create if not exists for all dialect #163
Conversation
There was a problem hiding this comment.
It's not in the standard, but at least PostgreSQL has it.
Thanks for the PR! Please see the comments inline. And please run cargo fmt and rebase on master, otherwise the CI fails.
src/parser.rs
Outdated
| @@ -932,6 +933,37 @@ impl Parser { | |||
| } | |||
|
|
|||
| pub fn parse_create_table(&mut self) -> Result<Statement, ParserError> { | |||
| let if_keyword = self.parse_keyword("IF"); | |||
There was a problem hiding this comment.
I think you can simply use let if_not_exists = self.parse_keywords(vec!["IF", "NOT", "EXISTS"]);, why not?
There was a problem hiding this comment.
If I use let if_not_exists = self.parse_keywords(vec!["IF", "NOT", "EXISTS"]); I wouldn't be able to say if user typed in just CREATE TABLE or missed one of keyword because self.parse_keywords returns bool
There was a problem hiding this comment.
Not sure I fully understand what you're saying.. When parse_keywords returns false the current token is left unchanged. So CREATE TABLE if (...) would be accepted (which seems more appropriate for the generic parser at least; I don't know if Postgres rejects that), and missing some of the keywords like CREATE TABLE IF NOT ... would fail with an error like "Expected end of statement, found: NOT" instead of more specific error message, which seems acceptable too.
Listing all possible errors for multiple dialects doesn't seem maintainable, and we already use parse_keywords in other places.
alex-dukhno
force-pushed
the
create-table-if-not-exists
branch
from
eb59356 to
112234d
Compare
|
@nickolay thank you for review, I've rebased over master and moved tests to only PostgreSQL dialect module, but I have general question how do I make sure that |
src/parser.rs
Outdated
| @@ -932,6 +933,37 @@ impl Parser { | |||
| } | |||
|
|
|||
| pub fn parse_create_table(&mut self) -> Result<Statement, ParserError> { | |||
| let if_keyword = self.parse_keyword("IF"); | |||
There was a problem hiding this comment.
Not sure I fully understand what you're saying.. When parse_keywords returns false the current token is left unchanged. So CREATE TABLE if (...) would be accepted (which seems more appropriate for the generic parser at least; I don't know if Postgres rejects that), and missing some of the keywords like CREATE TABLE IF NOT ... would fail with an error like "Expected end of statement, found: NOT" instead of more specific error message, which seems acceptable too.
Listing all possible errors for multiple dialects doesn't seem maintainable, and we already use parse_keywords in other places.
Note that CI still thinks you didn't run rustfmt before comitting: https://travis-ci.org/github/andygrove/sqlparser-rs/builds/677142460#L641 - have you? |
|
Thank you! |
This PR attempts to implement
CREATE TABLE IF NOT EXISTS table_namefor all dialects.However, I am not sure that it is a standard SQL statement - please correct me if I am wrong.