def push_incremental_data_to_es():
params_collection = mongo_helpers.get_mongo_db_con(
database=_MONGO_PARAMS_DB)[mongo_conf[_MONGO_PARAMS_COLLECTION]]
timestamp_range = obtain_time_ranges(params_collection)
LOGGER.debug(
"Started river to push data to ES for {0}".format(timestamp_range))
mongo_object_ids_range = mongo_helpers.get_server_object_ids(
timestamp_range=timestamp_range)
raw_data_collection = mongo_helpers.get_mongo_db_con(
database=_MONGO_RAW_DATA_DB)[mongo_conf[_MONGO_RAW_DATA_COLLECTION]]
query_filters = {'_id': {'$gte': mongo_object_ids_range[0], '$lte': mongo_object_ids_range[
1]}, 'category': {'$in': _categories}}
mongo_fetch_generator = mongo_helpers.create_mongo_fetch_generator(raw_data_collection,query_filters)
process_pipeline(raw_data_collection,mongo_fetch_generator,query_filters)
params_collection.update({'elasticsearch.lastUpdated': {'$exists': 'true'}}, {
'$set': {'elasticsearch.lastUpdated': str(timestamp_range[1])}})
def push_data_to_elasticsearch():
for collection in _collections:
for date_range in mongo_helpers.get_date_range_list(_NUMBER_OF_DAYS):
print "Started pushing events for {collection} of {date_range}".format(collection=collection, date_range=date_range)
raw_data_collection = mongo_helpers.get_mongo_db_con(
database=_MONGO_RAW_DATA_DB)[collection]
mongo_object_ids_range = mongo_helpers.get_server_object_ids(
date_range)
query_filters = {'_id': {'$gte': mongo_object_ids_range[0], '$lte': mongo_object_ids_range[
1]}}
mongo_fetch_generator = mongo_helpers.create_mongo_fetch_generator(
raw_data_collection, query_filters)
process_pipeline(
raw_data_collection, mongo_fetch_generator, query_filters)