Steven Talbot created CALCITE-5222:
--------------------------------------
Summary: Support more EXTRACT field values (or allow arbitrary in
parse?)
Key: CALCITE-5222
URL: https://issues.apache.org/jira/browse/CALCITE-5222
Project: Calcite
Issue Type: Improvement
Reporter: Steven Talbot
E.g. a number of BigQuery EXTRACTs from
[https://cloud.google.com/bigquery/docs/reference/standard-sql/timestamp_functions#extract]
do not parse with the babel parser
{noformat}
EXTRACT(DAYOFWEEK FROM <date_col>)
EXTRACT(TIME FROM <date_col>){noformat}
(also DATE, DATETIME, DAYOFYEAR... there may be others)
Other dialects may have similar problems, e.g. "EXTRACT(JULIAN...)" in Postgres
looks like it wouldn't parse either.
Calcite chasing down every value ever implemented by a DB vendor here seems
like it might be too tall of a task, but perhaps allowing for an arbitrary
keyword in that syntactic position, and then users of the parser would have to
handle it if the parse of the EXTRACT comes out with an arbitrary symbol,
rather than a TimeUnit enum value for one of the known EXTRACT fields.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
