BigQuery partition by in create table. #31
Conversation
| @@ -4083,6 +4083,13 @@ impl<'a> Parser<'a> { | |||
| None | |||
| }; | |||
|
|
|||
| // In BigQuery PARITION BY can be also defined after columns | |||
| let partitioned_by = if self.parse_keywords(&[Keyword::PARTITION, Keyword::BY]) { | |||
| Some(self.parse_expr()?) | |||
There was a problem hiding this comment.
Went through code base and parsing PARTITION BY or PARTITION in parts like window specification are always parsing just expr. Keeping the same parsing.
Also not guarding it by dialect as whole parse_create_table is dialect independent.
tests/sqlparser_common.rs
Outdated
| let res = Parser::new(&dialect) | ||
| .try_with_sql(&sql) | ||
| .expect("tokenize to work") | ||
| .with_recursion_limit(38) |
There was a problem hiding this comment.
The stack frame size increased in debug builds as this PR adds new local variables.
I fixed the test by decreasing recursion limit (39 is still raising an stack overflow error). @lustefaniak I could decrease it here as default value for logic https://github.com/getsynq/sqlparser-rs/blob/main/src/parser/mod.rs#L200, but not sure about it.
What is your advice?
There was a problem hiding this comment.
We need to decrease it in the default, as that is the one used by the gRPC service, if the code is going to fail otherwise we need to make sure it works and cleanly reports failure instead of crashing the app.
lustefaniak
deleted the
zdenko/syn-3024-bigquery-parse-errro-partition-by
branch
Why
BigQuery create table statement is supporting
PARTITION BY <expr>after column difinition. This PR is allowing us to parse these statements.BigQuery doc