Fix test_decoding schema parsing with quoted identifier #852
Conversation
|
The problem with quoted identifier is that our parser is not smart enough about identifiers that contain either dots or columns characters in their name, which now could happen. We need to implement an actual parser for a quoted identifier that knows how to walk through it and skip the right set of characters etc. I think it would be best to include that work in the current PR. |
@dimitri Should we build an equivalent parser like Postgres parse_ident? or may be we can assume we will a valid identifier here and implement some dumb parser which finds first dot after all escaping? |
We should assume a valid identifier here, for sure. I think all we need is to be able to navigate through the chars knowing if we are still within the quoted identifier text, reading the delimiter quote, or reading something that's not in the quoting (the dot, for instance, is never quoted). |
0d7729c to
9090857
Compare
|
@dimitri I attempted to parse the identifier with the assumption that it would be well-formed. |
9090857 to
995ea1b
Compare
bbd0da7 to
819bf3e
Compare
The current parser fails when the DML message table schema is escaped with quotes. Remove the quote check, as the existing logic already covers schemas with quotes. Signed-off-by: Arunprasad Rajkumar <ar.arunprasad@gmail.com>
819bf3e to
70badaa
Compare
The current parser fails when the DML message table schema is escaped with quotes.
Remove the quote check, as the existing logic already parses schemas with quotes.