def test_data_frame_type_check():
df = pd.DataFrame({
'String': pd.Categorical(['a', 'a', 'b', 'a']),
'Date': pd.Timestamp('20130102'),
'Value': np.array([3] * 4, dtype='int32')
})
config = __generate_cfg(df)
# Each referenced col must exist in the data frame
with pytest.raises(Exception) as e_info:
noDateFrame = pd.DataFrame({
'String':
pd.Categorical(['a', 'a', 'b', 'a']),
'Value':
np.array([3] * 4, dtype='int32')
})
validate_data_loader(config, noDateFrame)
assert 'data frame missing required field: Date' in str(e_info)
# Each referenced col must be the correct type
with pytest.raises(Exception) as e_info:
wrongTypeFrame = pd.DataFrame({
'String': pd.Timestamp('20130102'),
'Date': pd.Timestamp('20130102'),
'Value': np.array([3] * 4, dtype='int32')
})
validate_data_loader(config, wrongTypeFrame)
assert 'ata frame field String must be of type StringColumnConfig.' in str(
e_info)
def test_no_tuples():
# No time/value tuples
with pytest.raises(Exception) as e_info:
df = pd.DataFrame({
'String': pd.Categorical(['a', 'a', 'b', 'a']),
'Date': pd.Timestamp('20130102')
})
validate_data_loader(__generate_cfg(df))
assert 'Time tuples empty. No column loaded.' in str(e_info)
def test_valid_config():
# Completely valid config
df = pd.DataFrame({
'Date': pd.Timestamp('20130102'),
'Value': np.array([3] * 4, dtype='int32')
})
assert validate_data_loader(__generate_cfg(df), df)
def publish(session: dv_auth.Session, dataSourceId, df):
'''
Publish a DataFrame to Datavore that was configured on the client
'''
# Get the current loader config
currentConfig = get_data_loader_config(session, dataSourceId)
# Validate the config against the dataframe
if not dataload_validate.validate_data_loader(currentConfig, df):
raise Exception('Could not validate config.')
# Cancel load if it exists
try:
__cancel_data_load(session, dataSourceId)
except:
print('No job to cancel. Continuing.')
# @todo: log debug here -- exception if never loaded before
# Generate upload url
uploadUrl = __get_pre_signed_url(session, dataSourceId)
print('Uploading data frame...')
# Update the config with the URL @todo: use patch so we don't send the entire data brick back up :/
# currentConfig['uploadUrl'] = uploadUrl
# __setDatasourceLoaderConfig(session, dataSourceId, currentConfig)
# Put data to the uploadUrl
retries = 2
with tempfile.NamedTemporaryFile(mode='r+') as temp:
df.to_csv(temp.name, index=False, date_format='%Y-%m-%dT%H:%M:%SZ')
while retries > 0:
with open(temp.name, mode='rb') as csvFile:
res = requests.put(uploadUrl, data=csvFile)
if res.status_code == 200:
print('Data frame uploaded. Datavore load started.')
return res
elif res.status_code == 401:
retries -= 1
print(
f'Session upload error. Log in again. Attempts remaining: {retries}'
)
session = dv_auth.login(session.user_name,
session.env_conf)
else:
raise Exception(res.status_code,
res.content.decode('ascii'))
# Ran out of retries and never managed to upload :(
raise Exception('Failed to upload.')
def set_data_source_sample(session: dv_auth.Session, data_source_id, df):
'''
Sets the sample for a data source.
Once the sample is set, go to the Datavore client to configure the load.
Immutable on input DataFrame
:param session: Session - The Datavore session to use
:param data_source_id: String - The data source to set the sample for
:param df: - DataFrame to use as a sample source
:return: response - The result from setting the sample. Not very useful.
:raises Exception: on invalid DataFrame config or sample generation.
'''
# get the samples
sample = df_util.get_sample(df)
# Build our default config
column_configs = dataload_domain.get_column_configs(df)
data_source_meta = dataload_domain.ds_meta(data_source_id=data_source_id,
datasource='python',
publisher=session.user_name,
dataset='data frame')
loader_config = dataload_domain.csv_loader_config(
source_settings=dataload_domain.csv_source_settings(column_configs),
mapping=dataload_domain.simple_load_mapping(column_configs),
data_source_meta=data_source_meta,
sample_data=sample['sampleData'],
column_samples=sample['columnSamples'])
data_loader = dataload_domain.csv_data_loader(data_source_id,
loader_config)
# Validate the config against the dataframe
if not dataload_validate.validate_data_loader(data_loader, df):
raise Exception('Could not validate config.')
# save the loaderConfig
out = set_data_loader_config(session, data_source_id, data_loader)
print('Sample uploaded. Go to the Datavore client to review')
return out
def test_config_type_check():
df = pd.DataFrame({
'String': pd.Categorical(['a', 'a', 'b', 'a']),
'Date': pd.Timestamp('20130102'),
'Value': np.array([3] * 4, dtype='int32')
})
baseConfig = __generate_cfg(df)
# all keyColumns defined
with pytest.raises(Exception) as e_info:
local = copy.deepcopy(baseConfig)
local['loaderConfig']['mapping']['keyColumns'] = ['NotAField']
validate_data_loader(local)
assert 'key column NotAField not found' in str(e_info)
# all keyColumns are strings
with pytest.raises(Exception) as e_info:
local = copy.deepcopy(baseConfig)
local['loaderConfig']['mapping']['keyColumns'] = ['Date']
validate_data_loader(local)
assert 'key column Date must be a string' in str(e_info)
# all valueLabelColumn are defined
with pytest.raises(Exception) as e_info:
local = copy.deepcopy(baseConfig)
local['loaderConfig']['mapping']['valueLabelColumn'] = ['NotAField']
validate_data_loader(local)
assert 'value label NotAField not found' in str(e_info)
# all valueLabelColumn are strings
with pytest.raises(Exception) as e_info:
local = copy.deepcopy(baseConfig)
local['loaderConfig']['mapping']['valueLabelColumn'] = ['Date']
validate_data_loader(local)
assert 'value label Date must be a string' in str(e_info)
# all timeColumns are defined
with pytest.raises(Exception) as e_info:
local = copy.deepcopy(baseConfig)
local['loaderConfig']['mapping']['timeColumns'] = ['NotAField']
validate_data_loader(local)
assert 'time column NotAField not found' in str(e_info)
# all timeColumns are time
with pytest.raises(Exception) as e_info:
local = copy.deepcopy(baseConfig)
local['loaderConfig']['mapping']['timeColumns'] = ['String']
validate_data_loader(local)
assert 'time column String must be a time' in str(e_info)
# all timeTuples times are defined
with pytest.raises(Exception) as e_info:
local = copy.deepcopy(baseConfig)
local['loaderConfig']['mapping']['timeTuples'][0][
'timeColumn'] = 'NotAField'
validate_data_loader(local)
assert 'not found' in str(e_info)
# all timeTuples times are time
with pytest.raises(Exception) as e_info:
local = copy.deepcopy(baseConfig)
local['loaderConfig']['mapping']['timeTuples'][0][
'timeColumn'] = 'String'
validate_data_loader(local)
assert 'must be a time' in str(e_info)
# all timeTuples values are defined
with pytest.raises(Exception) as e_info:
local = copy.deepcopy(baseConfig)
local['loaderConfig']['mapping']['timeTuples'][0][
'valueColumn'] = 'NotAField'
validate_data_loader(local)
assert 'not found' in str(e_info)
# all timeTuples values are Number
with pytest.raises(Exception) as e_info:
local = copy.deepcopy(baseConfig)
local['loaderConfig']['mapping']['timeTuples'][0][
'valueColumn'] = 'String'
validate_data_loader(local)
assert 'must be a number' in str(e_info)
def test_no_time():
# No time columns
with pytest.raises(Exception) as e_info:
df = pd.DataFrame({})
validate_data_loader(__generate_cfg(df))
assert 'Loader config requires non-empty time columns.' in str(e_info)
def test_empty():
# No config
with pytest.raises(Exception) as e_info:
validate_data_loader({})
assert 'Empty loader config' in str(e_info)
