def add_filters_item(filters, es_url, es_in_index, es_out_index):
elastic_in = ElasticSearch(es_url, es_in_index)
elastic_out = ElasticSearch(es_url, es_out_index)
# Time to just copy from in_index to our_index
total = elastic_out.bulk_upload_sync(fetch(elastic_in, filters), "uuid")
ARGS = get_params()
if ARGS.debug:
logging.basicConfig(level=logging.DEBUG,
format='%(asctime)s %(message)s')
logging.debug("Debug mode activated")
else:
logging.basicConfig(level=logging.INFO,
format='%(asctime)s %(message)s')
logging.info("Importing items from %s to %s/%s", ARGS.collection,
ARGS.elastic_url, ARGS.index)
elastic = ElasticSearch(ARGS.elastic_url, ARGS.index)
if ARGS.collection:
mongo_items = fetch_mongodb_collection(ARGS.collection,
ARGS.mongo_host,
ARGS.mongo_port)
elif ARGS.project:
mongo_items = fetch_mongodb_project(ARGS.project, ARGS.mongo_host,
ARGS.mongo_port)
elif ARGS.all_collections:
mongo_items = fetch_mongodb_all(ARGS.mongo_host, ARGS.mongo_port)
else:
raise RuntimeError('Collection to be processed not provided')
if mongo_items:
logging.info("Loading collections in Elasticsearch")
elastic.bulk_upload_sync(mongo_items, "id")
help="Get metrics for data source")
return parser.parse_args()
if __name__ == '__main__':
args = get_params()
if args.debug:
logging.basicConfig(level=logging.DEBUG,
format='[%(asctime)s] %(message)s')
logging.debug("Debug mode activated")
else:
logging.basicConfig(level=logging.INFO,
format='%(asctime)s %(message)s')
logging.getLogger("urllib3").setLevel(logging.WARNING)
logging.getLogger("requests").setLevel(logging.WARNING)
data_source = args.data_source
index = "grimoirelab_metrics"
elastic = ElasticSearch(args.elastic_url, index)
if args.elastic_metrics_url:
elastic = ElasticSearch(args.elastic_metrics_url, index)
elastic.bulk_upload_sync(fetch_metric(args.elastic_url, data_source), "id")
# for metric in fetch_metric(es_url, data_source):
# print(metric)
def export_items(elastic_url,
in_index,
out_index,
elastic_url_out=None,
search_after=False,
search_after_value=None,
limit=None,
copy=False):
""" Export items from in_index to out_index using the correct mapping """
if not limit:
limit = DEFAULT_LIMIT
if search_after_value:
search_after_value_timestamp = int(search_after_value[0])
search_after_value_uuid = search_after_value[1]
search_after_value = [
search_after_value_timestamp, search_after_value_uuid
]
logging.info("Exporting items from %s/%s to %s", elastic_url, in_index,
out_index)
count_res = requests.get('%s/%s/_count' % (elastic_url, in_index))
try:
count_res.raise_for_status()
except requests.exceptions.HTTPError:
if count_res.status_code == 404:
logging.error("The index does not exists: %s", in_index)
else:
logging.error(count_res.text)
sys.exit(1)
logging.info("Total items to copy: %i", count_res.json()['count'])
# Time to upload the items with the correct mapping
elastic_in = ElasticSearch(elastic_url, in_index)
if not copy:
# Create the correct mapping for the data sources detected from in_index
ds_mapping = find_mapping(elastic_url, in_index)
else:
logging.debug('Using the input index mapping')
ds_mapping = extract_mapping(elastic_url, in_index)
if not elastic_url_out:
elastic_out = ElasticSearch(elastic_url,
out_index,
mappings=ds_mapping)
else:
elastic_out = ElasticSearch(elastic_url_out,
out_index,
mappings=ds_mapping)
# Time to just copy from in_index to our_index
uid_field = find_uuid(elastic_url, in_index)
backend = find_perceval_backend(elastic_url, in_index)
if search_after:
total = elastic_out.bulk_upload_sync(
fetch(elastic_in, backend, limit, search_after_value,
scroll=False), uid_field)
else:
total = elastic_out.bulk_upload_sync(fetch(elastic_in, backend, limit),
uid_field)
logging.info("Total items copied: %i", total)
for dependency in dependencies:
eitem = enrich_item(dependency)
eitem['project'] = project
yield eitem
if __name__ == '__main__':
args = get_params()
if args.debug:
logging.basicConfig(level=logging.DEBUG,
format='%(asctime)s %(message)s')
logging.debug("Debug mode activated")
else:
logging.basicConfig(level=logging.INFO,
format='%(asctime)s %(message)s')
logging.info("Importing items from %s to %s/%s", args.file,
args.elastic_url, args.index)
elastic = ElasticSearch(args.elastic_url, args.index)
items = fetch_dependencies(args.file, args.project)
if items:
logging.info("Loading dependencies in Elasticsearch ...")
elastic.bulk_upload_sync(items, "uuid")
logging.info("Import completed.")
