CSV file processing in MVC

Question

I am using a MVC framework and I have developed a module where .csv file is being imported into the database, I am little unsure about certain things which I'll try to explain below:

Step 1:

.csv file is uploaded and I do all the initial checks like valid .csv file, size etc

Step 2:

Next I go through the main headers of the .csv file and check if there are any new fields added to the file and if there are then I create an array of those fields. Before checking I already get list of existing fields from the db as an array. Then I create a string of all table field names, after properly sanitizing and escaping them using php

Step 3:

I go through the .csv file line by line, sanitize & escape values using php and create a string that is used as VALUES for my insert query

Step 4:

I pass the string of fields and values to the model for query execution.

Now thing is I am executing Steps 1 to 3 in controller and Step 4 in model, so question is should I sanitize & escape fields n values in controller or should I do it in the model? Which is technically & logically much better way of doing things in the MVC framework.

Answer 1

What you put in a controller and what finds its way in the model depends on the language and framework you use. Some prefer small models which consist of data structures and practically no logic; others prefer using a controller just as a glue, and moving as much logic as possible to the models. Given the code review you got, it seems that the habit in your company/ecosystem is to put data validation in the model.

This, by the way, makes perfect sense. Imagine a model Price which contains the price of a product. Would it be better to put the validation logic such as “the price amount could not be inferior or equal to zero” in the actual Price class, or do it in a controller? In OOP, you'd usually put this logic in the Price class.

Discussing validation specifically, for instance ASP.NET MVC where usually most of the code can be found in the controller, uses a special mechanism where input validation (and more specifically the validation of forms) is done through the attributes which decorate the members of the model. The controller is not involved.

but I don't understand why would it cause any error

It won't. The problems with your approach are only that:

• It makes difficult to follow the logic: the data belongs to a given class, but the validation of this data is made in a different class, which doesn't even have this responsibility in the first place.

• The model can be used by multiple controllers, leading to the duplication of the validation logic.

Probably both cases don't apply well or at all in your current case; however, they apply for the remaining part of the code base, and keeping your code consistent is important.

Answer 2

I prefer controllers and models (nowadays mostly Entities that are mistakenly referred to as Models) to be 'slim'. Do garbage collection and other gibberish elsewhere, but not there.

So you have to let the controller delegate the work:

(Controller object whispering) "I could do it, but that's too much trouble. I'll let my minion do that hard work and I'll just sit back, relax and wait for him to finish. When he's finished, I have another minion waiting to do finish the hard work!"

So I recommend to introduce a 'Helper' that would be called from the controller.

What it would basically do is:

• build the basic output and sanitize it in the process
• return the result

The result is a collection of values that can be parsed by your model (this is an important feature of the helper!).

I theory you'd now have two steps:

• The controller sends and receives data from and to the helper
• The controller sends data to the model and stores it

Makes it a lot cleaner and your code is more readable!

When dealing with really large heaps of data, this approach may not always work. You may have to integrate step 2 in step 1 (you helper acts without controller intervention an stores data autonomously).