def test_blob_info_json_mapping(self):
""" Tests serialization of json ingestion blob info. """
validation_policy = ValidationPolicy(
ValidationOptions.ValidateCsvInputConstantColumns,
ValidationImplications.BestEffort)
properties = IngestionProperties(
database="database",
table="table",
dataFormat=DataFormat.json,
mapping=[JsonColumnMapping("ColumnName", "jsonpath", "datatype")],
additionalTags=["tag"],
ingestIfNotExists=["ingestIfNotExistTags"],
ingestByTags=["ingestByTags"],
dropByTags=["dropByTags"],
flushImmediately=True,
reportLevel=ReportLevel.DoNotReport,
reportMethod=ReportMethod.QueueAndTable,
validationPolicy=validation_policy,
)
blob = BlobDescriptor("somepath", 10)
blob_info = _IngestionBlobInfo(blob,
properties,
deleteSourcesOnSuccess=True,
authContext="authorizationContextText")
self._verify_ingestion_blob_info_result(blob_info.to_json())
def test_blob_info_csv_mapping(self):
"""Tests serialization of csv ingestion blob info."""
validation_policy = ValidationPolicy(
ValidationOptions.ValidateCsvInputConstantColumns,
ValidationImplications.BestEffort)
columnMapping = ColumnMapping("ColumnName", "cslDataType", ordinal=1)
properties = IngestionProperties(
database="database",
table="table",
dataFormat=DataFormat.CSV,
ingestionMapping=[columnMapping],
additionalTags=["tag"],
ingestIfNotExists=["ingestIfNotExistTags"],
ingestByTags=["ingestByTags"],
dropByTags=["dropByTags"],
flushImmediately=True,
reportLevel=ReportLevel.DoNotReport,
reportMethod=ReportMethod.Queue,
validationPolicy=validation_policy,
)
blob = BlobDescriptor("somepath", 10)
blob_info = _IngestionBlobInfo(blob,
properties,
auth_context="authorizationContextText")
self._verify_ingestion_blob_info_result(blob_info.to_json())
def test_blob_info_json_mapping(self):
"""Tests serialization of json ingestion blob info."""
validation_policy = ValidationPolicy(
ValidationOptions.ValidateCsvInputConstantColumns,
ValidationImplications.BestEffort)
properties = IngestionProperties(
database="database",
table="table",
data_format=DataFormat.JSON,
column_mappings=[
ColumnMapping("ColumnName", "datatype", path="jsonpath")
],
additional_tags=["tag"],
ingest_if_not_exists=["ingestIfNotExistTags"],
ingest_by_tags=["ingestByTags"],
drop_by_tags=["dropByTags"],
flush_immediately=True,
report_level=ReportLevel.DoNotReport,
report_method=ReportMethod.Queue,
validation_policy=validation_policy,
)
blob = BlobDescriptor("somepath", 10)
blob_info = IngestionBlobInfo(blob,
properties,
auth_context="authorizationContextText")
self._verify_ingestion_blob_info_result(blob_info.to_json())
def test_ingest_complicated_props():
validation_policy = ValidationPolicy(
validation_options=ValidationOptions.ValidateCsvInputConstantColumns,
validation_implications=ValidationImplications.Fail)
json_ingestion_props = IngestionProperties(
test_db,
test_table,
data_format=DataFormat.JSON,
ingestion_mapping=TestData.test_table_json_mappings(),
additional_tags=["a", "b"],
ingest_if_not_exists=["aaaa", "bbbb"],
ingest_by_tags=["ingestByTag"],
drop_by_tags=["drop", "drop-by"],
flush_immediately=False,
report_level=ReportLevel.FailuresAndSuccesses,
report_method=ReportMethod.Queue,
validation_policy=validation_policy,
)
file_paths = [json_file_path, zipped_json_file_path]
fds = [FileDescriptor(fp, 0, uuid.uuid4()) for fp in file_paths]
for fd in fds:
ingest_client.ingest_from_file(fd, json_ingestion_props)
assert_rows_added(4)
def test_ingest_complicated_props():
validation_policy = ValidationPolicy(
validationOptions=ValidationOptions.ValidateCsvInputConstantColumns,
validationImplications=ValidationImplications.Fail)
json_ingestion_props = IngestionProperties(
db_name,
table_name,
dataFormat=DataFormat.JSON,
ingestionMapping=Helpers.create_test_table_json_mappings(),
additionalTags=["a", "b"],
ingestIfNotExists=["aaaa", "bbbb"],
ingestByTags=["ingestByTag"],
dropByTags=["drop", "drop-by"],
flushImmediately=False,
reportLevel=ReportLevel.FailuresAndSuccesses,
reportMethod=ReportMethod.Queue,
validationPolicy=validation_policy,
)
file_paths = [json_file_path, zipped_json_file_path]
fds = [FileDescriptor(fp, 0, uuid.uuid4()) for fp in file_paths]
for fd in fds:
ingest_client.ingest_from_file(fd, json_ingestion_props)
assert_success_mesagges_count(2)
assert_row_count(4)
def test_ingest_complicated_props():
# Test ingest with complicated ingestion properties
validation_policy = ValidationPolicy(
validationOptions=ValidationOptions.ValidateCsvInputConstantColumns,
validationImplications=ValidationImplications.Fail,
)
json_ingestion_props = IngestionProperties(
db_name,
table_name,
dataFormat=DataFormat.json,
mapping=Helpers.create_deft_table_json_mappings(),
additionalTags=["a", "b"],
ingestIfNotExists=["aaaa", "bbbb"],
ingestByTags=["ingestByTag"],
dropByTags=["drop", "drop-by"],
flushImmediately=False,
reportLevel=ReportLevel.FailuresAndSuccesses,
reportMethod=ReportMethod.Queue,
validationPolicy=validation_policy,
)
file_paths = [json_file_path, zipped_json_file_path]
fds = [FileDescriptor(fp, 0, uuid.uuid4()) for fp in file_paths]
source_ids = ["{}".format(fd.source_id) for fd in fds]
for fd in fds:
ingest_client.ingest_from_file(fd, json_ingestion_props)
successes = 0
timeout = 60
while successes != 2 and timeout > 0:
while ingest_status_q.success.is_empty() and timeout > 0:
time.sleep(1)
timeout -= 1
success_message = ingest_status_q.success.pop()
if success_message[0].IngestionSourceId in source_ids:
assert success_message[0].Database == db_name
assert success_message[0].Table == table_name
successes += 1
assert successes == 2
# TODO: status queues only mark ingestion was successful, but takes time for data to become available
time.sleep(20)
response = client.execute(db_name, "{} | count".format(table_name))
for row in response.primary_results[0]:
assert int(row["Count"]) == 28, "{0} | count = {1}".format(
table_name, text_type(row["Count"]))
def test_ingest_complicated_props():
# Test ingest with complicated ingestion properties
validation_policy = ValidationPolicy(
validationOptions=ValidationOptions.ValidateCsvInputConstantColumns,
validationImplications=ValidationImplications.Fail,
)
json_ingestion_props = IngestionProperties(
"PythonTest",
"Deft",
dataFormat=DataFormat.json,
mapping=Helpers.create_deft_table_json_mappings(),
additionalTags=["a", "b"],
ingestIfNotExists=["aaaa", "bbbb"],
ingestByTags=["ingestByTag"],
dropByTags=["drop", "drop-by"],
flushImmediately=False,
reportLevel=ReportLevel.FailuresAndSuccesses,
reportMethod=ReportMethod.Queue,
validationPolicy=validation_policy,
)
for f in [json_file_path, zipped_json_file_path]:
ingest_client.ingest_from_file(f, json_ingestion_props)
successes = 0
timeout = 60
while successes != 2 and timeout > 0:
while ingest_status_q.success.is_empty() and timeout > 0:
time.sleep(1)
timeout -= 1
success_message = ingest_status_q.success.pop()
assert success_message[0].Database == "PythonTest"
assert success_message[0].Table == "Deft"
successes += 1
assert successes == 2
# TODO: status queues only mark ingestion was successful, but takes time for data to become available
time.sleep(20)
response = client.execute("PythonTest", "Deft | count")
for row in response.primary_results[0]:
assert int(row["Count"]) == 28, "Deft | count = " + str(row["Count"])
def test_blob_json_mapping_reference(self):
"""Tests serialization of ingestion blob info with json mapping reference."""
validation_policy = ValidationPolicy(
ValidationOptions.ValidateCsvInputConstantColumns,
ValidationImplications.BestEffort)
properties = IngestionProperties(
database="database",
table="table",
dataFormat=DataFormat.JSON,
mappingReference="jsonMappingReference",
additionalTags=["tag"],
ingestIfNotExists=["ingestIfNotExistTags"],
ingestByTags=["ingestByTags"],
dropByTags=["dropByTags"],
flushImmediately=True,
reportLevel=ReportLevel.DoNotReport,
reportMethod=ReportMethod.Queue,
validationPolicy=validation_policy,
)
blob = BlobDescriptor("somepath", 10)
blob_info = _IngestionBlobInfo(blob,
properties,
auth_context="authorizationContextText")
self._verify_ingestion_blob_info_result(blob_info.to_json())
"input", "dataset.json")
ZIPPED_JSON_FILE_PATH = os.path.join(os.getcwd(), "azure-kusto-ingest",
"tests", "input", "dataset.jsonz.gz")
KUSTO_INGEST_CLIENT.ingest_from_multiple_files(
[JSON_FILE_PATH, ZIPPED_JSON_FILE_PATH], False, JSON_INGESTION_PROPERTIES)
time.sleep(60)
RESPONSE = KUSTO_CLIENT.execute("PythonTest", "Deft | count")
for row in RESPONSE.iter_all():
if int(row['Count']) == 24:
print("Completed ingest from json mapping successfully.")
else:
print("Deft | count = " + str(row['Count']))
# Test ingest with complicated ingestion properties
VALIDATION_POLICY = ValidationPolicy(
validationOptions=ValidationOptions.ValidateCsvInputConstantColumns,
validationImplications=ValidationImplications.Fail)
JSON_INGESTION_PROPERTIES = IngestionProperties(
"PythonTest",
"Deft",
dataFormat=DataFormat.json,
mapping=Helpers.create_deft_table_json_mappings(),
additionalTags=["a", "b"],
ingestIfNotExists=["aaaa", "bbbb"],
ingestByTags=["ingestByTag"],
dropByTags=["drop", "drop-by"],
flushImmediately=False,
reportLevel=ReportLevel.FailuresAndSuccesses,
reportMethod=ReportMethod.QueueAndTable,
validationPolicy=VALIDATION_POLICY)
KUSTO_INGEST_CLIENT.ingest_from_multiple_files(