DataFrame(serverless_client) fails with TypeError: 'Elasticsearch' object is not iterable

[pquentin]

Nick C on the Community Slack reported a bug when using DataFrame with an existing serverless Elasticsearch Python client.

from elasticsearch_serverless import Elasticsearch
import eland as ed
es = Elasticsearch(stuff)
print(es.info())  # <--- shows client connected successfully
df = ed.DataFrame(es, es_index_pattern=".ds-logs-endpoint.events*")

This fails with TypeError: 'Elasticsearch' object is not iterable because ensure_es_client tries to wrap the client in another one:

eland/eland/common.py

Lines 307 to 314 in 9cea238

	def ensure_es_client(
	es_client: Union[str, List[str], Tuple[str, ...], Elasticsearch]
	) -> Elasticsearch:
	if isinstance(es_client, tuple):
	es_client = list(es_client)
	if not isinstance(es_client, Elasticsearch):
	es_client = Elasticsearch(es_client)
	return es_client

The fix is probably to switch the isinstance check to check for str/list[str] instead of trying to check that it's not a client.

The workaround is to either pass connection data to DataFrame directly, or pass it a standard Elasticsearch client (useful to set credentials or other options).