Validation and conversion of data.

Platforms & distributions

  • Python package
  • platform-independent source code


Development status

Stable, handed off.

Latest version

See below.


This package has been taken over by TODO, who has substantially re-engineered it and chnaged some of the API. This page is maintained for largely historical purposes.


The problem of sanitizing data (checking correctness and transforming to a useful form) is widespread throughout programming:

  • How do I verify user input is correct?
  • How do I munge data from a spreadsheet into dates and numbers?
  • How do I convert raw database fields into a programmatic object?

Ian Bicking came up with a sensible idiom for this problem, embodied in his Formencode library: validation and conversion are one and the same thing, and can be handled by passing raw data through a chain of validators. Each validator checks and/or transforms the data and passes it on to the next.

In this spirit, konval is a package that provides:

  • a rich library of validation objects
  • base classes for easily producing custom validators
  • functions for easily using validators in a variety of ways

konval is in an exploratory state, having been produced to support another package and see if use can be got out of generalising conversion. As such. it is still an early release and the API may change. Comment is invited.


This package can be installed by the usual Pythonic methods:

  1. use your favourite installation tool:

    % easy_install rst2beamer
  2. or download the source, unpack it, change into the directory and call:

    % python install

Using konval

A full API is included in the source distribution.

Most commonly, konval will be used to check or clean values. Failures result in exceptions being thrown:

# convert user input to a actual integer
>>> from konval import *
>>> sanitize ('1.0', ToInt())
>>> sanitize ('one', ToInt())
Traceback (most recent call last)
ValueError: can't convert '1.0' to integer

A single validator or list can be passed to sanitize. Failure in any will result in any exception:

# check a list has no more than 3 members
>>> sanitize (['a', 'b', 'c'], [ToLength(), IsEqualOrLess(3)])
# check a password is long enough
>>> sanitize ('mypass', [ToLength(),IsEqualOrMore(8)])
Traceback (most recent call last)
ValueError: 6 is lower than 8

Any callable object that accepts and returns a single value can be used as a validator:

>>> from string import *
>>> sanitize (" my title ", [strip, capitalize])
'My title'

A rich library of prebuilt validators is supplied:

>>> sanitize ('abcde', IsNonblank()) 'abcde'
>>> sanitize (5, IsInRange(1,6)) 5
>>> sanitize ('foo', Synonyms({'foo': 'bar', 'baz': 'quux'}))

Custom validators can easily be subclassed from a supplied base class:

class IsFoo (BaseValidator):
        def validate_value (self, value):
                if value != 'foo':
                        self.raise_validation_error (value)
                return True


konval is aimed at a one-way transformation of data, turning user input or stored data into Python objects. Certainly it could be used in the reverse direction, but this is not a primary use case. FormEncode is based around two-way (round trip) conversion of data, so that may be a useful alternative.

The name konval was chosen because:

  1. there's already a Python library called "sanity"
  2. out of "valcon", "valkon", "conval" etc. it was the one with the fewest hits on Google