This can be useful in case your input JSON is complex, while you need
only few fields in it.
This behaviour is controlled by the
input_format_json_ignore_unknown_keys_in_named_tuple setting name, that
is turned OFF by default.
This will, almost, allow to parse gharchive dataset without jq. "almost"
because of two things:
- Tuple cannot be Nullable, so such keys with Tuple type in ClickHouse
cannot be `null` in JSON
- You cannot use dot.dot notation to extract columns for file() engine,
only tupleElement()
Signed-off-by: Azat Khuzhin <a.khuzhin@semrush.com>
