Configuration option to allow override/hook to identifier wrapping code

[elhigu]

First part of #2084 and #929 feature, which allows to write custom identifier wrapper methods.

So effectively you can describe function how each identifier name are treated, like for example id replacement in knex.raw('??', ['MyTable.myCol']) will behave. This is pretty nice feature for ORMs which would like to allow for example different naming scheme for database tables/columns and on javascript side.

Documentation PR is here knex/documentation#51

[stevenvachon]

By "first part", are you referring to the transformation of identifiers with no concept of direction? The code appears to always convert to snake_case, instead of conditionally to that or camelCase.

[elhigu]

@stevenvachon by the "first part" I mean converting identifiers and with "second part" I mean converting column names of the result rows read from database. Did you check the documentation pull request, which explains what the method does? If you have suggestions how to improve that all the comments are appreciated.

Also there is no hard coded snake_case <-> camelCase conversion in this PR. Just a tool for making implementation for that easier. Converting column names of result rows read from database is completely different feature, but together those two features can be used to create the complete snake_case <-> camelCase conversion.

[stevenvachon]

The documentation is in a script file which doesn't make for the clearest to comprehension. As for the snake_case -> camelCase, I was referring to the test file (which instead concatenates).

[elhigu]

As for the snake_case -> camelCase, I was referring to the test file (which instead concatenates).

Yes. That is very much intentional. It just tests that modifying and calling original wrapped callback actually works.

[elhigu]

@wubzz @rhys-vdw @tgriesser @koistya any comments on this one?

[wubzz]

Nice simple solution to something that has been requested for a long time. I have no objections!

[elhigu]

I'll still check if I'm able to add info, which part of the identifier is being transformed. Somewhat the same way that @koistya mentioned in issue #2084.

EDIT: I'm pretty sure that there is no robust way to this. For example with knex.raw('??', ['name']) there is no way to tell if the name is table name, column name, schema name or an alias...

[sdwrage]

Is there an estimated timeline for when a pull request will be made to the main knex repo for the last half of this feature? In other words, when will the full implementation be available for use?

[elhigu]

@sdwrage do you mean when this one is released or if there is going to be also hook for modifying column names read from database? Anyway, both features will be released in next release knex 0.14.0 hopefully next month. There are some other PRs still which I would like to get merged before it.

[sdwrage]

@elhigu both and thank you for the answer :)