File validation rules

Question

I have an application that can accept CSV files to run some operations. The files look like:

CREATE USER:username,last_name,first_name,age
user1,Smith,John,23
user2,Poppins,Mary,257

There are a number of commands that can be used, each with a specific "syntax" in terms of what columns are required / authorised, the type of data in each column, valid values as well as more complex rules (e.g. 2 users can't have the same username, the username must not already be present in the database, etc.).

At the moment, each of those commands correspond to a specific class in my OO model which runs the validation with code (if (type(column) != STRING) blabla) before executing the commands.

This is becoming cumbersome and I would like to extract the validation rules in human readable format, say:

Command=CREATE USER
Field='username', required, text, max_length=255, min_length=1,
      unique, not exists in db table 'User'
Field='age', optional, number, min=0, max=120
....

My code could then become:

List<Error> errors = validate(csvFile, rulesFile);
if (errors.isEmpty()) //good to go

However the more I think about it, the more it seems like it's going to be a major headache to code and effectively a whole project in itself, possibly not worth the time spent.

What approach could I take to extract rules in human readable format without unduly complicating the validation code?

Answer 1

Smart data structures and dumb code works a lot better than the other way around. -- Eric S. Raymond, The Cathedral and the Bazaar

You can use XML or JSON with schemas. Validation rules should be machine-readable first, and human-readable if possible. Consider not inventing your own format and using XML Schema or JSON Schema instead. Switching to XML/JSON should not be very difficult, since there are a lot of libs available.

What you have now is basically some form of RPC based on CSV format. Maybe you should look into JSON-RPC or something similar.