = ds_rows[0]
self.assertEqual(row.values[0], 'Bob')
self.assertEqual(row.values[1], 32)
self.assertEqual(row.values[2], '30K')
row = ds_rows[1]
self.assertEqual(row.values[0], 'Alice')
self.assertEqual(row.values[1], 23)
self.assertEqual(row.values[2], '35K')
# Ensure that row ids haven't changed
# ## July 16, 2020 by OK: Bug in mimir that is going to take a bunch of
# ## heavy lifting to fix: https://github.com/UBOdin/mimir-api/issues/11
# for i in range(len(ds_rows)):
# self.assertEqual(int(ds_rows[i].identifier), int(row_ids[i]))
# No changes if source and target position are the same
result = self.api.move_row(ds.identifier, row_ids[1], 1,
self.datastore)
# ## July 21, 2020 by OK: It would be fantastic if we could easily detect
# no-op vizual, but for now skip this check
#self.assertEqual(ds.identifier, result.dataset.identifier)
# Ensure exception is thrown if dataset identifier is unknown
with self.assertRaises(MimirError):
self.api.move_row('unknown:uri', 0, 1, self.datastore)
# Raise error if target position is out of bounds
# ## July 21, 2020 by OK: Skipping this check for now
# with self.assertRaises(ValueError):
# self.api.move_row(ds.identifier, 0, -1, self.datastore)
# with self.assertRaises(ValueError):
# self.api.move_row(ds.identifier, 1, 4, self.datastore)
def rename_column(self):
"""Test functionality to rename a column."""
# Create a new dataset
fh = self.filestore.upload_file(CSV_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
ds_rows = ds.fetch_rows()
# Keep track of column and row identifier
row_ids = [row.identifier for row in ds_rows]
# Rename first column to Firstname
result = self.api.rename_column(ds.identifier,
ds.column_by_name('Name').identifier,
'Firstname', self.datastore)
self.assertNotEqual(result.dataset.identifier, ds.identifier)
ds = self.datastore.get_dataset(result.dataset.identifier)
self.assertEqual(ds.columns[0].name.upper(), 'Firstname'.upper())
self.assertEqual(ds.columns[1].name.upper(), 'Age'.upper())
self.assertEqual(ds.columns[2].name.upper(), 'Salary'.upper())
result = self.api.rename_column(ds.identifier,
ds.column_by_name('Age').identifier,
'BDate', self.datastore)
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
self.assertEqual(ds.columns[0].name.upper(), 'Firstname'.upper())
self.assertEqual(ds.columns[1].name, 'BDate')
self.assertEqual(ds.columns[2].name.upper(), 'Salary'.upper())
# Ensure that row ids haven't changed
for i in range(len(ds_rows)):
self.assertEqual(int(ds_rows[i].identifier), int(row_ids[i]))
# No changes if the old and new column name are the same (with exception
# to upper and lower cases).
result = self.api.rename_column(ds.identifier,
ds.column_by_name('BDate').identifier,
'BDate', self.datastore)
# ## July 21, 2020 by OK: It would be fantastic if we could easily detect
# no-op vizual, but for now skip this check
# self.assertEqual(ds.identifier, result.dataset.identifier)
# Ensure exception is thrown if dataset identifier is unknown
with self.assertRaises(MimirError):
self.api.rename_column('unknown:uri', 0, 'Firstname',
self.datastore)
# Ensure exception is thrown for invalid column id
with self.assertRaises(ValueError):
self.api.rename_column(ds.identifier, 500, 'BDate', self.datastore)
def sequence_of_steps(self):
"""Test sequence of calls that modify a dataset."""
# Create a new dataset
fh = self.filestore.upload_file(CSV_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
ds = self.api.insert_row(ds.identifier, 1, self.datastore).dataset
row_ids = [row.identifier for row in ds.fetch_rows()]
row0 = row_ids[0]
row1 = row_ids[1]
row2 = row_ids[2]
ds = self.api.insert_column(ds.identifier, 3, 'HDate',
self.datastore).dataset
ds = self.api.update_cell(ds.identifier,
ds.column_by_name('HDate').identifier, row0,
'180', self.datastore).dataset
ds = self.api.update_cell(ds.identifier,
ds.column_by_name('HDate').identifier, row2,
'160', self.datastore).dataset
ds = self.api.rename_column(ds.identifier,
ds.column_by_name('HDate').identifier,
'Height', self.datastore).dataset
ds = self.api.update_cell(ds.identifier,
ds.column_by_name('Height').identifier, row1,
'170', self.datastore).dataset
ds = self.api.move_row(ds.identifier, row1, 2, self.datastore).dataset
ds = self.api.update_cell(ds.identifier,
ds.column_by_name('Name').identifier, row2,
'Carla', self.datastore).dataset
ds = self.api.update_cell(ds.identifier,
ds.column_by_name('Age').identifier, row2,
'45', self.datastore).dataset
ds = self.api.update_cell(ds.identifier,
ds.column_by_name('Salary').identifier, row2,
'56K', self.datastore).dataset
ds = self.api.move_column(ds.identifier,
ds.column_by_name('Salary').identifier, 4,
self.datastore).dataset
ds = self.api.delete_column(ds.identifier,
ds.column_by_name('Age').identifier,
self.datastore).dataset
ds = self.api.delete_row(ds.identifier, row0, self.datastore).dataset
ds = self.api.delete_row(ds.identifier, row1, self.datastore).dataset
ds = self.datastore.get_dataset(ds.identifier)
ds_rows = ds.fetch_rows()
names = ['Name', 'Height', 'Salary']
self.assertEqual(len(ds.columns), len(names))
for i in range(len(names)):
col = ds.columns[i]
self.assertEqual(col.name.upper(), names[i].upper())
self.assertEqual(len(ds_rows), 1)
self.assertEqual(ds_rows[0].values, ['Carla', '160', '56K'])
def sort_dataset(self):
"""Test sorting a dataset."""
# Create a new dataset
fh = self.filestore.upload_file(SORT_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
result = self.api.sort_dataset(ds.identifier, [1, 2, 0],
[False, False, True], self.datastore)
ds = self.datastore.get_dataset(result.dataset.identifier)
rows = ds.fetch_rows()
names = ['Alice', 'Bob', 'Dave', 'Gertrud', 'Frank']
result = list()
for row in rows:
name = row.values[0]
if name in names:
result.append(name)
for i in range(len(names)):
self.assertEqual(names[i], result[i])
result = self.api.sort_dataset(ds.identifier, [2, 1, 0],
[True, False, True], self.datastore)
ds = self.datastore.get_dataset(result.dataset.identifier)
rows = ds.fetch_rows()
names = ['Gertrud', 'Frank', 'Bob', 'Alice', 'Dave']
result = list()
for row in rows:
name = row.values[0]
if name in names:
result.append(name)
for i in range(len(names)):
self.assertEqual(names[i], result[i])
# Raises error for invalid column identifier
with self.assertRaises(ValueError):
self.api.sort_dataset(ds.identifier, [2, 10, 0],
[True, False, True], self.datastore)
def update_cell(self):
"""Test functionality to update a dataset cell."""
# Create a new dataset
fh = self.filestore.upload_file(CSV_FILE)
ds = self.api.load_dataset(datastore=self.datastore,
filestore=self.filestore,
file_id=fh.identifier).dataset
ds_rows = ds.fetch_rows()
# Keep track of column and row identifier
row_ids = [row.identifier for row in ds_rows]
# Update cell [0, 0]. Ensure that one row was updated and a new
# identifier is generated. Also ensure that the resulting datasets
# has the new value in cell [0, 0]
row_id = row_ids[0]
result = self.api.update_cell(ds.identifier, 0, row_id, 'MyValue',
self.datastore)
self.assertNotEqual(ds.identifier, result.dataset.identifier)
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
row = None
for r in ds.fetch_rows():
if r.identifier == row_id:
row = r
break
self.assertEqual(row.values[0], 'MyValue')
result = self.api.update_cell(ds.identifier,
ds.column_by_name('Name').identifier,
row_id, 'AValue', self.datastore)
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
row = None
for r in ds.fetch_rows():
if r.identifier == row_id:
row = r
break
self.assertEqual(row.values[0], 'AValue')
self.assertEqual(row.values[ds.column_index('Name')], 'AValue')
# Ensure that row ids haven't changed
for i in range(len(ds_rows)):
self.assertEqual(int(ds_rows[i].identifier), int(row_ids[i]))
# Set value to None
result = self.api.update_cell(ds.identifier,
ds.column_by_name('Name').identifier,
row_id, None, self.datastore)
ds = self.datastore.get_dataset(result.dataset.identifier)
ds_rows = ds.fetch_rows()
row = None
for r in ds.fetch_rows():
if r.identifier == row_id:
row = r
break
self.assertIsNone(row.values[0])
self.assertIsNone(row.values[ds.column_index('Name')])
# Ensure exception is thrown if dataset is unknown
with self.assertRaises(MimirError):
self.api.update_cell('unknown:uri', 0, 0, 'MyValue',
self.datastore)
class TestMimirProcessor(unittest.TestCase):
"""Individual test for Mimir lenses. Run separately since each test has to
initialize and shout down the Mimir gateway.
"""
def setUp(self):
"""Create an instance of the Mimir processor for an empty server
directory.
"""
# Drop directory if it exists
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
os.makedirs(SERVER_DIR)
self.processor = MimirProcessor()
self.datastore = MimirDatastore(DATASTORE_DIR)
self.filestore = FileSystemFilestore(FILESTORE_DIR)
self.available_lenses = set(mimir.getAvailableLensTypes())
def tearDown(self):
"""Clean-up by dropping the server directory.
"""
if os.path.isdir(SERVER_DIR):
shutil.rmtree(SERVER_DIR)
def compute_lens_result(self, ds, command):
return self.processor.compute(command_id=command.command_id,
arguments=command.arguments,
context=TaskContext(
project_id=1,
datastore=self.datastore,
filestore=self.filestore,
artifacts={DATASET_NAME: ds}))
def test_geocode_lens(self):
if lens_types.MIMIR_GEOCODE not in self.available_lenses:
self.skipTest("Mimir Geocoding Lens not initialized.")
"""Test GEOCODE lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(GEO_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
# Geocode Lens
command = cmd.mimir_geocode(
DATASET_NAME,
'GOOGLE',
house_nr=ds.column_by_name('STRNUMBER').identifier,
street=ds.column_by_name('STRNAME').identifier,
city=ds.column_by_name('CITY').identifier,
state=ds.column_by_name('STATE').identifier)
result = self.compute_lens_result(ds, command)
self.assertTrue(result.is_success)
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
columns = [c.name for c in ds.columns]
self.assertEqual(len(columns), 6)
self.assertTrue('LATITUDE' in columns)
self.assertTrue('LONGITUDE' in columns)
result = self.compute_lens_result(ds, command)
self.assertTrue(result.is_success)
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
columns = [c.name for c in ds.columns]
self.assertEqual(len(columns), 8)
self.assertTrue('LATITUDE_1' in columns)
self.assertTrue('LONGITUDE_1' in columns)
self.assertEqual(len(ds.columns), 8)
def test_key_repair_lens(self):
"""Test KEY REPAIR lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(KEY_REPAIR_FILE)
ds1 = self.datastore.load_dataset(f_handle=f_handle)
# Missing Value Lens
command = cmd.mimir_key_repair(DATASET_NAME,
ds1.column_by_name('Empid').identifier)
result = self.compute_lens_result(ds1, command)
self.assertTrue(result.is_success)
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
self.assertEqual(len(ds.columns), 4)
self.assertEqual(ds.row_count, 2)
names = set()
empids = set()
for row in ds.fetch_rows():
empids.add(int(row.values[0]))
names.add(row.values[1])
self.assertTrue(1 in empids)
self.assertTrue('Alice' in names or 'Bob' in names)
self.assertFalse('Alice' in names and 'Bob' in names)
self.assertTrue('Carla' in names)
# Test error case and command text
with self.assertRaises(ValueError):
command = cmd.mimir_key_repair('MY DS', 'MY COL')
result = self.compute_lens_result(ds, command)
def test_missing_value_lens(self):
"""Test MISSING_VALUE lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(INCOMPLETE_CSV_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
# Missing Value Lens
command = cmd.mimir_missing_value(
DATASET_NAME,
columns=[{
'column': ds.column_by_name('AGE').identifier
}])
result = self.compute_lens_result(ds, command)
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
rows = ds.fetch_rows()
for row in rows:
self.assertIsNotNone(row.values[1])
self.assertNotEqual(rows[2].values[ds.column_index('Age')], '')
# MISSING VALUE Lens with value constraint
command = cmd.mimir_missing_value(
DATASET_NAME,
columns=[{
'column': ds.column_by_name('AGE').identifier,
'constraint': '> 30'
}],
)
result = self.compute_lens_result(ds, command)
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
rows = ds.fetch_rows()
for row in rows:
self.assertIsNotNone(row.values[1])
print(rows[2].values)
# we shouldn't be imputing a value lower than the minimum value in the dataset
self.assertTrue(rows[2].values[ds.column_index('Age')] >= 23)
def test_missing_key_lens(self):
"""Test MISSING_KEY lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(INCOMPLETE_CSV_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
# Missing Value Lens
age_col = ds.column_by_name('Age').identifier
command = cmd.mimir_missing_key(DATASET_NAME, age_col)
result = self.compute_lens_result(ds, command)
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
self.assertEqual(len(ds.columns), 3)
rows = ds.fetch_rows()
# Depending on implementation this could be either 22 or 24, as there are two rows
# with missing values for the key column. Currently, Mimir discards such rows, but
# if this suddenly turns into a 24, that's not incorrect either.
self.assertEqual(len(rows), 22)
command = cmd.mimir_missing_key(DATASET_NAME,
ds.column_by_name('Salary').identifier)
result = self.compute_lens_result(ds, command)
self.assertTrue(result.is_success)
# Get dataset
ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
self.assertEqual(len(ds.columns), 3)
rows = ds.fetch_rows()
self.assertEqual(len(rows), 31)
def test_picker_lens(self):
"""Test PICKER lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(PICKER_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
command = cmd.mimir_picker(
DATASET_NAME, [{
'pickFrom': ds.column_by_name('Age').identifier
}, {
'pickFrom': ds.column_by_name('Salary').identifier
}])
result = self.compute_lens_result(ds, command)
self.assertTrue(result.is_success)
# Get dataset
result_ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
columns = [c.name for c in result_ds.columns]
# print(columns)
self.assertEqual(len(result_ds.columns), 3)
self.assertTrue('AGE_1' in columns)
# Pick another column, this time with custom name
command = cmd.mimir_picker(
DATASET_NAME, [{
'pickFrom': ds.column_by_name('Age').identifier
}, {
'pickFrom': ds.column_by_name('Salary').identifier
}],
pick_as='My_Column')
result = self.compute_lens_result(ds, command)
self.assertTrue(result.is_success)
# Get dataset
result_ds = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
columns = [c.name for c in result_ds.columns]
self.assertEqual(len(result_ds.columns), 3)
self.assertTrue('MY_COLUMN' in columns)
def test_type_inference_lens(self):
"""Test TYPE INFERENCE lens."""
# Create new work trail and retrieve the HEAD workflow of the default
# branch
f_handle = self.filestore.upload_file(INCOMPLETE_CSV_FILE)
ds = self.datastore.load_dataset(f_handle=f_handle)
# Infer type
command = cmd.mimir_type_inference(DATASET_NAME, 0.6)
result = self.compute_lens_result(ds, command)
self.assertTrue(result.is_success)
# Get dataset
ds2 = self.datastore.get_dataset(
result.provenance.write[DATASET_NAME].identifier)
self.assertEqual(len(ds2.columns), 3)
self.assertEqual(ds2.row_count, 7)
ds1_rows = ds.fetch_rows()
ds2_rows = ds2.fetch_rows()
for i in range(ds2.row_count):
self.assertEqual(ds1_rows[i].values, ds2_rows[i].values)
