[Dbix-class] modfiying data on retrieval

[Dave Howorth]

Ash Berlin wrote:
> Dave Howorth wrote:
>> Will Hawes wrote:
>>> On 05/03/07, Dave Howorth <dhoworth at mrc-lmb.cam.ac.uk> wrote:
>>>> I need to read data from a database where fields have been entered by
>>>> hand and have miscellaneous extraneous leading and trailing whitespace.
>>>>
>>>> I think the best approach is to suppress these characters as I retrieve
>>>> the fields from the database, so I guess I need to override some method
>>>> (in my Schema ?) but I haven't figured out which.
>>>>
>>>> I've looked around in the Cookbook and FAQ without spotting anything.
>>>> Could somebody give me a pointer to the right place?
>>>>
>>>> Thanks, Dave
>>> Wouldn't you just override the relevant accessors in your model
>>> classes to do this?
>>
>> I'm not quite sure what 'model classes' means in this context? I don't
>> think I want to override individual accessors; I think I want to
>> override whatever they inherit from (the equivalent of CDBI's get()
>> method). I just don't know where to read up about how to do that.
>>
>> Cheers, Dave
>>
> 
> 
> If you really want to do it that way - look at Inflators.

Where are they? I can't find the word inflator in the dozen or so DBIC
POD pages I've got open except at
<http://search.cpan.org/~jrobinson/DBIx-Class-0.07005/lib/DBIx/Class/Manual/Cookbook.pod#Overloading_methods>

The only place the word inflator appears in the code base is in
DBIx/Class/InflateColumn.pm. Do you mean to use inflate_column? That
also looks like it has to be done column-by-column? I'm looking for a
place to hook into the main retrieval path.

> What Will was talking about was adding something like the following:
> 
> sub my_col {
>   my ($self) = shift;
> 
>   if (@_)
>     return my_trim($self->next::my_col);
> 
>   $self->next::my_col(my_trim(@_));

I guess that should be something like

    return $self->next::my_col(my_trim(@_)) if @_;
    return my_trim($self->next::my_col);

> }
> 
> In your ResultSource classes.

But there are 31 tables with about ten columns per table. I'd rather not
write 300 or so identical methods! That's why I'm looking for a central
point.

Thanks, Dave