dataschema 0.1

Basic validation for Python data structures in a mostly declarative form (there is an escape hatch in “extraValidation” callables).

Validation errors are reported as both a path within the data structure (sequence of indices or keys) and a descriptive message (string).

Typical usage:

data = json.load(some_file) # or pickle, or ...
errors = dataschema.Validator(my_schema).validate(data)
if errors:
    for path, message in errors:
        # Report error `message` at path `path`.
else:
    # Any data access or application-specific validation can now
    # rely on properties of my_schema (e.g. minimum number of
    # elements in a sequence, data types of elements, presence of
    # certain keys in a dict, etc.).

See the unit tests for schema examples.

There are a few limitations (only string keys for any dictionaries in data) and a more fully Pythonic validator might focus on interfaces and abstract base classes over concrete types. However, dataschema is a great improvement over ad hoc validation code for many uses today.