def small_csv():
"""
Tests that we can select some data with the client
"""
cleanup()
logger.debug('starting setup')
# put the file onto s3
datafile = os.path.join(cur_dir, 'data/small_unquoted.csv')
s3.Bucket(bucket).upload_file(datafile, key)
logger.debug('uploaded file')
# create the table in Athena
query = '''CREATE EXTERNAL TABLE `{db}`.`{table}`
(`id` int, `date` date, `float` float, `message` string)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY ','
LINES TERMINATED BY '\\n'
LOCATION 's3://{bucket}/{path}/'
tblproperties ("skip.header.line.count"="1")'''.format(db=database,
table=table,
bucket=bucket,
path=path)
Client(results=bucket).athena_query(
'create database if not exists {}'.format(database))
logger.debug('created database')
Client(results=bucket).athena_query(query)
logger.debug('created table')
yield
cleanup()
def test_wait(mocker):
"""
Assert that for some good result, the wait function succeeds
"""
client = Client()
mocker.patch.object(client.client, 'get_query_execution')
response = {'QueryExecution': {'Status': {'State': 'SUCCEEDED'}}}
client.client.get_query_execution.return_value = response
result = client.wait_for_results('abc123')
assert result
def cleanup():
Client(results=bucket).athena_query(
'drop database if exists {} cascade'.format(database))
logger.debug('dropped database')
boto3.client('s3').delete_object(Bucket=bucket, Key=key)
logger.debug('deleted object')
def test_show_databases(small_csv):
"""
Test that we can inspect databases
"""
query = 'show databases'
result = Client(results='pythena-int').execute(query)
print(type(result))
def test_select_data(small_csv):
"""
Tests that we can select some data with the client
"""
query = 'select * from {}.{}'.format(database, table)
result = Client(results='pythena-int').execute(query)
print(result)
print(type(result))
def test_select_constant(small_csv):
"""
Tests that we can select a constant value (i.e. 1) - no data dependency
"""
query = 'select 1'
result = Client(results='pythena-int').execute(query)
print(result)
print(type(result))
def test_show_tables(small_csv):
"""
Test that we can inspect list of tables in our test database
"""
query = 'show tables in {}'.format(database)
result = Client(results='pythena-int').execute(query)
print(result)
print(type(result))
def test_select_single_row(small_csv):
"""
Tests that a where clause helps and works
"""
query = 'select * from {}.{} where id = 1'.format(database, table)
result = Client(results='pythena-int').athena_query(query)
assert len(result) == 1
assert result['message'][0] == 'something great!'
def test_show_databases(small_csv):
"""
Test that we can inspect databases
"""
query = 'show databases'
result = Client(results='pythena-int').athena_query(query)
print(type(result))
assert sum(result[0] == database) == 1
def test_select_constant(small_csv):
"""
Tests that we can select a constant value (i.e. 1) - no data dependency
"""
query = 'select 1'
result = Client(results='pythena-int').athena_query(query)
print(result)
print(type(result))
assert result['_col0'][0] == 1
def test_is_select_query():
""" Tests for select queries """
client = Client()
assert client._is_select_query('select 1')
assert client._is_select_query(' SELECT 1')
assert client._is_select_query('SELECT `insert`, `update`, from nothing')
def test_parse_create():
""" Test that a create query doesn't return columns """
client = Client()
columns = client._get_column_names('create table abc (id int) ')
assert columns is None
def test_parse_simple_select():
""" Test that a very simple select statement can be parsed """
client = Client()
columns = client._get_column_names('select a, b, c from table')
assert columns == ['a', 'b', 'c']
def test_is_not_select_query():
""" Tests for not select queries """
client = Client()
assert not client._is_select_query('update something set a = b')
assert not client._is_select_query('create table something awful')
assert not client._is_select_query('show databases')