Identifier case transform #175
Conversation
| val shouldQuote = database.alwaysQuoteIdentifiers | ||
| || !this.isIdentifier | ||
| || this.toUpperCase() in database.keywords | ||
| || this.isMixedCase && !database.supportsMixedCaseIdentifiers && database.supportsMixedCaseQuotedIdentifiers |
There was a problem hiding this comment.
@lyndsysimon We will quote an identifier when it meet any of these conditions:
- The
alwaysQuoteIdentifiersparam is set totrue. - The identifier is not valid.
- The identifier is a SQL keyword.
- The identifier is mixed case and the database doesn't support mixed case for unquoted identifiers but supports quoted ones.
There was a problem hiding this comment.
My apologies for being late to see this - it looks great!
My only thought is that there are a lot of flags on Database, and that they can be set in conflicting ways. My instinct would be to use an Enum instead of a series of flags - or more specifically, in this case, two Enums - one for how the DB stores mixed case identifiers and another for how the DB handles unquoted identifiers.
That said, given that a developer should not typically need to implement their own database subclasses I think this implementation is fine the way it is.
|
|
||
| /** | ||
| * Whether this database threats mixed case unquoted SQL identifiers as case sensitive and as a result | ||
| * stores them in mixed case. |
There was a problem hiding this comment.
Typo: s/threats/treats
@lyndsysimon Can you help review the code?
I just found that issue #166 is a very common case that also exists in other databases and I came up with a solution to handle it. I think this might be a better way.
I use
DatabaseMetaDatato detect whether the current database supports mixed case for quoted or unquoted SQL identifiers:If mixed case is not supported for unquoted identifiers but supported for quoted ones (e.g. PostgreSQL), I will quote the mixed case identifiers to fix #166
I also add some additional logic for the case that mixed case is not supported (no matter quoted or not). In this case, I will auto transform the idenfiers to the correct case that the database prefers. The perfered case can be known by these methods: