-
Notifications
You must be signed in to change notification settings - Fork 550
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ClickHouse: support alter table update ...
#1476
base: main
Are you sure you want to change the base?
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM, cc @iffyio
src/parser/mod.rs
Outdated
let target = self.parse_assignment_target()?; | ||
self.expect_token(&Token::Eq)?; | ||
let value = self.parse_subexpr(self.dialect.prec_value(Precedence::Between))?; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh was it something precluding us from calling self.parse_assignment()
to do this bit?
Also can we use self.parse_comma_separated()
in place of the loop
?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yep, self.parse_assignment()
fails to parse SQLs like ... UPDATE c = 1 IN PARTITION ...
due to it's not look ahead enough before parsing expressions. It's the same reason that I can't use parse_comma_separated
as well.
Also, I have to mention that there may still have bugs when parsing something like ... UPDATE c = 1 in [1,2,3] IN PARTITION ...
, which's valid in clickhouse. The good news is it throws errors instead of paring them into wrong ASTs.
|
||
#[test] | ||
fn test_parse_alter_table_update() { | ||
let dialects = all_dialects_where(|d| d.supports_alter_table_update()); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Since the change introduces a new node in the AST, one style of test I think could be useful is to assert the returned AST that it matches what we expect. i.e a test in this style for alter_policy
We can take one scenario like ALTER TABLE t UPDATE col1 = 1, col2 = 2 IN PARTITION abc WHERE col3 = 1
and verify that the assignments
, partition
and selection
are where we expect them to be in the node
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OK, I'll add it later.
@@ -1081,7 +1081,7 @@ impl<'a> Parser<'a> { | |||
self.parse_bigquery_struct_literal() | |||
} | |||
Keyword::PRIOR if matches!(self.state, ParserState::ConnectBy) => { | |||
let expr = self.parse_subexpr(self.dialect.prec_value(Precedence::PlusMinus))?; | |||
let expr = self.parse_subexpr(self.dialect.prec_value(Precedence::Pipe))?; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh was there a reason for this change?
@@ -7156,6 +7156,35 @@ impl<'a> Parser<'a> { | |||
partition, | |||
with_name, | |||
} | |||
} else if self.dialect.supports_alter_table_update() && self.parse_keyword(Keyword::UPDATE) | |||
{ | |||
let mut assignments = vec![]; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
One minor request, could we move this new block to its own function to keep the branch of the current function smaller/more-focused?
loop { | ||
let target = self.parse_assignment_target()?; | ||
self.expect_token(&Token::Eq)?; | ||
// NOTE: Maybe it's better to save the index before parse_subexpr to do a real look |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ah so regarding the earlier comment, I think what we could do would be to make self.parse_assignment
reusable:
pub fn parse_assignment(&mut self) -> Result<Assignment, ParserError> {
self.parse_assignment_with_prec(self.dialect.prec_unknown())
}
fn parse_assignment_with_precedence(self, precedence: u8) -> Result<Assignment, ParserError> {
let target = self.parse_assignment_target()?;
self.expect_token(&Token::Eq)?;
let value = self.parse_subexpr(precedence)?;
Ok(Assignment { target, value })
}
That would make it possible to use parse_comma_separated
as well, since the issue seems to be with the IN PARTITION
precedence when parsing the assigned expr
Setting this to draft since its not waiting for review in the meantime, please feel free to undraft when ready! |
It's to support
ALTER TABLE [<database>.]<table> UPDATE <column> = <expression> WHERE <filter_expr>
.