def get_manifest_data(bucket, team, dataset, manifest_key):
""" Returns a list of items from manifests control table """
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
s3_interface = S3Interface()
local_path = s3_interface.download_object(bucket, manifest_key)
ddb_keys = []
items = []
with open(local_path, "r") as raw_file:
file_names = [
file_name.strip().split("/")[-1] for file_name in raw_file
]
for file in file_names:
ddb_keys.append({
"dataset_name": team + "-" + dataset,
"manifest_file_name": manifest_key.split("/")[-1],
"datafile_name": file
})
for ddb_key in ddb_keys:
try:
items.append(
dynamo_interface.get_item_from_manifests_control_table(
ddb_key["dataset_name"], ddb_key["manifest_file_name"],
ddb_key["datafile_name"]))
except KeyError:
logger.error("The manifest file has not been processed in Stage A")
raise Exception("Manifest File has not been processed in Stage A")
return items
def lambda_handler(event, context):
"""Checks if the file to be processed is manifest driven
Arguments:
event {dict} -- Dictionary with details on previous processing step
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with Processed Bucket and Key(s)
"""
try:
logger.info('Fetching event data from previous step')
bucket = event['body']['bucket']
keys_to_process = event['body']['keysToProcess']
team = event['body']['team']
pipeline = event['body']['pipeline']
stage = event['body']['pipeline_stage']
dataset = event['body']['dataset']
ddb_key = team + "-" + dataset
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (octagon.OctagonClient().with_run_lambda(
True).with_configuration_instance(event['body']['env']).build())
peh_id = octagon_client.start_pipeline_execution(
pipeline_name='{}-{}-stage-{}'.format(team, pipeline,
stage[-1].lower()),
dataset_name='{}-{}'.format(team, dataset),
comment=event)
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
response = dynamo_interface.get_transform_table_item(ddb_key)
logger.info("Querying DynamoDB to check for manifest details")
event["body"]["manifest_enabled"] = response["manifest_enabled"]
event["body"]["manifest_details"] = response["manifest_details"]
# Call custom transform created by user and process the file
event['body']['peh_id'] = peh_id
remove_content_tmp()
octagon_client.update_pipeline_execution(
status="{} {} Processing".format(stage, component),
component=component)
except Exception as e:
logger.error("Fatal error", exc_info=True)
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment="{} {} Error: {}".format(stage, component, repr(e)))
remove_content_tmp()
raise e
return event
def lambda_handler(event, context):
""" Checks if a dataset is driven by manifest file
Arguments:
event {dict} -- Dictionary with details on previous processing step
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with outcome of the process
"""
try:
logger.info("Fetching event data from previous step")
team = event['body']['team']
pipeline = event['body']['pipeline']
stage = event['body']['pipeline_stage']
dataset = event['body']['dataset']
peh_id = event['body']['peh_id']
env = event['body']['env']
ddb_key = team+"-"+dataset
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (
octagon.OctagonClient()
.with_run_lambda(True)
.with_configuration_instance(event['body']['env'])
.build()
)
peh.PipelineExecutionHistoryAPI(
octagon_client).retrieve_pipeline_execution(peh_id)
logger.info('Initializing DynamoDB config and Interface')
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
response = dynamo_interface.get_transform_table_item(ddb_key)
event["body"]["manifest_enabled"] = response["manifest_enabled"]
event["body"]["manifest_details"] = response["manifest_details"]
octagon_client.update_pipeline_execution(status="{} {} Processing".format(stage, component),
component=component)
except Exception as e:
logger.error("Fatal error", exc_info=True)
octagon_client.end_pipeline_execution_failed(component=component,
issue_comment="{} {} Error: {}".format(stage, component, repr(e)))
raise e
return event
def lambda_handler(event, context):
"""Updates the objects metadata catalog
Arguments:
event {dict} -- Dictionary with details on S3 event
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with Processed Bucket and Key
"""
try:
logger.info('Fetching event data from previous step')
object_metadata = json.loads(event)
stage = object_metadata['pipeline_stage']
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (octagon.OctagonClient().with_run_lambda(
True).with_configuration_instance(object_metadata['env']).build())
object_metadata['peh_id'] = octagon_client.start_pipeline_execution(
pipeline_name='{}-{}-stage-{}'.format(object_metadata['team'],
object_metadata['pipeline'],
stage[-1].lower()),
dataset_name='{}-{}'.format(object_metadata['team'],
object_metadata['dataset']),
comment=event)
# Add business metadata (e.g. object_metadata['project'] = 'xyz')
logger.info('Initializing DynamoDB config and Interface')
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
logger.info('Storing metadata to DynamoDB')
dynamo_interface.update_object_metadata_catalog(object_metadata)
logger.info(
'Passing arguments to the next function of the state machine')
octagon_client.update_pipeline_execution(
status="{} {} Processing".format(stage, component),
component=component)
except Exception as e:
logger.error("Fatal error", exc_info=True)
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment="{} {} Error: {}".format(stage, component, repr(e)))
raise e
return {'statusCode': 200, 'body': object_metadata}
def stage_transform(self, team, dataset, stage):
"""Returns relevant stage Transformation
Arguments:
team {string} -- Team owning the transformation
dataset {string} -- Dataset targeted by transformation
Returns:
class -- Transform object
"""
stage_suffix = stage[-1].lower()
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
dataset_transforms = dynamo_interface.get_transform_table_item(
'{}-{}'.format(team, dataset))['transforms'][
'stage_{}_transform'.format(stage_suffix)]
transform_info = "datalake_library.transforms.stage_{}_transforms.{}".format(
stage_suffix, dataset_transforms)
return getattr(import_module(transform_info), 'CustomTransform')
def lambda_handler(event, context):
try:
logger.info('Fetching event data from previous step')
processed_keys = event['body']['keysToProcess']
team = event['body']['team']
pipeline = event['body']['pipeline']
stage = event['body']['pipeline_stage']
dataset = event['body']['dataset']
bucket = event['body']['bucket']
logger.info('Initializing DynamoDB config and Interface')
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
wlm_ddb_table = dynamo_interface.wlm_control_table
item = dynamo_interface.get_item(wlm_ddb_table, {"name":"{}-{}-{}".format(team, dataset,processed_keys[0].split("/")[-2])})
priority = item.get('priority', None)
print(priority)
print(''.join(
[stage[:-1], chr(ord(stage[-1]))]))
logger.info('Sending messages to right priority SQS queue')
sqs_config = SQSConfiguration(team, dataset, ''.join(
[stage[:-1], chr(ord(stage[-1]))]), priority) #Workload management changes
sqs_interface = SQSInterface(sqs_config.get_stage_queue_name_wlm) #Workload management changes
sqs_interface.send_message_to_fifo_queue(json.dumps(event), '{}-{}'.format(team, dataset))
logger.info("lambda Completed")
return {
'statusCode': 200
}
except Exception as e:
raise e
def lambda_handler(event, context):
"""Updates the S3 objects metadata catalog
Arguments:
event {dict} -- Dictionary with details on Bucket and Keys
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with response
"""
try:
logger.info('Fetching event data from previous step')
bucket = event['body']['bucket']
processed_keys_path = event['body']['job']['processedKeysPath']
processed_keys = S3Interface().list_objects(bucket,
processed_keys_path)
team = event['body']['team']
pipeline = event['body']['pipeline']
stage = event['body']['pipeline_stage']
dataset = event['body']['dataset']
peh_id = event['body']['job']['peh_id']
keys_to_process = event['body']['keysToProcess']
s3_path = "post-stage/{}/manifests/{}/{}".format(
team, dataset, keys_to_process[0].split("/")[-1])
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (octagon.OctagonClient().with_run_lambda(
True).with_configuration_instance(event['body']['env']).build())
peh.PipelineExecutionHistoryAPI(
octagon_client).retrieve_pipeline_execution(peh_id)
logger.info('Initializing DynamoDB config and Interface')
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
logger.info('Storing metadata to DynamoDB')
for key in processed_keys:
object_metadata = {
'bucket': bucket,
'key': key,
'size': S3Interface().get_size(bucket, key),
'last_modified_date':
S3Interface().get_last_modified(bucket, key),
'org': event['body']['org'],
'app': event['body']['app'],
'env': event['body']['env'],
'team': team,
'pipeline': pipeline,
'dataset': dataset,
'stage': 'stage',
'pipeline_stage': stage,
'peh_id': peh_id
}
dynamo_interface.update_object_metadata_catalog(object_metadata)
logger.info("Updating manifests control table")
items = get_manifest_data(bucket, team, dataset, keys_to_process[0])
ddb_keys = get_ddb_keys(items)
for ddb_key in ddb_keys:
dynamo_interface.update_manifests_control_table_stageb(
ddb_key, "COMPLETED")
logger.info("Move manifest file to post stage")
kms_key = KMSConfiguration(team).get_kms_arn
s3_interface = S3Interface()
s3_interface.copy_object(bucket,
keys_to_process[0],
bucket,
s3_path,
kms_key=kms_key)
logger.info("Removing manifest file from pre-stage")
s3_interface.delete_objects(bucket, keys_to_process[0])
# Only uncomment if a queue for the next stage exists
# logger.info('Sending messages to next SQS queue if it exists')
# sqs_config = SQSConfiguration(team, dataset, ''.join([stage[:-1], chr(ord(stage[-1]) + 1)]))
# sqs_interface = SQSInterface(sqs_config.get_stage_queue_name)
# sqs_interface.send_batch_messages_to_fifo_queue(processed_keys, 10, '{}-{}'.format(team, dataset))
octagon_client.update_pipeline_execution(
status="{} {} Processing".format(stage, component),
component=component)
octagon_client.end_pipeline_execution_success()
except Exception as e:
logger.error("Fatal error", exc_info=True)
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment="{} {} Error: {}".format(stage, component, repr(e)))
for ddb_key in ddb_keys:
dynamo_interface.update_manifests_control_table_stageb(
ddb_key, "FAILED", None, "Failed in Post Update")
raise e
return 200
def lambda_handler(event, context):
"""Updates the S3 objects metadata catalog
Arguments:
event {dict} -- Dictionary with details on previous processing step
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with outcome of the process
"""
try:
logger.info('Fetching event data from previous step')
processed_keys = event['body']['processedKeys']
team = event['body']['team']
pipeline = event['body']['pipeline']
stage = event['body']['pipeline_stage']
dataset = event['body']['dataset']
peh_id = event['body']['peh_id']
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (octagon.OctagonClient().with_run_lambda(
True).with_configuration_instance(event['body']['env']).build())
peh.PipelineExecutionHistoryAPI(
octagon_client).retrieve_pipeline_execution(peh_id)
logger.info('Initializing DynamoDB config and Interface')
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
logger.info('Storing metadata to DynamoDB')
bucket = S3Configuration().stage_bucket
for key in processed_keys:
object_metadata = {
'bucket': bucket,
'key': key,
'size': S3Interface().get_size(bucket, key),
'last_modified_date':
S3Interface().get_last_modified(bucket, key),
'org': event['body']['org'],
'app': event['body']['app'],
'env': event['body']['env'],
'team': team,
'pipeline': pipeline,
'dataset': dataset,
'stage': 'stage',
'pipeline_stage': stage,
'peh_id': peh_id
}
dynamo_interface.update_object_metadata_catalog(object_metadata)
#Workload management changes
#---------------------------
wlm_ddb_table = dynamo_interface.wlm_control_table
item = dynamo_interface.get_item(
wlm_ddb_table, {
"name":
"{}-{}-{}".format(team, dataset,
processed_keys[0].split("/")[-2])
})
priority = item.get('priority', None)
print(priority)
#---------------------------
logger.info('Sending messages to next SQS queue if it exists')
sqs_config = SQSConfiguration(team, dataset, ''.join(
[stage[:-1], chr(ord(stage[-1]) + 1)]),
priority) #Workload management changes
sqs_interface = SQSInterface(
sqs_config.get_stage_queue_name_wlm) #Workload management changes
sqs_interface.send_batch_messages_to_fifo_queue(
processed_keys, 10, '{}-{}'.format(team, dataset))
octagon_client.update_pipeline_execution(
status="{} {} Processing".format(stage, component),
component=component)
octagon_client.end_pipeline_execution_success()
except Exception as e:
logger.error("Fatal error", exc_info=True)
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment="{} {} Error: {}".format(stage, component, repr(e)))
raise e
return 200
def lambda_handler(event, context):
"""Updates the S3 objects metadata catalog
Arguments:
event {dict} -- Dictionary with details on Bucket and Keys
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with response
"""
def replace_decimals(obj):
if isinstance(obj, list):
for i in range(len(obj)):
obj[i] = replace_decimals(obj[i])
return obj
elif isinstance(obj, dict):
for k, v in obj.items():
obj[k] = replace_decimals(v)
return obj
elif isinstance(obj, set):
return set(replace_decimals(i) for i in obj)
elif isinstance(obj, decimal.Decimal):
if obj % 1 == 0:
return int(obj)
else:
return float(obj)
else:
return obj
def get_table_partitions(db, tbl):
glue_response = glue_client.get_table(DatabaseName=db, Name=tbl)
logger.debug('Glue get_table response: {}'.format(glue_response))
return glue_response['Table']['PartitionKeys']
try:
logger.info('Fetching event data from previous step')
bucket = event['body']['bucket']
processed_keys_path = event['body']['job']['processedKeysPath']
processed_keys = S3Interface().list_objects(bucket,
processed_keys_path)
team = event['body']['team']
pipeline = event['body']['pipeline']
stage = event['body']['pipeline_stage']
dataset1 = event['body']['dataset']
peh_id = event['body']['job']['peh_id']
env = event['body']['env']
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (octagon.OctagonClient().with_run_lambda(
True).with_configuration_instance(event['body']['env']).build())
peh.PipelineExecutionHistoryAPI(
octagon_client).retrieve_pipeline_execution(peh_id)
logger.info('Initializing DynamoDB config and Interface')
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
logger.info(
'Storing metadata to DynamoDB and tagging resulting S3 Objects')
for key in processed_keys:
object_metadata = {
'bucket': bucket,
'key': key,
'size': S3Interface().get_size(bucket, key),
'last_modified_date':
S3Interface().get_last_modified(bucket, key),
'org': event['body']['org'],
'app': event['body']['app'],
'env': event['body']['env'],
'team': team,
'pipeline': pipeline,
'dataset': dataset1,
'stage': 'stage',
'pipeline_stage': stage,
'peh_id': peh_id
}
dynamo_interface.update_object_metadata_catalog(object_metadata)
tag_keys = ['org', 'app', 'env', 'team', 'dataset']
tag_dict = {key: object_metadata[key] for key in tag_keys}
S3Interface().tag_object(bucket, key, tag_dict)
# Only uncomment if a queue for the next stage exists
# logger.info('Sending messages to next SQS queue if it exists')
# sqs_config = SQSConfiguration(team, dataset, ''.join([stage[:-1], chr(ord(stage[-1]) + 1)]))
# sqs_interface = SQSInterface(sqs_config.get_stage_queue_name)
# sqs_interface.send_batch_messages_to_fifo_queue(processed_keys, 10, '{}-{}'.format(team, dataset))
prestage_table = event['body']['dest_table']['name']
prestage_db = event['body']['dest_db']
dest_part_name = event['body']['dest_table']['part_name']
dest_part_value = event['body']['dest_table']['part_value']
processOutput = {}
if dest_part_name is not '' and dest_part_value is not '':
partitions = []
part_dict = {"name": dest_part_name, "value": dest_part_value}
partitions.append(part_dict)
processOutput['partitions'] = partitions
processOutput['processed_keys'] = processed_keys
ssmresponse = ssmcli.get_parameter(
Name=f'/SDLF/DDB/{team}/{pipeline}/DependenciesByTable')
ddb_dependencies_by_table = ssmresponse['Parameter']['Value']
ddb_table = dynamodb.Table(ddb_dependencies_by_table)
ssmresponse = ssmcli.get_parameter(
Name=f'/SDLF/DDB/{team}/{pipeline}/Dependencies')
ddb_dependencies = ssmresponse['Parameter']['Value']
consulta = f'{prestage_db.lower()}.{prestage_table.lower()}'
logger.info(consulta)
response = ddb_table.get_item(Key={'table_name': consulta})
logger.info(f'Response {response}')
if 'Item' in response:
list_transforms = response['Item']['list_transforms']
num_of_transforms = len(list_transforms)
logger.debug(f'Response {response}')
logger.info(f'This table triggers {num_of_transforms} datasets')
next_stage = 'B'
stage_b_message = {}
for dataset in list_transforms:
ddb_steps = dynamodb.Table(ddb_dependencies)
logger.info(dataset)
response = ddb_steps.get_item(Key={'dataset': dataset})
logger.info(f'Response {response}')
num_of_transforms = len(list_transforms)
item = response['Item']
dest_table = item['dataset'].split('.')[1]
dest_db = item['dataset'].split('.')[0]
dependencies = item['dependencies']
date_substitutions = replace_decimals(
item.get('date_substitutions', []))
logger.info(f'Dependencies: {dependencies}')
partition = item.get('partitionColumn', '')
partition_mask = item.get('partitionPythonMask', None)
partition_value_formatted = None
table_check = []
for table in dependencies:
table_name = table['TableName'].split('.')[1]
table_db = table['TableName'].split('.')[0]
table_partition = table.get('FieldColumn', '')
table_partition_format = table.get('DateExpression', None)
relativedelta_attributes = replace_decimals(
table.get('relativedelta_attributes', None))
table_partitions = processOutput.get('partitions', [])
usage = table.get('Usage', 'validate').lower()
if usage == 'validate':
if prestage_db == table_db and prestage_table == table_name:
logger.info(
f'This table does not update/overwrite {dataset} dataset'
)
break
else:
logger.debug(
f'Table {table_db}.{table_name} is not the trigger table'
)
else:
if prestage_db.lower() == table_db.lower(
) and prestage_table.lower() == table_name.lower():
# dst_tbl_partitions = get_table_partitions(prestage_db,prestage_table)
partition_value_formatted = ''
# If dest table has partitions and source table has partitions
logger.debug(
f'Partition: {partition}, table_partitions: {table_partitions}'
)
if table_partitions and table_partition_format is not None:
table_partition_value = table_partitions[0][
'value']
value = datetime.strptime(
table_partition_value,
table_partition_format)
target_value = value - relativedelta(
**relativedelta_attributes)
partition_value_formatted = target_value.strftime(
partition_mask)
logger.info(
f'This table {usage.upper()} dataset {dest_table} '
f' Partition {partition} = {partition_value_formatted}'
)
# validate(table_db, table_name, table_partitions)
stage_b_message[
'prev_stage_processed_keys'] = processed_keys
stage_b_message['team'] = team
stage_b_message['pipeline'] = pipeline
stage_b_message['pipeline_stage'] = ''.join(
[stage[:-1], next_stage])
stage_b_message['dataset'] = dataset1
stage_b_message['org'] = event['body']['org']
stage_b_message['app'] = event['body']['app']
stage_b_message['env'] = event['body']['env']
stage_b_message['behaviour'] = table[
'Usage'].lower()
stage_b_message['dest_db'] = dest_db
stage_b_message['dest_table'] = {}
stage_b_message['dest_table']['name'] = dest_table
stage_b_message['dest_table'][
'part_name'] = partition
stage_b_message['dest_table'][
'part_value'] = partition_value_formatted
stage_b_message['steps'] = item['steps']
stage_b_message[
'date_substitutions'] = date_substitutions
logger.info(
'Sending messages to next SQS queue if it exists'
)
# GEt queue by SSM
logger.info(stage_b_message)
sqs_config = SQSConfiguration(
team, pipeline, stage)
sqs_interface = SQSInterface(
sqs_config.get_stage_queue_name)
sqs_interface.send_message_to_fifo_queue(
json.dumps(stage_b_message),
'{}-{}'.format(team, pipeline))
break
else:
logger.info(f'This table triggers 0 datasets')
octagon_client.update_pipeline_execution(
status=f'{stage} {component} Processing', component=component)
octagon_client.end_pipeline_execution_success()
except Exception as e:
logger.error("Fatal error", exc_info=True)
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment=f'{stage} {component} Error: {repr(e)}')
raise e
return 200
def transform_object(self, bucket, keys, team, dataset):
#######################################################
# We assume a Glue Job has already been created based on
# customer needs. This function makes an API call to start it
#######################################################
job_name = 'sdlf-{}-{}-glue-job'.format(
team, dataset) # Name of the Glue Job
### Create the list of s3 keys to be processed by the glue job
### keys will contain a single file for manifest processing
items = get_manifest_data(bucket, team, dataset, keys[0])
s3_keys = get_s3_keys(items)
files = []
file_names = ""
for key in s3_keys:
files.append(key.split('/')[-1])
if file_names is not None:
file_names = file_names + "|" + key
else:
file_names = key
### Update Manifests Control Table
ddb_keys = get_ddb_keys(items)
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
for ddb_key in ddb_keys:
dynamo_interface.update_manifests_control_table_stageb(
ddb_key, "PROCESSING")
# S3 Path where Glue Job outputs processed keys
# IMPORTANT: Build the output s3_path without the s3://stage-bucket/
processed_keys_path = 'post-stage/{}/{}'.format(team, dataset)
# Submitting a new Glue Job
job_response = client.start_job_run(
JobName=job_name,
Arguments={
# Specify any arguments needed based on bucket and keys (e.g. input/output S3 locations)
'--JOB_NAME':
'sdlf-{}-{}-glue-job'.format(team, dataset),
'--SOURCE_LOCATION':
's3://{}/'.format(bucket),
'--OUTPUT_LOCATION':
's3://{}/{}'.format(bucket, processed_keys_path),
'--FILE_NAMES':
file_names,
'--job-bookmark-option':
'job-bookmark-enable'
},
MaxCapacity=2.0)
# Collecting details about Glue Job after submission (e.g. jobRunId for Glue)
json_data = json.loads(
json.dumps(job_response, default=datetimeconverter))
job_details = {
"jobName": job_name,
"jobRunId": json_data.get('JobRunId'),
"jobStatus": 'STARTED',
"files": list(set(files))
}
#######################################################
# IMPORTANT
# This function must return a dictionary object with at least a reference to:
# 1) processedKeysPath (i.e. S3 path where job outputs data without the s3://stage-bucket/ prefix)
# 2) jobDetails (i.e. a Dictionary holding information about the job
# e.g. jobName and jobId for Glue or clusterId and stepId for EMR
# A jobStatus key MUST be present in jobDetails as it's used to determine the status of the job)
# Example: {processedKeysPath' = 'post-stage/engineering/legislators',
# 'jobDetails': {'jobName': 'sdlf-engineering-legislators-glue-job', 'jobId': 'jr-2ds438nfinev34', 'jobStatus': 'STARTED'}}
#######################################################
response = {
'processedKeysPath': processed_keys_path,
'jobDetails': job_details
}
return response
def get_dependent_datasets(team_name, dataset_name):
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
transform_info = dynamo_interface.get_transform_table_item("{}-{}".format(
team_name, dataset_name))
return transform_info["dependencies"]
def lambda_handler(event, context):
""" Load Datafile metadata in manifests control table
Check if manifest file is available within the threshold
Arguments:
event {dict} -- Dictionary with details on previous processing step
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with outcome of the process
"""
s3_interface = S3Interface()
stage_bucket = S3Configuration().stage_bucket
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
current_time = dt.datetime.utcnow()
current_timestamp = current_time.timestamp()
try:
logger.info("Fetching event data from previous step")
team = event['body']['team']
pipeline = event['body']['pipeline']
stage = event['body']['pipeline_stage']
dataset = event['body']['dataset']
peh_id = event['body']['peh_id']
env = event['body']['env']
bucket = event['body']['bucket']
input_file_key = event['body']['key']
input_file_name = input_file_key.split("/")[-1]
manifest_file_pattern = event['body']['manifest_details'][
'regex_pattern']
manifest_timeout = int(
event['body']['manifest_details']['manifest_timeout'])
if 'manifest_interval' in event['body']:
manifest_interval = event['body']['manifest_interval']
else:
manifest_interval = current_timestamp
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (octagon.OctagonClient().with_run_lambda(
True).with_configuration_instance(env).build())
peh.PipelineExecutionHistoryAPI(
octagon_client).retrieve_pipeline_execution(peh_id)
octagon_client.update_pipeline_execution(
status="{} {} Processing".format(stage, component),
component=component)
### List S3 Objects for the manifest file in the manifest prefix
### For this to work the manifest should have been loaded into DynamoDB
manifest_key = "pre-stage/{}/manifests/{}/".format(team, dataset)
processed_manifest_keys = s3_interface.list_objects(
stage_bucket, manifest_key)
matched_keys = []
items = []
if not processed_manifest_keys:
logger.info(
"Manifest File has not been loaded, sleeping for 5 mins")
time.sleep(300)
manifest_file_loaded = "False"
else:
for manifest_file_key in processed_manifest_keys:
manifest_file_name = manifest_file_key.split("/")[-1]
match = re.match(manifest_file_pattern, manifest_file_name)
if match:
matched_keys.append(manifest_file_name)
### Query Manifests Control table
for keys in matched_keys:
dataset_name = team + "-" + dataset
try:
items.append(
dynamo_interface.
get_item_from_manifests_control_table(
dataset_name, keys, input_file_name))
except KeyError:
logger.info(
"Manifest File has not been loaded, sleeping for 5 mins"
)
manifest_file_loaded = "False"
### Update Manifests Control table
if not items:
logger.info(
"Manifest File has not been loaded, sleeping for 5 mins"
)
time.sleep(300)
manifest_file_loaded = "False"
else:
ddb_key = {
'dataset_name': items[0]['dataset_name'],
'datafile_name': items[0]['datafile_name']
}
STATUS = "STARTED"
dynamo_interface.update_manifests_control_table_stagea(
ddb_key, STATUS)
manifest_file_loaded = "True"
event['body']['manifest_ddb_key'] = ddb_key
### Check if Manifest threshold has exceeded
if current_timestamp == manifest_interval:
current_timestamp = dt.datetime.utcnow().timestamp()
if int(
(current_timestamp - manifest_interval) / 60) >= manifest_timeout:
logger.error("Manifest Threshold Breached")
raise Exception("Manifest Threshold Breached")
event['body']['manifest_interval'] = manifest_interval
event['body']['manifest_file_loaded'] = manifest_file_loaded
except Exception as e:
logger.error("Fatal error", exc_info=True)
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment="{} {} Error: {}".format(stage, component, repr(e)))
raise e
return event
def lambda_handler(event, context):
""" Process the manifest file and loads into DynamoDB
Arguments:
event {dict} -- Dictionary with details on previous processing step
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with outcome of the process
"""
s3_interface = S3Interface()
stage_bucket = S3Configuration().stage_bucket
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
try:
logger.info("Fetching event data from previous step")
team = event['body']['team']
pipeline = event['body']['pipeline']
stage = event['body']['pipeline_stage']
dataset = event['body']['dataset']
peh_id = event['body']['peh_id']
env = event['body']['env']
bucket = event['body']['bucket']
manifest_file_key = event['body']['key']
manifest_file_name = manifest_file_key.split("/")[-1]
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (octagon.OctagonClient().with_run_lambda(
True).with_configuration_instance(env).build())
peh.PipelineExecutionHistoryAPI(
octagon_client).retrieve_pipeline_execution(peh_id)
### Download the manifest file to local
local_path = s3_interface.download_object(bucket, manifest_file_key)
### Process the manifest file
with open(local_path, "r") as raw_file:
file_names = [
file_name.strip().split("/")[-1] for file_name in raw_file
]
### Load data into manifests control table
for file in file_names:
item = {
"dataset_name":
team + "-" + dataset + "-" + manifest_file_name,
"datafile_name": manifest_file_name + "-" + file
}
dynamo_interface.put_item_in_manifests_control_table(item)
### Set s3 path for Copy
s3_path = 'pre-stage/{}/manifests/{}/{}'.format(
team, dataset, manifest_file_name)
kms_key = KMSConfiguration(team).get_kms_arn
### Copy Manifest File to team/manifest/dataset location
s3_interface.copy_object(bucket,
manifest_file_key,
stage_bucket,
s3_path,
kms_key=kms_key)
octagon_client.update_pipeline_execution(
status="{} {} Processing".format(stage, component),
component=component)
processed_keys = [s3_path]
except Exception as e:
logger.error("Fatal error", exc_info=True)
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment="{} {} Error: {}".format(stage, component, repr(e)))
raise e
return processed_keys
def lambda_handler(event, context):
"""Checks if the file to be processed is manifest driven
Arguments:
event {dict} -- Dictionary with details on previous processing step
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with Processed Bucket and Key(s)
"""
try:
logger.info('Fetching event data from previous step')
bucket = event['body']['bucket']
keys_to_process = event['body']['keysToProcess']
team = event['body']['team']
pipeline = event['body']['pipeline']
stage = event['body']['pipeline_stage']
dataset = event['body']['dataset']
peh_id = event['body']['peh_id']
manifest_data_timeout = int(
event['body']['manifest_details']['manifest_data_timeout'])
current_time = dt.datetime.utcnow()
current_timestamp = current_time.timestamp()
if 'manifest_interval' in event['body']:
manifest_interval = event['body']['manifest_interval']
else:
manifest_interval = current_timestamp
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (octagon.OctagonClient().with_run_lambda(
True).with_configuration_instance(event['body']['env']).build())
peh.PipelineExecutionHistoryAPI(
octagon_client).retrieve_pipeline_execution(peh_id)
### Set max_items_process in datasets table so that the statemachine only processes 1 manifest file at a time
ddb_keys = get_ddb_keys(keys_to_process, bucket, team, dataset)
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
### Query Manifest Control Table to get the status
items = []
logger.info(
"Querying DynamoDB to check data in manifests control table for Stage A status"
)
for ddb_key in ddb_keys:
try:
items.append(
dynamo_interface.get_item_from_manifests_control_table(
ddb_key["dataset_name"], ddb_key["manifest_file_name"],
ddb_key["datafile_name"]))
except KeyError:
logger.error(
"The manifest file has not been processed in Stage A")
raise Exception(
"Manifest File has not been processed in Stage A")
### Check stage a status for data files
logger.info(
"Checking to see if all the files have been processed in Stage A")
status_message_list = []
failed_status_message_list = []
wait_message_counter = 0
failed_message_counter = 0
for item in items:
if "stage_a_status" in item:
stage_a_status = item["stage_a_status"]
else:
stage_a_status = "NOT STARTED"
if stage_a_status != "COMPLETED" and stage_a_status != "FAILED":
status_message_list.append("Waiting for Data File {}".format(
item["datafile_name"].split("-")[-1]))
wait_message_counter += 1
elif stage_a_status == "FAILED":
failed_status_message_list.append(
"Data Files Failed in Stage A {}".format(
item["datafile_name"].split("-")[-1]))
failed_message_counter += 1
if failed_message_counter > 0:
logger.error("Data File Failure in Stage A, Processing will stop")
logger.error("The following files have failed in Stage A")
for message in failed_status_message_list:
logger.error(message)
### Update manifest control table, mark all files as failed in Stage B
for ddb_key in ddb_keys:
update_key = dynamo_interface.manifest_keys(
ddb_key["dataset_name"], ddb_key["manifest_file_name"],
ddb_key["datafile_name"])
dynamo_interface.update_manifests_control_table_stageb(
update_key, "FAILED", None, "Datafile Failed in Stage A")
raise Exception("Data File Failure in Stage A")
if wait_message_counter > 0:
logger.info("Waiting for Data Files to be processed in Stage A")
for message in status_message_list:
logger.info(message)
logger.info("Will sleep for 5 mins")
time.sleep(300)
data_file_wait = "True"
if manifest_interval == current_timestamp:
current_timestamp = dt.datetime.utcnow().timestamp()
if int((current_timestamp - manifest_interval) /
60) >= manifest_data_timeout:
logger.error("Data File Threshold Breached")
logger.error("Stage B Processing Will Stop Now")
data_file_wait = "False"
for message in status_message_list:
logger.error(message)
### Update manifest control table, mark all files as failed in Stage B
for ddb_key in ddb_keys:
update_key = dynamo_interface.manifest_keys(
ddb_key["dataset_name"], ddb_key["manifest_file_name"],
ddb_key["datafile_name"])
dynamo_interface.update_manifests_control_table_stageb(
update_key, "FAILED", None,
"Datafile threshold Breached")
raise Exception("Data File Threshold Breached")
else:
logger.info("All files processed in Stage A")
data_file_wait = "False"
for ddb_key in ddb_keys:
update_key = dynamo_interface.manifest_keys(
ddb_key["dataset_name"], ddb_key["manifest_file_name"],
ddb_key["datafile_name"])
dynamo_interface.update_manifests_control_table_stageb(
update_key, "STARTED")
event["body"]["manifest_interval"] = manifest_interval
event["body"]["data_file_wait"] = data_file_wait
remove_content_tmp()
octagon_client.update_pipeline_execution(
status="{} {} Processing".format(stage, component),
component=component)
except Exception as e:
logger.error("Fatal error", exc_info=True)
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment="{} {} Error: {}".format(stage, component, repr(e)))
remove_content_tmp()
raise e
return event
def lambda_handler(event, context):
"""Updates the S3 objects metadata catalog
Arguments:
event {dict} -- Dictionary with details on Bucket and Keys
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with response
"""
try:
logger.info('Fetching event data from previous step')
bucket = event['body']['bucket']
processed_keys_path = event['body']['job']['processedKeysPath']
processed_keys = S3Interface().list_objects(bucket,
processed_keys_path)
team = event['body']['team']
pipeline = event['body']['pipeline']
stage = event['body']['pipeline_stage']
dataset = event['body']['dataset']
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (octagon.OctagonClient().with_run_lambda(
True).with_configuration_instance(event['body']['env']).build())
peh.PipelineExecutionHistoryAPI(
octagon_client).retrieve_pipeline_execution(
event['body']['job']['peh_id'])
logger.info('Initializing DynamoDB config and Interface')
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
logger.info('Storing metadata to DynamoDB')
for key in processed_keys:
object_metadata = {
'bucket': bucket,
'key': key,
'team': team,
'pipeline': pipeline,
'dataset': dataset,
'peh_id': event['body']['job']['peh_id'],
'stage': 'post-stage'
}
dynamo_interface.update_object_metadata_catalog(object_metadata)
# Add Tables to Result Path to Enable Deequ Job
table_path = "compile_topics_data_csv"
tables = [table_path]
# Only uncomment if using Kendra and index and data source ALREADY created
# Data Sync Job
# kendra_client = boto3.client('kendra')
# response = kendra_client.start_data_source_sync_job(
# Id='ENTER_DATASOURCE_ID',
# IndexId='ENTER_INDEX_ID''
# )
# Only uncomment if a queue for the next stage exists
# logger.info('Sending messages to next SQS queue if it exists')
# sqs_config = SQSConfiguration(team, dataset, ''.join([stage[:-1], chr(ord(stage[-1]) + 1)]))
# sqs_interface = SQSInterface(sqs_config.get_stage_queue_name)
# sqs_interface.send_batch_messages_to_fifo_queue(processed_keys, 10, '{}-{}'.format(team, dataset))
octagon_client.update_pipeline_execution(
status="{} {} Processing".format(stage, component),
component=component)
octagon_client.end_pipeline_execution_success()
except Exception as e:
logger.error("Fatal error", exc_info=True)
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment="{} {} Error: {}".format(stage, component, repr(e)))
raise e
return tables
import os
import shutil
from datalake_library.commons import init_logger
from datalake_library.transforms.transform_handler import TransformHandler
from datalake_library import octagon
from datalake_library.octagon import Artifact, EventReasonEnum, peh
from datalake_library.configuration.resource_configs import DynamoConfiguration
from datalake_library.interfaces.dynamo_interface import DynamoInterface
logger = init_logger(__name__)
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
def remove_content_tmp():
# Remove contents of the Lambda /tmp folder (Not released by default)
for root, dirs, files in os.walk('/tmp'):
for f in files:
os.unlink(os.path.join(root, f))
for d in dirs:
shutil.rmtree(os.path.join(root, d))
def lambda_handler(event, context):
"""Calls custom transform developed by user
Arguments:
event {dict} -- Dictionary with details on previous processing step
context {dict} -- Dictionary with details on Lambda context
def lambda_handler(event, context):
"""Calls custom job waiter developed by user
Arguments:
event {dict} -- Dictionary with details on previous processing step
context {dict} -- Dictionary with details on Lambda context
Returns:
{dict} -- Dictionary with Processed Bucket, Key(s) and Job Details
"""
try:
logger.info('Fetching event data from previous step')
bucket = event['body']['bucket']
keys_to_process = event['body']['keysToProcess']
team = event['body']['team']
stage = event['body']['pipeline_stage']
dataset = event['body']['dataset']
job_details = event['body']['job']['jobDetails']
processed_keys_path = event['body']['job']['processedKeysPath']
logger.info('Initializing Octagon client')
component = context.function_name.split('-')[-2].title()
octagon_client = (octagon.OctagonClient().with_run_lambda(
True).with_configuration_instance(event['body']['env']).build())
logger.info('Querying manifests control table ')
items = get_manifest_data(bucket, team, dataset, keys_to_process[0])
ddb_keys = get_ddb_keys(items)
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
logger.info('Checking Job Status with user custom code')
transform_handler = TransformHandler().stage_transform(
team, dataset, stage)
response = transform_handler().check_job_status(
bucket, keys_to_process, processed_keys_path,
job_details) # custom user code called
response['peh_id'] = event['body']['job']['peh_id']
if event['body']['job']['jobDetails']['jobStatus'] == 'FAILED':
peh.PipelineExecutionHistoryAPI(
octagon_client).retrieve_pipeline_execution(response['peh_id'])
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment="{} {} Error: Check Job Logs".format(
stage, component))
for ddb_key in ddb_keys:
dynamo_interface.update_manifests_control_table_stageb(
ddb_key, "FAILED", None, "Glue Job Failed, Check Logs")
except Exception as e:
logger.error("Fatal error", exc_info=True)
peh.PipelineExecutionHistoryAPI(
octagon_client).retrieve_pipeline_execution(
event['body']['job']['peh_id'])
octagon_client.end_pipeline_execution_failed(
component=component,
issue_comment="{} {} Error: {}".format(stage, component, repr(e)))
for ddb_key in ddb_keys:
dynamo_interface.update_manifests_control_table_stageb(
ddb_key, "FAILED", None, "Glue Job Failed, Check Logs")
raise e
return response
def lambda_handler(event, context):
"""Checks if any items need processing and triggers state machine
Arguments:
event {dict} -- Dictionary with no relevant details
context {dict} -- Dictionary with details on Lambda context
"""
# TODO Implement Redrive Logic (through message_group_id)
try:
team = event['team']
pipeline = event['pipeline']
stage = event['pipeline_stage']
dataset = event['dataset']
org = event['org']
app = event['app']
env = event['env']
stage_bucket = S3Configuration().stage_bucket
dynamo_config = DynamoConfiguration()
dynamo_interface = DynamoInterface(dynamo_config)
transform_info = dynamo_interface.get_transform_table_item(
'{}-{}'.format(team, dataset))
MIN_ITEMS_TO_PROCESS = int(
transform_info['min_items_process']['stage_{}'.format(
stage[-1].lower())])
MAX_ITEMS_TO_PROCESS = int(
transform_info['max_items_process']['stage_{}'.format(
stage[-1].lower())])
sqs_config = SQSConfiguration(team, dataset, stage)
queue_interface = SQSInterface(sqs_config.get_stage_queue_name)
keys_to_process = []
logger.info('Querying {}-{} objects waiting for processing'.format(
team, dataset))
keys_to_process = queue_interface.receive_min_max_messages(
MIN_ITEMS_TO_PROCESS, MAX_ITEMS_TO_PROCESS)
# If no keys to process, break
if not keys_to_process:
return
logger.info('{} Objects ready for processing'.format(
len(keys_to_process)))
keys_to_process = list(set(keys_to_process))
response = {
'statusCode': 200,
'body': {
"bucket": stage_bucket,
"keysToProcess": keys_to_process,
"team": team,
"pipeline": pipeline,
"pipeline_stage": stage,
"dataset": dataset,
"org": org,
"app": app,
"env": env
}
}
logger.info('Starting State Machine Execution')
state_config = StateMachineConfiguration(team, pipeline, stage)
StatesInterface().run_state_machine(
state_config.get_stage_state_machine_arn, response)
except Exception as e:
# If failure send to DLQ
if keys_to_process:
dlq_interface = SQSInterface(sqs_config.get_stage_dlq_name)
dlq_interface.send_message_to_fifo_queue(json.dumps(response),
'failed')
logger.error("Fatal error", exc_info=True)
raise e
return