[Dbix-class] best way to cleanse / validate data

[Seth Daniel]

On Mon, Jan 17, 2011 at 09:19:13AM -0200, Eden Cardim wrote:
> >>>>> "Seth" == Seth Daniel <dbix-class at sethdaniel.org> writes:
> 
>     Seth> Hi, I was wondering what is the preferred way to cleanse and
>     Seth> validate incoming data when using DBIx::Class?
[...snip...]
> 
> Personally, I've found that a generic validation solution is only
> adequate for the most trivial cases, like constraining an int to a
> certain range of values or checking for email well-formedness. For
> anything more complex than that, like checking for uniqueness or
> building a list of the possible values that are allowed for a certain
> field, I'd say you're better off writing your own logic, YMMV.

I was mostly thinking of trivial cases.  i.e. a simple check to verify
the column meets a certain pattern or is a certain type.  Curiously, why
would I need to check for uniqueness?  Every database I plan on using
does this for me (which isn't to say that every database DBIx::Class
supports does this).


>     Seth> So beyond data validation I'd like any opinion on the *proper* way to
>     Seth> cleanse data before it gets placed in the database.  Mostly I'm just
>     Seth> looking for a way to make certain varchar columns are all lower case.
>     Seth> I'm also looking for a good way to validate the incoming data for
>     Seth> certain columns.  Again, I was hoping using MooseColumns would allow me
>     Seth> to use Moose's type system, but since the accessors don't get called for
>     Seth> some common DBIx::Class methods the usefulness is less than I had hoped.
> 
> How about wrapping the relevant API calls with Params::Validate?

It seemed to me the lowest level I could do this type of checking was
the accessor.  This turned out to be incorrect.  Certainly, doing the
checking at a higher level will work, but seems like a lot more effort
since I'd have to check the same columns over and over in different API
calls.

-- 
seth /\ sethdaniel.org